public-mirrors/linux
1
0
Fork 0
mirror of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git synced 2025-08-05 16:54:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
linux/drivers/net/dsa/ocelot/felix_vsc9959.c

3024 lines
84 KiB
C
Raw Normal View History

net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
// SPDX-License-Identifier: (GPL-2.0 OR MIT)
/* Copyright 2017 Microsemi Corporation
net: update NXP copyright text NXP Legal insists that the following are not fine: - Saying "NXP Semiconductors" instead of "NXP", since the company's registered name is "NXP" - Putting a "(c)" sign in the copyright string - Putting a comma in the copyright string The only accepted copyright string format is "Copyright <year-range> NXP". This patch changes the copyright headers in the networking files that were sent by me, or derived from code sent by me. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-17 14:17:35 +03:00
* Copyright 2018-2019 NXP
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
*/
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
#include <linux/fsl/enetc_mdio.h>
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
#include <soc/mscc/ocelot_qsys.h>
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
#include <soc/mscc/ocelot_vcap.h>
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
#include <soc/mscc/ocelot_ana.h>
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
#include <soc/mscc/ocelot_ptp.h>
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#include <soc/mscc/ocelot_sys.h>
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
#include <net/tc_act/tc_gate.h>
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#include <soc/mscc/ocelot.h>
net: mscc: ocelot: use common tag parsing code with DSA The Injection Frame Header and Extraction Frame Header that the switch prepends to frames over the NPI port is also prepended to frames delivered over the CPU port module's queues. Let's unify the handling of the frame headers by making the ocelot driver call some helpers exported by the DSA tagger. Among other things, this allows us to get rid of the strange cpu_to_be32 when transmitting the Injection Frame Header on ocelot, since the packing API uses network byte order natively (when "quirks" is 0). The comments above ocelot_gen_ifh talk about setting pop_cnt to 3, and the cpu extraction queue mask to something, but the code doesn't do it, so we don't do it either. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-14 00:37:56 +02:00
#include <linux/dsa/ocelot.h>
net: dsa: ocelot: use the Lynx PCS helpers in Felix and Seville Use the helper functions introduced by the newly added Lynx PCS MDIO module in the Felix VSC9959 and Seville VSC9953. Instead of representing the PCS as a phy_device, a mdio_device structure will be passed to the Lynx module which is now actually implementing all the PCS configuration and status reporting. All code previously used for PCS monitoring and runtime configuration is removed and replaced will calls to the Lynx PCS operations. Tested on the following SERDES protocols of LS1028A: 0x7777 (2500Base-X), 0x85bb (QSGMII), 0x9999 (SGMII) and 0x13bb (USXGMII). Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-08-30 11:34:02 +03:00
#include <linux/pcs-lynx.h>
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
#include <net/pkt_sched.h>
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#include <linux/iopoll.h>
net: dsa: felix: (re)use already existing constants Now that there are USXGMII constants available, drop the old definitions and reuse the generic ones. Signed-off-by: Michael Walle <michael@walle.cc> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-20 00:03:34 +02:00
#include <linux/mdio.h>
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#include <linux/pci.h>
time64.h: consolidate uses of PSEC_PER_NSEC Time-sensitive networking code needs to work with PTP times expressed in nanoseconds, and with packet transmission times expressed in picoseconds, since those would be fractional at higher than gigabit speed when expressed in nanoseconds. Convert the existing uses in tc-taprio and the ocelot/felix DSA driver to a PSEC_PER_NSEC macro. This macro is placed in include/linux/time64.h as opposed to its relatives (PSEC_PER_SEC etc) from include/vdso/time64.h because the vDSO library does not (yet) need/use it. Cc: Andy Lutomirski <luto@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vincenzo Frascino <vincenzo.frascino@arm.com> # for the vDSO parts Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:38 +03:00
#include <linux/time.h>
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#include "felix.h"
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
#define VSC9959_NUM_PORTS 6
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
#define VSC9959_TAS_GCL_ENTRY_MAX 63
net: mscc: ocelot: use index to set vcap policer Policer was previously automatically assigned from the highest index to the lowest index from policer pool. But police action of tc flower now uses index to set an police entry. This patch uses the police index to set vcap policers, so that one policer can be shared by multiple rules. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:02 +08:00
#define VSC9959_VCAP_POLICER_BASE 63
#define VSC9959_VCAP_POLICER_MAX 383
net: dsa: ocelot: remove unnecessary pci_bar variables The pci_bar variables for the switch and imdio don't make sense for the generic felix driver. Moving them to felix_vsc9959 to limit scope and simplify the felix_info struct. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 09:00:27 -08:00
#define VSC9959_SWITCH_PCI_BAR 4
#define VSC9959_IMDIO_PCI_BAR 0
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
#define VSC9959_PORT_MODE_SERDES (OCELOT_PORT_MODE_SGMII | \
OCELOT_PORT_MODE_QSGMII | \
net: dsa: ocelot: accept 1000base-X for VSC9959 and VSC9953 Switches using the Lynx PCS driver support 1000base-X optical SFP modules. Accept this interface type on a port. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220510164320.10313-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-05-10 19:43:20 +03:00
OCELOT_PORT_MODE_1000BASEX | \
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
OCELOT_PORT_MODE_2500BASEX | \
OCELOT_PORT_MODE_USXGMII)
static const u32 vsc9959_port_modes[VSC9959_NUM_PORTS] = {
VSC9959_PORT_MODE_SERDES,
VSC9959_PORT_MODE_SERDES,
VSC9959_PORT_MODE_SERDES,
VSC9959_PORT_MODE_SERDES,
OCELOT_PORT_MODE_INTERNAL,
net: dsa: felix: allow PHY_INTERFACE_MODE_INTERNAL on port 5 The Felix switch has 6 ports, 2 of which are internal. Due to some misunderstanding, my initial suggestion for vsc9959_port_modes[]: https://patchwork.kernel.org/project/netdevbpf/patch/20220129220221.2823127-10-colin.foster@in-advantage.com/#24718277 got translated by Colin into a 5-port array, leading to an all-zero port mode mask for port 5. Fixes: acf242fc739e ("net: dsa: felix: remove prevalidate_phy_mode interface") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Link: https://lore.kernel.org/r/20220318195812.276276-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-03-18 21:58:12 +02:00
OCELOT_PORT_MODE_INTERNAL,
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
};
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
static const u32 vsc9959_ana_regmap[] = {
REG(ANA_ADVLEARN, 0x0089a0),
REG(ANA_VLANMASK, 0x0089a4),
REG_RESERVED(ANA_PORT_B_DOMAIN),
REG(ANA_ANAGEFIL, 0x0089ac),
REG(ANA_ANEVENTS, 0x0089b0),
REG(ANA_STORMLIMIT_BURST, 0x0089b4),
REG(ANA_STORMLIMIT_CFG, 0x0089b8),
REG(ANA_ISOLATED_PORTS, 0x0089c8),
REG(ANA_COMMUNITY_PORTS, 0x0089cc),
REG(ANA_AUTOAGE, 0x0089d0),
REG(ANA_MACTOPTIONS, 0x0089d4),
REG(ANA_LEARNDISC, 0x0089d8),
REG(ANA_AGENCTRL, 0x0089dc),
REG(ANA_MIRRORPORTS, 0x0089e0),
REG(ANA_EMIRRORPORTS, 0x0089e4),
REG(ANA_FLOODING, 0x0089e8),
REG(ANA_FLOODING_IPMC, 0x008a08),
REG(ANA_SFLOW_CFG, 0x008a0c),
REG(ANA_PORT_MODE, 0x008a28),
REG(ANA_CUT_THRU_CFG, 0x008a48),
REG(ANA_PGID_PGID, 0x008400),
REG(ANA_TABLES_ANMOVED, 0x007f1c),
REG(ANA_TABLES_MACHDATA, 0x007f20),
REG(ANA_TABLES_MACLDATA, 0x007f24),
REG(ANA_TABLES_STREAMDATA, 0x007f28),
REG(ANA_TABLES_MACACCESS, 0x007f2c),
REG(ANA_TABLES_MACTINDX, 0x007f30),
REG(ANA_TABLES_VLANACCESS, 0x007f34),
REG(ANA_TABLES_VLANTIDX, 0x007f38),
REG(ANA_TABLES_ISDXACCESS, 0x007f3c),
REG(ANA_TABLES_ISDXTIDX, 0x007f40),
REG(ANA_TABLES_ENTRYLIM, 0x007f00),
REG(ANA_TABLES_PTP_ID_HIGH, 0x007f44),
REG(ANA_TABLES_PTP_ID_LOW, 0x007f48),
REG(ANA_TABLES_STREAMACCESS, 0x007f4c),
REG(ANA_TABLES_STREAMTIDX, 0x007f50),
REG(ANA_TABLES_SEQ_HISTORY, 0x007f54),
REG(ANA_TABLES_SEQ_MASK, 0x007f58),
REG(ANA_TABLES_SFID_MASK, 0x007f5c),
REG(ANA_TABLES_SFIDACCESS, 0x007f60),
REG(ANA_TABLES_SFIDTIDX, 0x007f64),
REG(ANA_MSTI_STATE, 0x008600),
REG(ANA_OAM_UPM_LM_CNT, 0x008000),
REG(ANA_SG_ACCESS_CTRL, 0x008a64),
REG(ANA_SG_CONFIG_REG_1, 0x007fb0),
REG(ANA_SG_CONFIG_REG_2, 0x007fb4),
REG(ANA_SG_CONFIG_REG_3, 0x007fb8),
REG(ANA_SG_CONFIG_REG_4, 0x007fbc),
REG(ANA_SG_CONFIG_REG_5, 0x007fc0),
REG(ANA_SG_GCL_GS_CONFIG, 0x007f80),
REG(ANA_SG_GCL_TI_CONFIG, 0x007f90),
REG(ANA_SG_STATUS_REG_1, 0x008980),
REG(ANA_SG_STATUS_REG_2, 0x008984),
REG(ANA_SG_STATUS_REG_3, 0x008988),
REG(ANA_PORT_VLAN_CFG, 0x007800),
REG(ANA_PORT_DROP_CFG, 0x007804),
REG(ANA_PORT_QOS_CFG, 0x007808),
REG(ANA_PORT_VCAP_CFG, 0x00780c),
REG(ANA_PORT_VCAP_S1_KEY_CFG, 0x007810),
REG(ANA_PORT_VCAP_S2_CFG, 0x00781c),
REG(ANA_PORT_PCP_DEI_MAP, 0x007820),
REG(ANA_PORT_CPU_FWD_CFG, 0x007860),
REG(ANA_PORT_CPU_FWD_BPDU_CFG, 0x007864),
REG(ANA_PORT_CPU_FWD_GARP_CFG, 0x007868),
REG(ANA_PORT_CPU_FWD_CCM_CFG, 0x00786c),
REG(ANA_PORT_PORT_CFG, 0x007870),
REG(ANA_PORT_POL_CFG, 0x007874),
REG(ANA_PORT_PTP_CFG, 0x007878),
REG(ANA_PORT_PTP_DLY1_CFG, 0x00787c),
REG(ANA_PORT_PTP_DLY2_CFG, 0x007880),
REG(ANA_PORT_SFID_CFG, 0x007884),
REG(ANA_PFC_PFC_CFG, 0x008800),
REG_RESERVED(ANA_PFC_PFC_TIMER),
REG_RESERVED(ANA_IPT_OAM_MEP_CFG),
REG_RESERVED(ANA_IPT_IPT),
REG_RESERVED(ANA_PPT_PPT),
REG_RESERVED(ANA_FID_MAP_FID_MAP),
REG(ANA_AGGR_CFG, 0x008a68),
REG(ANA_CPUQ_CFG, 0x008a6c),
REG_RESERVED(ANA_CPUQ_CFG2),
REG(ANA_CPUQ_8021_CFG, 0x008a74),
REG(ANA_DSCP_CFG, 0x008ab4),
REG(ANA_DSCP_REWR_CFG, 0x008bb4),
REG(ANA_VCAP_RNG_TYPE_CFG, 0x008bf4),
REG(ANA_VCAP_RNG_VAL_CFG, 0x008c14),
REG_RESERVED(ANA_VRAP_CFG),
REG_RESERVED(ANA_VRAP_HDR_DATA),
REG_RESERVED(ANA_VRAP_HDR_MASK),
REG(ANA_DISCARD_CFG, 0x008c40),
REG(ANA_FID_CFG, 0x008c44),
REG(ANA_POL_PIR_CFG, 0x004000),
REG(ANA_POL_CIR_CFG, 0x004004),
REG(ANA_POL_MODE_CFG, 0x004008),
REG(ANA_POL_PIR_STATE, 0x00400c),
REG(ANA_POL_CIR_STATE, 0x004010),
REG_RESERVED(ANA_POL_STATE),
REG(ANA_POL_FLOWC, 0x008c48),
REG(ANA_POL_HYST, 0x008cb4),
REG_RESERVED(ANA_POL_MISC_CFG),
};
static const u32 vsc9959_qs_regmap[] = {
REG(QS_XTR_GRP_CFG, 0x000000),
REG(QS_XTR_RD, 0x000008),
REG(QS_XTR_FRM_PRUNING, 0x000010),
REG(QS_XTR_FLUSH, 0x000018),
REG(QS_XTR_DATA_PRESENT, 0x00001c),
REG(QS_XTR_CFG, 0x000020),
REG(QS_INJ_GRP_CFG, 0x000024),
REG(QS_INJ_WR, 0x00002c),
REG(QS_INJ_CTRL, 0x000034),
REG(QS_INJ_STATUS, 0x00003c),
REG(QS_INJ_ERR, 0x000040),
REG_RESERVED(QS_INH_DBG),
};
net: mscc: ocelot: generalize existing code for VCAP In the Ocelot switches there are 3 TCAMs: VCAP ES0, IS1 and IS2, which have the same configuration interface, but different sets of keys and actions. The driver currently only supports VCAP IS2. In preparation of VCAP IS1 and ES0 support, the existing code must be generalized to work with any VCAP. In that direction, we should move the structures that depend upon VCAP instantiation, like vcap_is2_keys and vcap_is2_actions, out of struct ocelot and into struct vcap_props .keys and .actions, a structure that is replicated 3 times, once per VCAP. We'll pass that structure as an argument to each function that does the key and action packing - only the control logic needs to distinguish between ocelot->vcap[VCAP_IS2] or IS1 or ES0. Another change is to make use of the newly introduced ocelot_target_read and ocelot_target_write API, since the 3 VCAPs have the same registers but put at different addresses. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:23 +03:00
static const u32 vsc9959_vcap_regmap[] = {
/* VCAP_CORE_CFG */
REG(VCAP_CORE_UPDATE_CTRL, 0x000000),
REG(VCAP_CORE_MV_CFG, 0x000004),
/* VCAP_CORE_CACHE */
REG(VCAP_CACHE_ENTRY_DAT, 0x000008),
REG(VCAP_CACHE_MASK_DAT, 0x000108),
REG(VCAP_CACHE_ACTION_DAT, 0x000208),
REG(VCAP_CACHE_CNT_DAT, 0x000308),
REG(VCAP_CACHE_TG_DAT, 0x000388),
net: mscc: ocelot: automatically detect VCAP constants The numbers in struct vcap_props are not intuitive to derive, because they are not a straightforward copy-and-paste from the reference manual but instead rely on a fairly detailed level of understanding of the layout of an entry in the TCAM and in the action RAM. For this reason, bugs are very easy to introduce here. Ease the work of hardware porters and read from hardware the constants that were exported for this particular purpose. Note that this implies that struct vcap_props can no longer be const. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:26 +03:00
/* VCAP_CONST */
REG(VCAP_CONST_VCAP_VER, 0x000398),
REG(VCAP_CONST_ENTRY_WIDTH, 0x00039c),
REG(VCAP_CONST_ENTRY_CNT, 0x0003a0),
REG(VCAP_CONST_ENTRY_SWCNT, 0x0003a4),
REG(VCAP_CONST_ENTRY_TG_WIDTH, 0x0003a8),
REG(VCAP_CONST_ACTION_DEF_CNT, 0x0003ac),
REG(VCAP_CONST_ACTION_WIDTH, 0x0003b0),
REG(VCAP_CONST_CNT_WIDTH, 0x0003b4),
REG(VCAP_CONST_CORE_CNT, 0x0003b8),
REG(VCAP_CONST_IF_CNT, 0x0003bc),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
};
static const u32 vsc9959_qsys_regmap[] = {
REG(QSYS_PORT_MODE, 0x00f460),
REG(QSYS_SWITCH_PORT_MODE, 0x00f480),
REG(QSYS_STAT_CNT_CFG, 0x00f49c),
REG(QSYS_EEE_CFG, 0x00f4a0),
REG(QSYS_EEE_THRES, 0x00f4b8),
REG(QSYS_IGR_NO_SHARING, 0x00f4bc),
REG(QSYS_EGR_NO_SHARING, 0x00f4c0),
REG(QSYS_SW_STATUS, 0x00f4c4),
REG(QSYS_EXT_CPU_CFG, 0x00f4e0),
REG_RESERVED(QSYS_PAD_CFG),
REG(QSYS_CPU_GROUP_MAP, 0x00f4e8),
REG_RESERVED(QSYS_QMAP),
REG_RESERVED(QSYS_ISDX_SGRP),
REG_RESERVED(QSYS_TIMED_FRAME_ENTRY),
REG(QSYS_TFRM_MISC, 0x00f50c),
REG(QSYS_TFRM_PORT_DLY, 0x00f510),
REG(QSYS_TFRM_TIMER_CFG_1, 0x00f514),
REG(QSYS_TFRM_TIMER_CFG_2, 0x00f518),
REG(QSYS_TFRM_TIMER_CFG_3, 0x00f51c),
REG(QSYS_TFRM_TIMER_CFG_4, 0x00f520),
REG(QSYS_TFRM_TIMER_CFG_5, 0x00f524),
REG(QSYS_TFRM_TIMER_CFG_6, 0x00f528),
REG(QSYS_TFRM_TIMER_CFG_7, 0x00f52c),
REG(QSYS_TFRM_TIMER_CFG_8, 0x00f530),
REG(QSYS_RED_PROFILE, 0x00f534),
REG(QSYS_RES_QOS_MODE, 0x00f574),
REG(QSYS_RES_CFG, 0x00c000),
REG(QSYS_RES_STAT, 0x00c004),
REG(QSYS_EGR_DROP_MODE, 0x00f578),
REG(QSYS_EQ_CTRL, 0x00f57c),
REG_RESERVED(QSYS_EVENTS_CORE),
REG(QSYS_QMAXSDU_CFG_0, 0x00f584),
REG(QSYS_QMAXSDU_CFG_1, 0x00f5a0),
REG(QSYS_QMAXSDU_CFG_2, 0x00f5bc),
REG(QSYS_QMAXSDU_CFG_3, 0x00f5d8),
REG(QSYS_QMAXSDU_CFG_4, 0x00f5f4),
REG(QSYS_QMAXSDU_CFG_5, 0x00f610),
REG(QSYS_QMAXSDU_CFG_6, 0x00f62c),
REG(QSYS_QMAXSDU_CFG_7, 0x00f648),
REG(QSYS_PREEMPTION_CFG, 0x00f664),
net: dsa: felix: add support Credit Based Shaper(CBS) for hardware offload VSC9959 hardware support the Credit Based Shaper(CBS) which part of the IEEE-802.1Qav. This patch support sch_cbs set for VSC9959. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:10 +08:00
REG(QSYS_CIR_CFG, 0x000000),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(QSYS_EIR_CFG, 0x000004),
REG(QSYS_SE_CFG, 0x000008),
REG(QSYS_SE_DWRR_CFG, 0x00000c),
REG_RESERVED(QSYS_SE_CONNECT),
REG(QSYS_SE_DLB_SENSE, 0x000040),
REG(QSYS_CIR_STATE, 0x000044),
REG(QSYS_EIR_STATE, 0x000048),
REG_RESERVED(QSYS_SE_STATE),
REG(QSYS_HSCH_MISC_CFG, 0x00f67c),
REG(QSYS_TAG_CONFIG, 0x00f680),
REG(QSYS_TAS_PARAM_CFG_CTRL, 0x00f698),
REG(QSYS_PORT_MAX_SDU, 0x00f69c),
REG(QSYS_PARAM_CFG_REG_1, 0x00f440),
REG(QSYS_PARAM_CFG_REG_2, 0x00f444),
REG(QSYS_PARAM_CFG_REG_3, 0x00f448),
REG(QSYS_PARAM_CFG_REG_4, 0x00f44c),
REG(QSYS_PARAM_CFG_REG_5, 0x00f450),
REG(QSYS_GCL_CFG_REG_1, 0x00f454),
REG(QSYS_GCL_CFG_REG_2, 0x00f458),
REG(QSYS_PARAM_STATUS_REG_1, 0x00f400),
REG(QSYS_PARAM_STATUS_REG_2, 0x00f404),
REG(QSYS_PARAM_STATUS_REG_3, 0x00f408),
REG(QSYS_PARAM_STATUS_REG_4, 0x00f40c),
REG(QSYS_PARAM_STATUS_REG_5, 0x00f410),
REG(QSYS_PARAM_STATUS_REG_6, 0x00f414),
REG(QSYS_PARAM_STATUS_REG_7, 0x00f418),
REG(QSYS_PARAM_STATUS_REG_8, 0x00f41c),
REG(QSYS_PARAM_STATUS_REG_9, 0x00f420),
REG(QSYS_GCL_STATUS_REG_1, 0x00f424),
REG(QSYS_GCL_STATUS_REG_2, 0x00f428),
};
static const u32 vsc9959_rew_regmap[] = {
REG(REW_PORT_VLAN_CFG, 0x000000),
REG(REW_TAG_CFG, 0x000004),
REG(REW_PORT_CFG, 0x000008),
REG(REW_DSCP_CFG, 0x00000c),
REG(REW_PCP_DEI_QOS_MAP_CFG, 0x000010),
REG(REW_PTP_CFG, 0x000050),
REG(REW_PTP_DLY1_CFG, 0x000054),
REG(REW_RED_TAG_CFG, 0x000058),
REG(REW_DSCP_REMAP_DP1_CFG, 0x000410),
REG(REW_DSCP_REMAP_CFG, 0x000510),
REG_RESERVED(REW_STAT_CFG),
REG_RESERVED(REW_REW_STICKY),
REG_RESERVED(REW_PPT),
};
static const u32 vsc9959_sys_regmap[] = {
REG(SYS_COUNT_RX_OCTETS, 0x000000),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_RX_UNICAST, 0x000004),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_RX_MULTICAST, 0x000008),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_RX_BROADCAST, 0x00000c),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_RX_SHORTS, 0x000010),
REG(SYS_COUNT_RX_FRAGMENTS, 0x000014),
REG(SYS_COUNT_RX_JABBERS, 0x000018),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_RX_CRC_ALIGN_ERRS, 0x00001c),
REG(SYS_COUNT_RX_SYM_ERRS, 0x000020),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_RX_64, 0x000024),
REG(SYS_COUNT_RX_65_127, 0x000028),
REG(SYS_COUNT_RX_128_255, 0x00002c),
net: mscc: ocelot: fix incorrect ndo_get_stats64 packet counters Reading stats using the SYS_COUNT_* register definitions is only used by ocelot_get_stats64() from the ocelot switchdev driver, however, currently the bucket definitions are incorrect. Separately, on both RX and TX, we have the following problems: - a 256-1023 bucket which actually tracks the 256-511 packets - the 1024-1526 bucket actually tracks the 512-1023 packets - the 1527-max bucket actually tracks the 1024-1526 packets => nobody tracks the packets from the real 1527-max bucket Additionally, the RX_PAUSE, RX_CONTROL, RX_LONGS and RX_CLASSIFIED_DROPS all track the wrong thing. However this doesn't seem to have any consequence, since ocelot_get_stats64() doesn't use these. Even though this problem only manifests itself for the switchdev driver, we cannot split the fix for ocelot and for DSA, since it requires fixing the bucket definitions from enum ocelot_reg, which makes us necessarily adapt the structures from felix and seville as well. Fixes: 84705fc16552 ("net: dsa: felix: introduce support for Seville VSC9953 switch") Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Fixes: a556c76adc05 ("net: mscc: Add initial Ocelot switch support") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:46 +03:00
REG(SYS_COUNT_RX_256_511, 0x000030),
REG(SYS_COUNT_RX_512_1023, 0x000034),
REG(SYS_COUNT_RX_1024_1526, 0x000038),
REG(SYS_COUNT_RX_1527_MAX, 0x00003c),
REG(SYS_COUNT_RX_PAUSE, 0x000040),
REG(SYS_COUNT_RX_CONTROL, 0x000044),
REG(SYS_COUNT_RX_LONGS, 0x000048),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_RX_CLASSIFIED_DROPS, 0x00004c),
REG(SYS_COUNT_RX_RED_PRIO_0, 0x000050),
REG(SYS_COUNT_RX_RED_PRIO_1, 0x000054),
REG(SYS_COUNT_RX_RED_PRIO_2, 0x000058),
REG(SYS_COUNT_RX_RED_PRIO_3, 0x00005c),
REG(SYS_COUNT_RX_RED_PRIO_4, 0x000060),
REG(SYS_COUNT_RX_RED_PRIO_5, 0x000064),
REG(SYS_COUNT_RX_RED_PRIO_6, 0x000068),
REG(SYS_COUNT_RX_RED_PRIO_7, 0x00006c),
REG(SYS_COUNT_RX_YELLOW_PRIO_0, 0x000070),
REG(SYS_COUNT_RX_YELLOW_PRIO_1, 0x000074),
REG(SYS_COUNT_RX_YELLOW_PRIO_2, 0x000078),
REG(SYS_COUNT_RX_YELLOW_PRIO_3, 0x00007c),
REG(SYS_COUNT_RX_YELLOW_PRIO_4, 0x000080),
REG(SYS_COUNT_RX_YELLOW_PRIO_5, 0x000084),
REG(SYS_COUNT_RX_YELLOW_PRIO_6, 0x000088),
REG(SYS_COUNT_RX_YELLOW_PRIO_7, 0x00008c),
REG(SYS_COUNT_RX_GREEN_PRIO_0, 0x000090),
REG(SYS_COUNT_RX_GREEN_PRIO_1, 0x000094),
REG(SYS_COUNT_RX_GREEN_PRIO_2, 0x000098),
REG(SYS_COUNT_RX_GREEN_PRIO_3, 0x00009c),
REG(SYS_COUNT_RX_GREEN_PRIO_4, 0x0000a0),
REG(SYS_COUNT_RX_GREEN_PRIO_5, 0x0000a4),
REG(SYS_COUNT_RX_GREEN_PRIO_6, 0x0000a8),
REG(SYS_COUNT_RX_GREEN_PRIO_7, 0x0000ac),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_TX_OCTETS, 0x000200),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_TX_UNICAST, 0x000204),
REG(SYS_COUNT_TX_MULTICAST, 0x000208),
REG(SYS_COUNT_TX_BROADCAST, 0x00020c),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_TX_COLLISION, 0x000210),
REG(SYS_COUNT_TX_DROPS, 0x000214),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_TX_PAUSE, 0x000218),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_TX_64, 0x00021c),
REG(SYS_COUNT_TX_65_127, 0x000220),
net: mscc: ocelot: fix incorrect ndo_get_stats64 packet counters Reading stats using the SYS_COUNT_* register definitions is only used by ocelot_get_stats64() from the ocelot switchdev driver, however, currently the bucket definitions are incorrect. Separately, on both RX and TX, we have the following problems: - a 256-1023 bucket which actually tracks the 256-511 packets - the 1024-1526 bucket actually tracks the 512-1023 packets - the 1527-max bucket actually tracks the 1024-1526 packets => nobody tracks the packets from the real 1527-max bucket Additionally, the RX_PAUSE, RX_CONTROL, RX_LONGS and RX_CLASSIFIED_DROPS all track the wrong thing. However this doesn't seem to have any consequence, since ocelot_get_stats64() doesn't use these. Even though this problem only manifests itself for the switchdev driver, we cannot split the fix for ocelot and for DSA, since it requires fixing the bucket definitions from enum ocelot_reg, which makes us necessarily adapt the structures from felix and seville as well. Fixes: 84705fc16552 ("net: dsa: felix: introduce support for Seville VSC9953 switch") Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Fixes: a556c76adc05 ("net: mscc: Add initial Ocelot switch support") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:46 +03:00
REG(SYS_COUNT_TX_128_255, 0x000224),
REG(SYS_COUNT_TX_256_511, 0x000228),
REG(SYS_COUNT_TX_512_1023, 0x00022c),
REG(SYS_COUNT_TX_1024_1526, 0x000230),
REG(SYS_COUNT_TX_1527_MAX, 0x000234),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_TX_YELLOW_PRIO_0, 0x000238),
REG(SYS_COUNT_TX_YELLOW_PRIO_1, 0x00023c),
REG(SYS_COUNT_TX_YELLOW_PRIO_2, 0x000240),
REG(SYS_COUNT_TX_YELLOW_PRIO_3, 0x000244),
REG(SYS_COUNT_TX_YELLOW_PRIO_4, 0x000248),
REG(SYS_COUNT_TX_YELLOW_PRIO_5, 0x00024c),
REG(SYS_COUNT_TX_YELLOW_PRIO_6, 0x000250),
REG(SYS_COUNT_TX_YELLOW_PRIO_7, 0x000254),
REG(SYS_COUNT_TX_GREEN_PRIO_0, 0x000258),
REG(SYS_COUNT_TX_GREEN_PRIO_1, 0x00025c),
REG(SYS_COUNT_TX_GREEN_PRIO_2, 0x000260),
REG(SYS_COUNT_TX_GREEN_PRIO_3, 0x000264),
REG(SYS_COUNT_TX_GREEN_PRIO_4, 0x000268),
REG(SYS_COUNT_TX_GREEN_PRIO_5, 0x00026c),
REG(SYS_COUNT_TX_GREEN_PRIO_6, 0x000270),
REG(SYS_COUNT_TX_GREEN_PRIO_7, 0x000274),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_COUNT_TX_AGING, 0x000278),
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
REG(SYS_COUNT_DROP_LOCAL, 0x000400),
REG(SYS_COUNT_DROP_TAIL, 0x000404),
REG(SYS_COUNT_DROP_YELLOW_PRIO_0, 0x000408),
REG(SYS_COUNT_DROP_YELLOW_PRIO_1, 0x00040c),
REG(SYS_COUNT_DROP_YELLOW_PRIO_2, 0x000410),
REG(SYS_COUNT_DROP_YELLOW_PRIO_3, 0x000414),
REG(SYS_COUNT_DROP_YELLOW_PRIO_4, 0x000418),
REG(SYS_COUNT_DROP_YELLOW_PRIO_5, 0x00041c),
REG(SYS_COUNT_DROP_YELLOW_PRIO_6, 0x000420),
REG(SYS_COUNT_DROP_YELLOW_PRIO_7, 0x000424),
REG(SYS_COUNT_DROP_GREEN_PRIO_0, 0x000428),
REG(SYS_COUNT_DROP_GREEN_PRIO_1, 0x00042c),
REG(SYS_COUNT_DROP_GREEN_PRIO_2, 0x000430),
REG(SYS_COUNT_DROP_GREEN_PRIO_3, 0x000434),
REG(SYS_COUNT_DROP_GREEN_PRIO_4, 0x000438),
REG(SYS_COUNT_DROP_GREEN_PRIO_5, 0x00043c),
REG(SYS_COUNT_DROP_GREEN_PRIO_6, 0x000440),
REG(SYS_COUNT_DROP_GREEN_PRIO_7, 0x000444),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_RESET_CFG, 0x000e00),
REG(SYS_SR_ETYPE_CFG, 0x000e04),
REG(SYS_VLAN_ETYPE_CFG, 0x000e08),
REG(SYS_PORT_MODE, 0x000e0c),
REG(SYS_FRONT_PORT_MODE, 0x000e2c),
REG(SYS_FRM_AGING, 0x000e44),
REG(SYS_STAT_CFG, 0x000e48),
REG(SYS_SW_STATUS, 0x000e4c),
REG_RESERVED(SYS_MISC_CFG),
REG(SYS_REW_MAC_HIGH_CFG, 0x000e6c),
REG(SYS_REW_MAC_LOW_CFG, 0x000e84),
REG(SYS_TIMESTAMP_OFFSET, 0x000e9c),
REG(SYS_PAUSE_CFG, 0x000ea0),
REG(SYS_PAUSE_TOT_CFG, 0x000ebc),
REG(SYS_ATOP, 0x000ec0),
REG(SYS_ATOP_TOT_CFG, 0x000edc),
REG(SYS_MAC_FC_CFG, 0x000ee0),
REG(SYS_MMGT, 0x000ef8),
REG_RESERVED(SYS_MMGT_FAST),
REG_RESERVED(SYS_EVENTS_DIF),
REG_RESERVED(SYS_EVENTS_CORE),
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
REG(SYS_CNT, 0x000000),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
REG(SYS_PTP_STATUS, 0x000f14),
REG(SYS_PTP_TXSTAMP, 0x000f18),
REG(SYS_PTP_NXT, 0x000f1c),
REG(SYS_PTP_CFG, 0x000f20),
REG(SYS_RAM_INIT, 0x000f24),
REG_RESERVED(SYS_CM_ADDR),
REG_RESERVED(SYS_CM_DATA_WR),
REG_RESERVED(SYS_CM_DATA_RD),
REG_RESERVED(SYS_CM_OP),
REG_RESERVED(SYS_CM_DATA),
};
net: dsa: ocelot: define PTP registers for felix_vsc9959 This patch is to define PTP registers for felix_vsc9959. Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-20 16:23:17 +08:00
static const u32 vsc9959_ptp_regmap[] = {
net: dsa: felix: replace tabs with spaces Over the time, some patches have introduced structures aligned with spaces, near structures aligned with tabs. Fix the inconsistencies. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:49 +03:00
REG(PTP_PIN_CFG, 0x000000),
REG(PTP_PIN_TOD_SEC_MSB, 0x000004),
REG(PTP_PIN_TOD_SEC_LSB, 0x000008),
REG(PTP_PIN_TOD_NSEC, 0x00000c),
REG(PTP_PIN_WF_HIGH_PERIOD, 0x000014),
REG(PTP_PIN_WF_LOW_PERIOD, 0x000018),
REG(PTP_CFG_MISC, 0x0000a0),
REG(PTP_CLK_CFG_ADJ_CFG, 0x0000a4),
REG(PTP_CLK_CFG_ADJ_FREQ, 0x0000a8),
net: dsa: ocelot: define PTP registers for felix_vsc9959 This patch is to define PTP registers for felix_vsc9959. Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-20 16:23:17 +08:00
};
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
static const u32 vsc9959_gcb_regmap[] = {
REG(GCB_SOFT_RST, 0x000004),
};
net: mscc: ocelot: convert port registers to regmap At the moment, there are some minimal register differences between VSC7514 Ocelot and VSC9959 Felix. To be precise, the PCS1G registers are missing from Felix because it was integrated with an NXP PCS. But with VSC9953 Seville (not yet introduced), the register differences are more pronounced. The MAC registers are located at different offsets within the DEV_GMII target. So we need to refactor the driver to keep a regmap even for per-port registers. The callers of the ocelot_port_readl and ocelot_port_writel were kept unchanged, only the implementation is now more generic. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:01 +03:00
static const u32 vsc9959_dev_gmii_regmap[] = {
REG(DEV_CLOCK_CFG, 0x0),
REG(DEV_PORT_MISC, 0x4),
REG(DEV_EVENTS, 0x8),
REG(DEV_EEE_CFG, 0xc),
REG(DEV_RX_PATH_DELAY, 0x10),
REG(DEV_TX_PATH_DELAY, 0x14),
REG(DEV_PTP_PREDICT_CFG, 0x18),
REG(DEV_MAC_ENA_CFG, 0x1c),
REG(DEV_MAC_MODE_CFG, 0x20),
REG(DEV_MAC_MAXLEN_CFG, 0x24),
REG(DEV_MAC_TAGS_CFG, 0x28),
REG(DEV_MAC_ADV_CHK_CFG, 0x2c),
REG(DEV_MAC_IFG_CFG, 0x30),
REG(DEV_MAC_HDX_CFG, 0x34),
REG(DEV_MAC_DBG_CFG, 0x38),
REG(DEV_MAC_FC_MAC_LOW_CFG, 0x3c),
REG(DEV_MAC_FC_MAC_HIGH_CFG, 0x40),
REG(DEV_MAC_STICKY, 0x44),
REG_RESERVED(PCS1G_CFG),
REG_RESERVED(PCS1G_MODE_CFG),
REG_RESERVED(PCS1G_SD_CFG),
REG_RESERVED(PCS1G_ANEG_CFG),
REG_RESERVED(PCS1G_ANEG_NP_CFG),
REG_RESERVED(PCS1G_LB_CFG),
REG_RESERVED(PCS1G_DBG_CFG),
REG_RESERVED(PCS1G_CDET_CFG),
REG_RESERVED(PCS1G_ANEG_STATUS),
REG_RESERVED(PCS1G_ANEG_NP_STATUS),
REG_RESERVED(PCS1G_LINK_STATUS),
REG_RESERVED(PCS1G_LINK_DOWN_CNT),
REG_RESERVED(PCS1G_STICKY),
REG_RESERVED(PCS1G_DEBUG_STATUS),
REG_RESERVED(PCS1G_LPI_CFG),
REG_RESERVED(PCS1G_LPI_WAKE_ERROR_CNT),
REG_RESERVED(PCS1G_LPI_STATUS),
REG_RESERVED(PCS1G_TSTPAT_MODE_CFG),
REG_RESERVED(PCS1G_TSTPAT_STATUS),
REG_RESERVED(DEV_PCS_FX100_CFG),
REG_RESERVED(DEV_PCS_FX100_STATUS),
};
static const u32 *vsc9959_regmap[TARGET_MAX] = {
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[ANA] = vsc9959_ana_regmap,
[QS] = vsc9959_qs_regmap,
[QSYS] = vsc9959_qsys_regmap,
[REW] = vsc9959_rew_regmap,
[SYS] = vsc9959_sys_regmap,
net: mscc: ocelot: add definitions for VCAP ES0 keys, actions and target As a preparation step for the offloading to ES0, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:25 +03:00
[S0] = vsc9959_vcap_regmap,
net: mscc: ocelot: add definitions for VCAP IS1 keys, actions and target As a preparation step for the offloading to IS1, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:24 +03:00
[S1] = vsc9959_vcap_regmap,
net: mscc: ocelot: generalize existing code for VCAP In the Ocelot switches there are 3 TCAMs: VCAP ES0, IS1 and IS2, which have the same configuration interface, but different sets of keys and actions. The driver currently only supports VCAP IS2. In preparation of VCAP IS1 and ES0 support, the existing code must be generalized to work with any VCAP. In that direction, we should move the structures that depend upon VCAP instantiation, like vcap_is2_keys and vcap_is2_actions, out of struct ocelot and into struct vcap_props .keys and .actions, a structure that is replicated 3 times, once per VCAP. We'll pass that structure as an argument to each function that does the key and action packing - only the control logic needs to distinguish between ocelot->vcap[VCAP_IS2] or IS1 or ES0. Another change is to make use of the newly introduced ocelot_target_read and ocelot_target_write API, since the 3 VCAPs have the same registers but put at different addresses. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:23 +03:00
[S2] = vsc9959_vcap_regmap,
net: dsa: ocelot: define PTP registers for felix_vsc9959 This patch is to define PTP registers for felix_vsc9959. Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-20 16:23:17 +08:00
[PTP] = vsc9959_ptp_regmap,
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[GCB] = vsc9959_gcb_regmap,
net: mscc: ocelot: convert port registers to regmap At the moment, there are some minimal register differences between VSC7514 Ocelot and VSC9959 Felix. To be precise, the PCS1G registers are missing from Felix because it was integrated with an NXP PCS. But with VSC9953 Seville (not yet introduced), the register differences are more pronounced. The MAC registers are located at different offsets within the DEV_GMII target. So we need to refactor the driver to keep a regmap even for per-port registers. The callers of the ocelot_port_readl and ocelot_port_writel were kept unchanged, only the implementation is now more generic. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:01 +03:00
[DEV_GMII] = vsc9959_dev_gmii_regmap,
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
};
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
/* Addresses are relative to the PCI device's base address */
net: mscc: ocelot: convert port registers to regmap At the moment, there are some minimal register differences between VSC7514 Ocelot and VSC9959 Felix. To be precise, the PCS1G registers are missing from Felix because it was integrated with an NXP PCS. But with VSC9953 Seville (not yet introduced), the register differences are more pronounced. The MAC registers are located at different offsets within the DEV_GMII target. So we need to refactor the driver to keep a regmap even for per-port registers. The callers of the ocelot_port_readl and ocelot_port_writel were kept unchanged, only the implementation is now more generic. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:01 +03:00
static const struct resource vsc9959_target_io_res[TARGET_MAX] = {
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[ANA] = {
.start = 0x0280000,
.end = 0x028ffff,
.name = "ana",
},
[QS] = {
.start = 0x0080000,
.end = 0x00800ff,
.name = "qs",
},
[QSYS] = {
.start = 0x0200000,
.end = 0x021ffff,
.name = "qsys",
},
[REW] = {
.start = 0x0030000,
.end = 0x003ffff,
.name = "rew",
},
[SYS] = {
.start = 0x0010000,
.end = 0x001ffff,
.name = "sys",
},
net: mscc: ocelot: add definitions for VCAP ES0 keys, actions and target As a preparation step for the offloading to ES0, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:25 +03:00
[S0] = {
.start = 0x0040000,
.end = 0x00403ff,
.name = "s0",
},
net: mscc: ocelot: add definitions for VCAP IS1 keys, actions and target As a preparation step for the offloading to IS1, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:24 +03:00
[S1] = {
.start = 0x0050000,
.end = 0x00503ff,
.name = "s1",
},
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[S2] = {
.start = 0x0060000,
.end = 0x00603ff,
.name = "s2",
},
net: dsa: ocelot: define PTP registers for felix_vsc9959 This patch is to define PTP registers for felix_vsc9959. Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-20 16:23:17 +08:00
[PTP] = {
.start = 0x0090000,
.end = 0x00900cb,
.name = "ptp",
},
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[GCB] = {
.start = 0x0070000,
.end = 0x00701ff,
.name = "devcpu_gcb",
},
};
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
static const struct resource vsc9959_port_io_res[] = {
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
{
.start = 0x0100000,
.end = 0x010ffff,
.name = "port0",
},
{
.start = 0x0110000,
.end = 0x011ffff,
.name = "port1",
},
{
.start = 0x0120000,
.end = 0x012ffff,
.name = "port2",
},
{
.start = 0x0130000,
.end = 0x013ffff,
.name = "port3",
},
{
.start = 0x0140000,
.end = 0x014ffff,
.name = "port4",
},
{
.start = 0x0150000,
.end = 0x015ffff,
.name = "port5",
},
};
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
/* Port MAC 0 Internal MDIO bus through which the SerDes acting as an
* SGMII/QSGMII MAC PCS can be found.
*/
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
static const struct resource vsc9959_imdio_res = {
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
.start = 0x8030,
.end = 0x8040,
.name = "imdio",
};
soc: mscc: ocelot: add MII registers description Add the register definitions for the MSCC MIIM MDIO controller in preparation for seville_vsc9959.c to create its accessors for the internal MDIO bus. Since we've introduced elements to ocelot_regfields that are not instantiated by felix and ocelot, we need to define the size of the regfields arrays explicitly, otherwise ocelot_regfields_init, which iterates up to REGFIELD_MAX, will fault on the undefined regfield entries (if we're lucky). Signed-off-by: Maxim Kochetkov <fido_max@inbox.ru> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:02 +03:00
static const struct reg_field vsc9959_regfields[REGFIELD_MAX] = {
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
[ANA_ADVLEARN_VLAN_CHK] = REG_FIELD(ANA_ADVLEARN, 6, 6),
[ANA_ADVLEARN_LEARN_MIRROR] = REG_FIELD(ANA_ADVLEARN, 0, 5),
[ANA_ANEVENTS_FLOOD_DISCARD] = REG_FIELD(ANA_ANEVENTS, 30, 30),
[ANA_ANEVENTS_AUTOAGE] = REG_FIELD(ANA_ANEVENTS, 26, 26),
[ANA_ANEVENTS_STORM_DROP] = REG_FIELD(ANA_ANEVENTS, 24, 24),
[ANA_ANEVENTS_LEARN_DROP] = REG_FIELD(ANA_ANEVENTS, 23, 23),
[ANA_ANEVENTS_AGED_ENTRY] = REG_FIELD(ANA_ANEVENTS, 22, 22),
[ANA_ANEVENTS_CPU_LEARN_FAILED] = REG_FIELD(ANA_ANEVENTS, 21, 21),
[ANA_ANEVENTS_AUTO_LEARN_FAILED] = REG_FIELD(ANA_ANEVENTS, 20, 20),
[ANA_ANEVENTS_LEARN_REMOVE] = REG_FIELD(ANA_ANEVENTS, 19, 19),
[ANA_ANEVENTS_AUTO_LEARNED] = REG_FIELD(ANA_ANEVENTS, 18, 18),
[ANA_ANEVENTS_AUTO_MOVED] = REG_FIELD(ANA_ANEVENTS, 17, 17),
[ANA_ANEVENTS_CLASSIFIED_DROP] = REG_FIELD(ANA_ANEVENTS, 15, 15),
[ANA_ANEVENTS_CLASSIFIED_COPY] = REG_FIELD(ANA_ANEVENTS, 14, 14),
[ANA_ANEVENTS_VLAN_DISCARD] = REG_FIELD(ANA_ANEVENTS, 13, 13),
[ANA_ANEVENTS_FWD_DISCARD] = REG_FIELD(ANA_ANEVENTS, 12, 12),
[ANA_ANEVENTS_MULTICAST_FLOOD] = REG_FIELD(ANA_ANEVENTS, 11, 11),
[ANA_ANEVENTS_UNICAST_FLOOD] = REG_FIELD(ANA_ANEVENTS, 10, 10),
[ANA_ANEVENTS_DEST_KNOWN] = REG_FIELD(ANA_ANEVENTS, 9, 9),
[ANA_ANEVENTS_BUCKET3_MATCH] = REG_FIELD(ANA_ANEVENTS, 8, 8),
[ANA_ANEVENTS_BUCKET2_MATCH] = REG_FIELD(ANA_ANEVENTS, 7, 7),
[ANA_ANEVENTS_BUCKET1_MATCH] = REG_FIELD(ANA_ANEVENTS, 6, 6),
[ANA_ANEVENTS_BUCKET0_MATCH] = REG_FIELD(ANA_ANEVENTS, 5, 5),
[ANA_ANEVENTS_CPU_OPERATION] = REG_FIELD(ANA_ANEVENTS, 4, 4),
[ANA_ANEVENTS_DMAC_LOOKUP] = REG_FIELD(ANA_ANEVENTS, 3, 3),
[ANA_ANEVENTS_SMAC_LOOKUP] = REG_FIELD(ANA_ANEVENTS, 2, 2),
[ANA_ANEVENTS_SEQ_GEN_ERR_0] = REG_FIELD(ANA_ANEVENTS, 1, 1),
[ANA_ANEVENTS_SEQ_GEN_ERR_1] = REG_FIELD(ANA_ANEVENTS, 0, 0),
[ANA_TABLES_MACACCESS_B_DOM] = REG_FIELD(ANA_TABLES_MACACCESS, 16, 16),
[ANA_TABLES_MACTINDX_BUCKET] = REG_FIELD(ANA_TABLES_MACTINDX, 11, 12),
[ANA_TABLES_MACTINDX_M_INDEX] = REG_FIELD(ANA_TABLES_MACTINDX, 0, 10),
[SYS_RESET_CFG_CORE_ENA] = REG_FIELD(SYS_RESET_CFG, 0, 0),
[GCB_SOFT_RST_SWC_RST] = REG_FIELD(GCB_SOFT_RST, 0, 0),
net: mscc: ocelot: convert QSYS_SWITCH_PORT_MODE and SYS_PORT_MODE to regfields Currently Felix and Ocelot share the same bit layout in these per-port registers, but Seville does not. So we need reg_fields for that. Actually since these are per-port registers, we need to also specify the number of ports, and register size per port, and use the regmap API for multiple ports. There's a more subtle point to be made about the other 2 register fields: - QSYS_SWITCH_PORT_MODE_SCH_NEXT_CFG - QSYS_SWITCH_PORT_MODE_INGRESS_DROP_MODE which we are not writing any longer, for 2 reasons: - Using the previous API (ocelot_write_rix), we were only writing 1 for Felix and Ocelot, which was their hardware-default value, and which there wasn't any intention in changing. - In the case of SCH_NEXT_CFG, in fact Seville does not have this register field at all, and therefore, if we want to have common code we would be required to not write to it. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:03 +03:00
/* Replicated per number of ports (7), register size 4 per port */
[QSYS_SWITCH_PORT_MODE_PORT_ENA] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 14, 14, 7, 4),
[QSYS_SWITCH_PORT_MODE_SCH_NEXT_CFG] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 11, 13, 7, 4),
[QSYS_SWITCH_PORT_MODE_YEL_RSRVD] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 10, 10, 7, 4),
[QSYS_SWITCH_PORT_MODE_INGRESS_DROP_MODE] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 9, 9, 7, 4),
[QSYS_SWITCH_PORT_MODE_TX_PFC_ENA] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 1, 8, 7, 4),
[QSYS_SWITCH_PORT_MODE_TX_PFC_MODE] = REG_FIELD_ID(QSYS_SWITCH_PORT_MODE, 0, 0, 7, 4),
[SYS_PORT_MODE_DATA_WO_TS] = REG_FIELD_ID(SYS_PORT_MODE, 5, 6, 7, 4),
[SYS_PORT_MODE_INCL_INJ_HDR] = REG_FIELD_ID(SYS_PORT_MODE, 3, 4, 7, 4),
[SYS_PORT_MODE_INCL_XTR_HDR] = REG_FIELD_ID(SYS_PORT_MODE, 1, 2, 7, 4),
[SYS_PORT_MODE_INCL_HDR_ERR] = REG_FIELD_ID(SYS_PORT_MODE, 0, 0, 7, 4),
net: mscc: ocelot: convert SYS_PAUSE_CFG register access to regfield Seville has a different bitwise layout than Ocelot and Felix. Signed-off-by: Maxim Kochetkov <fido_max@inbox.ru> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:07 +03:00
[SYS_PAUSE_CFG_PAUSE_START] = REG_FIELD_ID(SYS_PAUSE_CFG, 10, 18, 7, 4),
[SYS_PAUSE_CFG_PAUSE_STOP] = REG_FIELD_ID(SYS_PAUSE_CFG, 1, 9, 7, 4),
[SYS_PAUSE_CFG_PAUSE_ENA] = REG_FIELD_ID(SYS_PAUSE_CFG, 0, 1, 7, 4),
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
};
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
static const struct ocelot_stat_layout vsc9959_stats_layout[OCELOT_NUM_STATS] = {
[OCELOT_STAT_RX_OCTETS] = {
.name = "rx_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_OCTETS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_UNICAST] = {
.name = "rx_unicast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_UNICAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_MULTICAST] = {
.name = "rx_multicast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_MULTICAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_BROADCAST] = {
.name = "rx_broadcast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_BROADCAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_SHORTS] = {
.name = "rx_shorts",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_SHORTS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_FRAGMENTS] = {
.name = "rx_fragments",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_FRAGMENTS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_JABBERS] = {
.name = "rx_jabbers",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_JABBERS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_CRC_ALIGN_ERRS] = {
.name = "rx_crc_align_errs",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_CRC_ALIGN_ERRS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_SYM_ERRS] = {
.name = "rx_sym_errs",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_SYM_ERRS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_64] = {
.name = "rx_frames_below_65_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_64,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_65_127] = {
.name = "rx_frames_65_to_127_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_65_127,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_128_255] = {
.name = "rx_frames_128_to_255_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_128_255,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_256_511] = {
.name = "rx_frames_256_to_511_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_256_511,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_512_1023] = {
.name = "rx_frames_512_to_1023_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_512_1023,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_1024_1526] = {
.name = "rx_frames_1024_to_1526_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_1024_1526,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_1527_MAX] = {
.name = "rx_frames_over_1526_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_1527_MAX,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_PAUSE] = {
.name = "rx_pause",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_PAUSE,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_CONTROL] = {
.name = "rx_control",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_CONTROL,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_LONGS] = {
.name = "rx_longs",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_LONGS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_CLASSIFIED_DROPS] = {
.name = "rx_classified_drops",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_CLASSIFIED_DROPS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_0] = {
.name = "rx_red_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_1] = {
.name = "rx_red_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_2] = {
.name = "rx_red_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_3] = {
.name = "rx_red_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_4] = {
.name = "rx_red_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_5] = {
.name = "rx_red_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_6] = {
.name = "rx_red_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_RED_PRIO_7] = {
.name = "rx_red_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_RED_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_0] = {
.name = "rx_yellow_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_1] = {
.name = "rx_yellow_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_2] = {
.name = "rx_yellow_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_3] = {
.name = "rx_yellow_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_4] = {
.name = "rx_yellow_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_5] = {
.name = "rx_yellow_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_6] = {
.name = "rx_yellow_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_YELLOW_PRIO_7] = {
.name = "rx_yellow_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_YELLOW_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_0] = {
.name = "rx_green_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_1] = {
.name = "rx_green_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_2] = {
.name = "rx_green_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_3] = {
.name = "rx_green_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_4] = {
.name = "rx_green_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_5] = {
.name = "rx_green_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_6] = {
.name = "rx_green_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_RX_GREEN_PRIO_7] = {
.name = "rx_green_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_RX_GREEN_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_OCTETS] = {
.name = "tx_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_OCTETS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_UNICAST] = {
.name = "tx_unicast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_UNICAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_MULTICAST] = {
.name = "tx_multicast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_MULTICAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_BROADCAST] = {
.name = "tx_broadcast",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_BROADCAST,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_COLLISION] = {
.name = "tx_collision",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_COLLISION,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_DROPS] = {
.name = "tx_drops",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_DROPS,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_PAUSE] = {
.name = "tx_pause",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_PAUSE,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_64] = {
.name = "tx_frames_below_65_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_64,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_65_127] = {
.name = "tx_frames_65_to_127_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_65_127,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_128_255] = {
.name = "tx_frames_128_255_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_128_255,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_256_511] = {
.name = "tx_frames_256_511_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_256_511,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_512_1023] = {
.name = "tx_frames_512_1023_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_512_1023,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_1024_1526] = {
.name = "tx_frames_1024_1526_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_1024_1526,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_1527_MAX] = {
.name = "tx_frames_over_1526_octets",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_1527_MAX,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_0] = {
.name = "tx_yellow_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_1] = {
.name = "tx_yellow_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_2] = {
.name = "tx_yellow_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_3] = {
.name = "tx_yellow_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_4] = {
.name = "tx_yellow_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_5] = {
.name = "tx_yellow_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_6] = {
.name = "tx_yellow_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_YELLOW_PRIO_7] = {
.name = "tx_yellow_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_YELLOW_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_0] = {
.name = "tx_green_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_1] = {
.name = "tx_green_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_2] = {
.name = "tx_green_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_3] = {
.name = "tx_green_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_4] = {
.name = "tx_green_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_5] = {
.name = "tx_green_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_6] = {
.name = "tx_green_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_GREEN_PRIO_7] = {
.name = "tx_green_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_GREEN_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_TX_AGED] = {
.name = "tx_aged",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_TX_AGING,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_LOCAL] = {
.name = "drop_local",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_LOCAL,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_TAIL] = {
.name = "drop_tail",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_TAIL,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_0] = {
.name = "drop_yellow_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_1] = {
.name = "drop_yellow_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_2] = {
.name = "drop_yellow_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_3] = {
.name = "drop_yellow_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_4] = {
.name = "drop_yellow_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_5] = {
.name = "drop_yellow_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_6] = {
.name = "drop_yellow_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_YELLOW_PRIO_7] = {
.name = "drop_yellow_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_YELLOW_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_0] = {
.name = "drop_green_prio_0",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_0,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_1] = {
.name = "drop_green_prio_1",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_1,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_2] = {
.name = "drop_green_prio_2",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_2,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_3] = {
.name = "drop_green_prio_3",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_3,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_4] = {
.name = "drop_green_prio_4",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_4,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_5] = {
.name = "drop_green_prio_5",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_5,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_6] = {
.name = "drop_green_prio_6",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_6,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
[OCELOT_STAT_DROP_GREEN_PRIO_7] = {
.name = "drop_green_prio_7",
net: mscc: ocelot: keep ocelot_stat_layout by reg address, not offset With so many counter addresses recently discovered as being wrong, it is desirable to at least have a central database of information, rather than two: one through the SYS_COUNT_* registers (used for ndo_get_stats64), and the other through the offset field of struct ocelot_stat_layout elements (used for ethtool -S). The strategy will be to keep the SYS_COUNT_* definitions as the single source of truth, but for that we need to expand our current definitions to cover all registers. Then we need to convert the ocelot region creation logic, and stats worker, to the read semantics imposed by going through SYS_COUNT_* absolute register addresses, rather than offsets of 32-bit words relative to SYS_COUNT_RX_OCTETS (which should have been SYS_CNT, by the way). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:51 +03:00
.reg = SYS_COUNT_DROP_GREEN_PRIO_7,
net: mscc: ocelot: make struct ocelot_stat_layout array indexable The ocelot counters are 32-bit and require periodic reading, every 2 seconds, by ocelot_port_update_stats(), so that wraparounds are detected. Currently, the counters reported by ocelot_get_stats64() come from the 32-bit hardware counters directly, rather than from the 64-bit accumulated ocelot->stats, and this is a problem for their integrity. The strategy is to make ocelot_get_stats64() able to cherry-pick individual stats from ocelot->stats the way in which it currently reads them out from SYS_COUNT_* registers. But currently it can't, because ocelot->stats is an opaque u64 array that's used only to feed data into ethtool -S. To solve that problem, we need to make ocelot->stats indexable, and associate each element with an element of struct ocelot_stat_layout used by ethtool -S. This makes ocelot_stat_layout a fat (and possibly sparse) array, so we need to change the way in which we access it. We no longer need OCELOT_STAT_END as a sentinel, because we know the array's size (OCELOT_NUM_STATS). We just need to skip the array elements that were left unpopulated for the switch revision (ocelot, felix, seville). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:50 +03:00
},
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
};
net: mscc: ocelot: add definitions for VCAP ES0 keys, actions and target As a preparation step for the offloading to ES0, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:25 +03:00
static const struct vcap_field vsc9959_vcap_es0_keys[] = {
[VCAP_ES0_EGR_PORT] = { 0, 3},
[VCAP_ES0_IGR_PORT] = { 3, 3},
[VCAP_ES0_RSV] = { 6, 2},
[VCAP_ES0_L2_MC] = { 8, 1},
[VCAP_ES0_L2_BC] = { 9, 1},
[VCAP_ES0_VID] = { 10, 12},
[VCAP_ES0_DP] = { 22, 1},
[VCAP_ES0_PCP] = { 23, 3},
};
static const struct vcap_field vsc9959_vcap_es0_actions[] = {
[VCAP_ES0_ACT_PUSH_OUTER_TAG] = { 0, 2},
[VCAP_ES0_ACT_PUSH_INNER_TAG] = { 2, 1},
[VCAP_ES0_ACT_TAG_A_TPID_SEL] = { 3, 2},
[VCAP_ES0_ACT_TAG_A_VID_SEL] = { 5, 1},
[VCAP_ES0_ACT_TAG_A_PCP_SEL] = { 6, 2},
[VCAP_ES0_ACT_TAG_A_DEI_SEL] = { 8, 2},
[VCAP_ES0_ACT_TAG_B_TPID_SEL] = { 10, 2},
[VCAP_ES0_ACT_TAG_B_VID_SEL] = { 12, 1},
[VCAP_ES0_ACT_TAG_B_PCP_SEL] = { 13, 2},
[VCAP_ES0_ACT_TAG_B_DEI_SEL] = { 15, 2},
[VCAP_ES0_ACT_VID_A_VAL] = { 17, 12},
[VCAP_ES0_ACT_PCP_A_VAL] = { 29, 3},
[VCAP_ES0_ACT_DEI_A_VAL] = { 32, 1},
[VCAP_ES0_ACT_VID_B_VAL] = { 33, 12},
[VCAP_ES0_ACT_PCP_B_VAL] = { 45, 3},
[VCAP_ES0_ACT_DEI_B_VAL] = { 48, 1},
[VCAP_ES0_ACT_RSV] = { 49, 23},
[VCAP_ES0_ACT_HIT_STICKY] = { 72, 1},
};
net: mscc: ocelot: add definitions for VCAP IS1 keys, actions and target As a preparation step for the offloading to IS1, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:24 +03:00
static const struct vcap_field vsc9959_vcap_is1_keys[] = {
[VCAP_IS1_HK_TYPE] = { 0, 1},
[VCAP_IS1_HK_LOOKUP] = { 1, 2},
[VCAP_IS1_HK_IGR_PORT_MASK] = { 3, 7},
[VCAP_IS1_HK_RSV] = { 10, 9},
[VCAP_IS1_HK_OAM_Y1731] = { 19, 1},
[VCAP_IS1_HK_L2_MC] = { 20, 1},
[VCAP_IS1_HK_L2_BC] = { 21, 1},
[VCAP_IS1_HK_IP_MC] = { 22, 1},
[VCAP_IS1_HK_VLAN_TAGGED] = { 23, 1},
[VCAP_IS1_HK_VLAN_DBL_TAGGED] = { 24, 1},
[VCAP_IS1_HK_TPID] = { 25, 1},
[VCAP_IS1_HK_VID] = { 26, 12},
[VCAP_IS1_HK_DEI] = { 38, 1},
[VCAP_IS1_HK_PCP] = { 39, 3},
/* Specific Fields for IS1 Half Key S1_NORMAL */
[VCAP_IS1_HK_L2_SMAC] = { 42, 48},
[VCAP_IS1_HK_ETYPE_LEN] = { 90, 1},
[VCAP_IS1_HK_ETYPE] = { 91, 16},
[VCAP_IS1_HK_IP_SNAP] = {107, 1},
[VCAP_IS1_HK_IP4] = {108, 1},
/* Layer-3 Information */
[VCAP_IS1_HK_L3_FRAGMENT] = {109, 1},
[VCAP_IS1_HK_L3_FRAG_OFS_GT0] = {110, 1},
[VCAP_IS1_HK_L3_OPTIONS] = {111, 1},
[VCAP_IS1_HK_L3_DSCP] = {112, 6},
[VCAP_IS1_HK_L3_IP4_SIP] = {118, 32},
/* Layer-4 Information */
[VCAP_IS1_HK_TCP_UDP] = {150, 1},
[VCAP_IS1_HK_TCP] = {151, 1},
[VCAP_IS1_HK_L4_SPORT] = {152, 16},
[VCAP_IS1_HK_L4_RNG] = {168, 8},
/* Specific Fields for IS1 Half Key S1_5TUPLE_IP4 */
[VCAP_IS1_HK_IP4_INNER_TPID] = { 42, 1},
[VCAP_IS1_HK_IP4_INNER_VID] = { 43, 12},
[VCAP_IS1_HK_IP4_INNER_DEI] = { 55, 1},
[VCAP_IS1_HK_IP4_INNER_PCP] = { 56, 3},
[VCAP_IS1_HK_IP4_IP4] = { 59, 1},
[VCAP_IS1_HK_IP4_L3_FRAGMENT] = { 60, 1},
[VCAP_IS1_HK_IP4_L3_FRAG_OFS_GT0] = { 61, 1},
[VCAP_IS1_HK_IP4_L3_OPTIONS] = { 62, 1},
[VCAP_IS1_HK_IP4_L3_DSCP] = { 63, 6},
[VCAP_IS1_HK_IP4_L3_IP4_DIP] = { 69, 32},
[VCAP_IS1_HK_IP4_L3_IP4_SIP] = {101, 32},
[VCAP_IS1_HK_IP4_L3_PROTO] = {133, 8},
[VCAP_IS1_HK_IP4_TCP_UDP] = {141, 1},
[VCAP_IS1_HK_IP4_TCP] = {142, 1},
[VCAP_IS1_HK_IP4_L4_RNG] = {143, 8},
[VCAP_IS1_HK_IP4_IP_PAYLOAD_S1_5TUPLE] = {151, 32},
};
static const struct vcap_field vsc9959_vcap_is1_actions[] = {
[VCAP_IS1_ACT_DSCP_ENA] = { 0, 1},
[VCAP_IS1_ACT_DSCP_VAL] = { 1, 6},
[VCAP_IS1_ACT_QOS_ENA] = { 7, 1},
[VCAP_IS1_ACT_QOS_VAL] = { 8, 3},
[VCAP_IS1_ACT_DP_ENA] = { 11, 1},
[VCAP_IS1_ACT_DP_VAL] = { 12, 1},
[VCAP_IS1_ACT_PAG_OVERRIDE_MASK] = { 13, 8},
[VCAP_IS1_ACT_PAG_VAL] = { 21, 8},
[VCAP_IS1_ACT_RSV] = { 29, 9},
net: mscc: ocelot: offload ingress skbedit and vlan actions to VCAP IS1 VCAP IS1 is a VCAP module which can filter on the most common L2/L3/L4 Ethernet keys, and modify the results of the basic QoS classification and VLAN classification based on those flow keys. There are 3 VCAP IS1 lookups, mapped over chains 10000, 11000 and 12000. Currently the driver is hardcoded to use IS1_ACTION_TYPE_NORMAL half keys. Note that the VLAN_MANGLE has been omitted for now. In hardware, the VCAP_IS1_ACT_VID_REPLACE_ENA field replaces the classified VLAN (metadata associated with the frame) and not the VLAN from the header itself. There are currently some issues which need to be addressed when operating in standalone, or in bridge with vlan_filtering=0 modes, because in those cases the switch ports have VLAN awareness disabled, and changing the classified VLAN to anything other than the pvid causes the packets to be dropped. Another issue is that on egress, we expect port tagging to push the classified VLAN, but port tagging is disabled in the modes mentioned above, so although the classified VLAN is replaced, it is not visible in the packet transmitted by the switch. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-10-02 15:02:23 +03:00
/* The fields below are incorrectly shifted by 2 in the manual */
net: mscc: ocelot: add definitions for VCAP IS1 keys, actions and target As a preparation step for the offloading to IS1, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:24 +03:00
[VCAP_IS1_ACT_VID_REPLACE_ENA] = { 38, 1},
[VCAP_IS1_ACT_VID_ADD_VAL] = { 39, 12},
[VCAP_IS1_ACT_FID_SEL] = { 51, 2},
[VCAP_IS1_ACT_FID_VAL] = { 53, 13},
[VCAP_IS1_ACT_PCP_DEI_ENA] = { 66, 1},
[VCAP_IS1_ACT_PCP_VAL] = { 67, 3},
[VCAP_IS1_ACT_DEI_VAL] = { 70, 1},
[VCAP_IS1_ACT_VLAN_POP_CNT_ENA] = { 71, 1},
[VCAP_IS1_ACT_VLAN_POP_CNT] = { 72, 2},
[VCAP_IS1_ACT_CUSTOM_ACE_TYPE_ENA] = { 74, 4},
[VCAP_IS1_ACT_HIT_STICKY] = { 78, 1},
};
net: dsa: felix: make vcap is2 keys and actions static Get rid of some sparse warnings. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-06-20 18:43:36 +03:00
static struct vcap_field vsc9959_vcap_is2_keys[] = {
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
/* Common: 41 bits */
[VCAP_IS2_TYPE] = { 0, 4},
[VCAP_IS2_HK_FIRST] = { 4, 1},
[VCAP_IS2_HK_PAG] = { 5, 8},
[VCAP_IS2_HK_IGR_PORT_MASK] = { 13, 7},
[VCAP_IS2_HK_RSV2] = { 20, 1},
[VCAP_IS2_HK_HOST_MATCH] = { 21, 1},
[VCAP_IS2_HK_L2_MC] = { 22, 1},
[VCAP_IS2_HK_L2_BC] = { 23, 1},
[VCAP_IS2_HK_VLAN_TAGGED] = { 24, 1},
[VCAP_IS2_HK_VID] = { 25, 12},
[VCAP_IS2_HK_DEI] = { 37, 1},
[VCAP_IS2_HK_PCP] = { 38, 3},
/* MAC_ETYPE / MAC_LLC / MAC_SNAP / OAM common */
[VCAP_IS2_HK_L2_DMAC] = { 41, 48},
[VCAP_IS2_HK_L2_SMAC] = { 89, 48},
/* MAC_ETYPE (TYPE=000) */
[VCAP_IS2_HK_MAC_ETYPE_ETYPE] = {137, 16},
[VCAP_IS2_HK_MAC_ETYPE_L2_PAYLOAD0] = {153, 16},
[VCAP_IS2_HK_MAC_ETYPE_L2_PAYLOAD1] = {169, 8},
[VCAP_IS2_HK_MAC_ETYPE_L2_PAYLOAD2] = {177, 3},
/* MAC_LLC (TYPE=001) */
[VCAP_IS2_HK_MAC_LLC_L2_LLC] = {137, 40},
/* MAC_SNAP (TYPE=010) */
[VCAP_IS2_HK_MAC_SNAP_L2_SNAP] = {137, 40},
/* MAC_ARP (TYPE=011) */
[VCAP_IS2_HK_MAC_ARP_SMAC] = { 41, 48},
[VCAP_IS2_HK_MAC_ARP_ADDR_SPACE_OK] = { 89, 1},
[VCAP_IS2_HK_MAC_ARP_PROTO_SPACE_OK] = { 90, 1},
[VCAP_IS2_HK_MAC_ARP_LEN_OK] = { 91, 1},
[VCAP_IS2_HK_MAC_ARP_TARGET_MATCH] = { 92, 1},
[VCAP_IS2_HK_MAC_ARP_SENDER_MATCH] = { 93, 1},
[VCAP_IS2_HK_MAC_ARP_OPCODE_UNKNOWN] = { 94, 1},
[VCAP_IS2_HK_MAC_ARP_OPCODE] = { 95, 2},
[VCAP_IS2_HK_MAC_ARP_L3_IP4_DIP] = { 97, 32},
[VCAP_IS2_HK_MAC_ARP_L3_IP4_SIP] = {129, 32},
[VCAP_IS2_HK_MAC_ARP_DIP_EQ_SIP] = {161, 1},
/* IP4_TCP_UDP / IP4_OTHER common */
[VCAP_IS2_HK_IP4] = { 41, 1},
[VCAP_IS2_HK_L3_FRAGMENT] = { 42, 1},
[VCAP_IS2_HK_L3_FRAG_OFS_GT0] = { 43, 1},
[VCAP_IS2_HK_L3_OPTIONS] = { 44, 1},
[VCAP_IS2_HK_IP4_L3_TTL_GT0] = { 45, 1},
[VCAP_IS2_HK_L3_TOS] = { 46, 8},
[VCAP_IS2_HK_L3_IP4_DIP] = { 54, 32},
[VCAP_IS2_HK_L3_IP4_SIP] = { 86, 32},
[VCAP_IS2_HK_DIP_EQ_SIP] = {118, 1},
/* IP4_TCP_UDP (TYPE=100) */
[VCAP_IS2_HK_TCP] = {119, 1},
net: dsa: felix: fix some key offsets for IP4_TCP_UDP VCAP IS2 entries Some of the IS2 IP4_TCP_UDP keys are not correct, like L4_DPORT, L4_SPORT and other L4 keys. This prevents offloaded tc-flower rules from matching on src_port and dst_port for TCP and UDP packets. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-22 01:56:36 +03:00
[VCAP_IS2_HK_L4_DPORT] = {120, 16},
[VCAP_IS2_HK_L4_SPORT] = {136, 16},
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
[VCAP_IS2_HK_L4_RNG] = {152, 8},
[VCAP_IS2_HK_L4_SPORT_EQ_DPORT] = {160, 1},
[VCAP_IS2_HK_L4_SEQUENCE_EQ0] = {161, 1},
net: dsa: felix: fix some key offsets for IP4_TCP_UDP VCAP IS2 entries Some of the IS2 IP4_TCP_UDP keys are not correct, like L4_DPORT, L4_SPORT and other L4 keys. This prevents offloaded tc-flower rules from matching on src_port and dst_port for TCP and UDP packets. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-22 01:56:36 +03:00
[VCAP_IS2_HK_L4_FIN] = {162, 1},
[VCAP_IS2_HK_L4_SYN] = {163, 1},
[VCAP_IS2_HK_L4_RST] = {164, 1},
[VCAP_IS2_HK_L4_PSH] = {165, 1},
[VCAP_IS2_HK_L4_ACK] = {166, 1},
[VCAP_IS2_HK_L4_URG] = {167, 1},
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
[VCAP_IS2_HK_L4_1588_DOM] = {168, 8},
[VCAP_IS2_HK_L4_1588_VER] = {176, 4},
/* IP4_OTHER (TYPE=101) */
[VCAP_IS2_HK_IP4_L3_PROTO] = {119, 8},
[VCAP_IS2_HK_L3_PAYLOAD] = {127, 56},
/* IP6_STD (TYPE=110) */
[VCAP_IS2_HK_IP6_L3_TTL_GT0] = { 41, 1},
[VCAP_IS2_HK_L3_IP6_SIP] = { 42, 128},
[VCAP_IS2_HK_IP6_L3_PROTO] = {170, 8},
/* OAM (TYPE=111) */
[VCAP_IS2_HK_OAM_MEL_FLAGS] = {137, 7},
[VCAP_IS2_HK_OAM_VER] = {144, 5},
[VCAP_IS2_HK_OAM_OPCODE] = {149, 8},
[VCAP_IS2_HK_OAM_FLAGS] = {157, 8},
[VCAP_IS2_HK_OAM_MEPID] = {165, 16},
[VCAP_IS2_HK_OAM_CCM_CNTS_EQ0] = {181, 1},
[VCAP_IS2_HK_OAM_IS_Y1731] = {182, 1},
};
net: dsa: felix: make vcap is2 keys and actions static Get rid of some sparse warnings. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-06-20 18:43:36 +03:00
static struct vcap_field vsc9959_vcap_is2_actions[] = {
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
[VCAP_IS2_ACT_HIT_ME_ONCE] = { 0, 1},
[VCAP_IS2_ACT_CPU_COPY_ENA] = { 1, 1},
[VCAP_IS2_ACT_CPU_QU_NUM] = { 2, 3},
[VCAP_IS2_ACT_MASK_MODE] = { 5, 2},
[VCAP_IS2_ACT_MIRROR_ENA] = { 7, 1},
[VCAP_IS2_ACT_LRN_DIS] = { 8, 1},
[VCAP_IS2_ACT_POLICE_ENA] = { 9, 1},
[VCAP_IS2_ACT_POLICE_IDX] = { 10, 9},
[VCAP_IS2_ACT_POLICE_VCAP_ONLY] = { 19, 1},
net: dsa: felix: fix incorrect action offsets for VCAP IS2 The port mask width was larger than the actual number of ports, and therefore, all fields following this one were also shifted by the number of excess bits. But the driver doesn't use the REW_OP, SMAC_REPLACE_ENA or ACL_ID bits from the action vector, so the bug was inconsequential. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-29 14:20:24 +03:00
[VCAP_IS2_ACT_PORT_MASK] = { 20, 6},
[VCAP_IS2_ACT_REW_OP] = { 26, 9},
[VCAP_IS2_ACT_SMAC_REPLACE_ENA] = { 35, 1},
[VCAP_IS2_ACT_RSV] = { 36, 2},
[VCAP_IS2_ACT_ACL_ID] = { 38, 6},
[VCAP_IS2_ACT_HIT_CNT] = { 44, 32},
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
};
net: mscc: ocelot: automatically detect VCAP constants The numbers in struct vcap_props are not intuitive to derive, because they are not a straightforward copy-and-paste from the reference manual but instead rely on a fairly detailed level of understanding of the layout of an entry in the TCAM and in the action RAM. For this reason, bugs are very easy to introduce here. Ease the work of hardware porters and read from hardware the constants that were exported for this particular purpose. Note that this implies that struct vcap_props can no longer be const. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:26 +03:00
static struct vcap_props vsc9959_vcap_props[] = {
net: mscc: ocelot: add definitions for VCAP ES0 keys, actions and target As a preparation step for the offloading to ES0, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:25 +03:00
[VCAP_ES0] = {
.action_type_width = 0,
.action_table = {
[ES0_ACTION_TYPE_NORMAL] = {
.width = 72, /* HIT_STICKY not included */
.count = 1,
},
},
.target = S0,
.keys = vsc9959_vcap_es0_keys,
.actions = vsc9959_vcap_es0_actions,
},
net: mscc: ocelot: add definitions for VCAP IS1 keys, actions and target As a preparation step for the offloading to IS1, let's create the infrastructure for talking with this hardware block. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:24 +03:00
[VCAP_IS1] = {
.action_type_width = 0,
.action_table = {
[IS1_ACTION_TYPE_NORMAL] = {
.width = 78, /* HIT_STICKY not included */
.count = 4,
},
},
.target = S1,
.keys = vsc9959_vcap_is1_keys,
.actions = vsc9959_vcap_is1_actions,
},
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
[VCAP_IS2] = {
.action_type_width = 1,
.action_table = {
[IS2_ACTION_TYPE_NORMAL] = {
.width = 44,
.count = 2
},
[IS2_ACTION_TYPE_SMAC_SIP] = {
.width = 6,
.count = 4
},
},
net: mscc: ocelot: generalize existing code for VCAP In the Ocelot switches there are 3 TCAMs: VCAP ES0, IS1 and IS2, which have the same configuration interface, but different sets of keys and actions. The driver currently only supports VCAP IS2. In preparation of VCAP IS1 and ES0 support, the existing code must be generalized to work with any VCAP. In that direction, we should move the structures that depend upon VCAP instantiation, like vcap_is2_keys and vcap_is2_actions, out of struct ocelot and into struct vcap_props .keys and .actions, a structure that is replicated 3 times, once per VCAP. We'll pass that structure as an argument to each function that does the key and action packing - only the control logic needs to distinguish between ocelot->vcap[VCAP_IS2] or IS1 or ES0. Another change is to make use of the newly introduced ocelot_target_read and ocelot_target_write API, since the 3 VCAPs have the same registers but put at different addresses. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-30 01:27:23 +03:00
.target = S2,
.keys = vsc9959_vcap_is2_keys,
.actions = vsc9959_vcap_is2_actions,
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
},
};
net: dsa: felix: move the PTP clock structure to felix_vsc9959.c Not only does Sevile not have a PTP clock, but with separate modules, this structure cannot even live in felix.c, due to the .owner = THIS_MODULE assignment causing this link time error: drivers/net/dsa/ocelot/felix.o:(.data+0x0): undefined reference to `__this_module' Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:52 +03:00
static const struct ptp_clock_info vsc9959_ptp_caps = {
.owner = THIS_MODULE,
.name = "felix ptp",
.max_adj = 0x7fffffff,
.n_alarm = 0,
.n_ext_ts = 0,
.n_per_out = OCELOT_PTP_PINS_NUM,
.n_pins = OCELOT_PTP_PINS_NUM,
.pps = 0,
.gettime64 = ocelot_ptp_gettime64,
.settime64 = ocelot_ptp_settime64,
.adjtime = ocelot_ptp_adjtime,
.adjfine = ocelot_ptp_adjfine,
.verify = ocelot_ptp_verify,
.enable = ocelot_ptp_enable,
};
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
#define VSC9959_INIT_TIMEOUT 50000
#define VSC9959_GCB_RST_SLEEP 100
#define VSC9959_SYS_RAMINIT_SLEEP 80
static int vsc9959_gcb_soft_rst_status(struct ocelot *ocelot)
{
int val;
net: dsa: felix: use ocelot_field_{read,write} helpers consistently Since these helpers for regmap fields are available, use them. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:43 +03:00
ocelot_field_read(ocelot, GCB_SOFT_RST_SWC_RST, &val);
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
return val;
}
static int vsc9959_sys_ram_init_status(struct ocelot *ocelot)
{
return ocelot_read(ocelot, SYS_RAM_INIT);
}
net: dsa: ocelot: document why reset procedure is different for felix/seville The overall idea (issue soft reset, enable memories, initialize memories, enable core) is the same, so it would make sense that an attempt is made to unify the procedures. It is not immediately obvious that the fields are not part of the same register targets, though. So add a comment. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:46 +03:00
/* CORE_ENA is in SYS:SYSTEM:RESET_CFG
* RAM_INIT is in SYS:RAM_CTRL:RAM_INIT
*/
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
static int vsc9959_reset(struct ocelot *ocelot)
{
int val, err;
/* soft-reset the switch core */
net: dsa: felix: use ocelot_field_{read,write} helpers consistently Since these helpers for regmap fields are available, use them. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:43 +03:00
ocelot_field_write(ocelot, GCB_SOFT_RST_SWC_RST, 1);
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
err = readx_poll_timeout(vsc9959_gcb_soft_rst_status, ocelot, val, !val,
VSC9959_GCB_RST_SLEEP, VSC9959_INIT_TIMEOUT);
if (err) {
dev_err(ocelot->dev, "timeout: switch core reset\n");
return err;
}
/* initialize switch mem ~40us */
ocelot_write(ocelot, SYS_RAM_INIT_RAM_INIT, SYS_RAM_INIT);
err = readx_poll_timeout(vsc9959_sys_ram_init_status, ocelot, val, !val,
VSC9959_SYS_RAMINIT_SLEEP,
VSC9959_INIT_TIMEOUT);
if (err) {
dev_err(ocelot->dev, "timeout: switch sram init\n");
return err;
}
/* enable switch core */
net: dsa: felix: use ocelot_field_{read,write} helpers consistently Since these helpers for regmap fields are available, use them. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:43 +03:00
ocelot_field_write(ocelot, SYS_RESET_CFG_CORE_ENA, 1);
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
return 0;
}
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
static void vsc9959_phylink_validate(struct ocelot *ocelot, int port,
unsigned long *supported,
struct phylink_link_state *state)
{
__ETHTOOL_DECLARE_LINK_MODE_MASK(mask) = { 0, };
phylink_set_port_modes(mask);
phylink_set(mask, Autoneg);
phylink_set(mask, Pause);
phylink_set(mask, Asym_Pause);
phylink_set(mask, 10baseT_Half);
phylink_set(mask, 10baseT_Full);
phylink_set(mask, 100baseT_Half);
phylink_set(mask, 100baseT_Full);
phylink_set(mask, 1000baseT_Half);
phylink_set(mask, 1000baseT_Full);
net: dsa: ocelot: accept 1000base-X for VSC9959 and VSC9953 Switches using the Lynx PCS driver support 1000base-X optical SFP modules. Accept this interface type on a port. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220510164320.10313-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-05-10 19:43:20 +03:00
phylink_set(mask, 1000baseX_Full);
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
if (state->interface == PHY_INTERFACE_MODE_INTERNAL ||
state->interface == PHY_INTERFACE_MODE_2500BASEX ||
state->interface == PHY_INTERFACE_MODE_USXGMII) {
phylink_set(mask, 2500baseT_Full);
phylink_set(mask, 2500baseX_Full);
}
net: convert users of bitmap_foo() to linkmode_foo() This converts instances of bitmap_foo(args..., __ETHTOOL_LINK_MODE_MASK_NBITS) to linkmode_foo(args...) I manually fixed up some lines to prevent them from being excessively long. Otherwise, this change was generated with the following semantic patch: // Generated with // echo linux/linkmode.h > includes // git grep -Flf includes include/ | cut -f 2- -d / | cat includes - \ // | sort | uniq | tee new_includes | wc -l && mv new_includes includes // and repeating until the number stopped going up @i@ @@ ( #include <linux/acpi_mdio.h> | #include <linux/brcmphy.h> | #include <linux/dsa/loop.h> | #include <linux/dsa/sja1105.h> | #include <linux/ethtool.h> | #include <linux/ethtool_netlink.h> | #include <linux/fec.h> | #include <linux/fs_enet_pd.h> | #include <linux/fsl/enetc_mdio.h> | #include <linux/fwnode_mdio.h> | #include <linux/linkmode.h> | #include <linux/lsm_audit.h> | #include <linux/mdio-bitbang.h> | #include <linux/mdio.h> | #include <linux/mdio-mux.h> | #include <linux/mii.h> | #include <linux/mii_timestamper.h> | #include <linux/mlx5/accel.h> | #include <linux/mlx5/cq.h> | #include <linux/mlx5/device.h> | #include <linux/mlx5/driver.h> | #include <linux/mlx5/eswitch.h> | #include <linux/mlx5/fs.h> | #include <linux/mlx5/port.h> | #include <linux/mlx5/qp.h> | #include <linux/mlx5/rsc_dump.h> | #include <linux/mlx5/transobj.h> | #include <linux/mlx5/vport.h> | #include <linux/of_mdio.h> | #include <linux/of_net.h> | #include <linux/pcs-lynx.h> | #include <linux/pcs/pcs-xpcs.h> | #include <linux/phy.h> | #include <linux/phy_led_triggers.h> | #include <linux/phylink.h> | #include <linux/platform_data/bcmgenet.h> | #include <linux/platform_data/xilinx-ll-temac.h> | #include <linux/pxa168_eth.h> | #include <linux/qed/qed_eth_if.h> | #include <linux/qed/qed_fcoe_if.h> | #include <linux/qed/qed_if.h> | #include <linux/qed/qed_iov_if.h> | #include <linux/qed/qed_iscsi_if.h> | #include <linux/qed/qed_ll2_if.h> | #include <linux/qed/qed_nvmetcp_if.h> | #include <linux/qed/qed_rdma_if.h> | #include <linux/sfp.h> | #include <linux/sh_eth.h> | #include <linux/smsc911x.h> | #include <linux/soc/nxp/lpc32xx-misc.h> | #include <linux/stmmac.h> | #include <linux/sunrpc/svc_rdma.h> | #include <linux/sxgbe_platform.h> | #include <net/cfg80211.h> | #include <net/dsa.h> | #include <net/mac80211.h> | #include <net/selftests.h> | #include <rdma/ib_addr.h> | #include <rdma/ib_cache.h> | #include <rdma/ib_cm.h> | #include <rdma/ib_hdrs.h> | #include <rdma/ib_mad.h> | #include <rdma/ib_marshall.h> | #include <rdma/ib_pack.h> | #include <rdma/ib_pma.h> | #include <rdma/ib_sa.h> | #include <rdma/ib_smi.h> | #include <rdma/ib_umem.h> | #include <rdma/ib_umem_odp.h> | #include <rdma/ib_verbs.h> | #include <rdma/iw_cm.h> | #include <rdma/mr_pool.h> | #include <rdma/opa_addr.h> | #include <rdma/opa_port_info.h> | #include <rdma/opa_smi.h> | #include <rdma/opa_vnic.h> | #include <rdma/rdma_cm.h> | #include <rdma/rdma_cm_ib.h> | #include <rdma/rdmavt_cq.h> | #include <rdma/rdma_vt.h> | #include <rdma/rdmavt_qp.h> | #include <rdma/rw.h> | #include <rdma/tid_rdma_defs.h> | #include <rdma/uverbs_ioctl.h> | #include <rdma/uverbs_named_ioctl.h> | #include <rdma/uverbs_std_types.h> | #include <rdma/uverbs_types.h> | #include <soc/mscc/ocelot.h> | #include <soc/mscc/ocelot_ptp.h> | #include <soc/mscc/ocelot_vcap.h> | #include <trace/events/ib_mad.h> | #include <trace/events/rdma_core.h> | #include <trace/events/rdma.h> | #include <trace/events/rpcrdma.h> | #include <uapi/linux/ethtool.h> | #include <uapi/linux/ethtool_netlink.h> | #include <uapi/linux/mdio.h> | #include <uapi/linux/mii.h> ) @depends on i@ expression list args; @@ ( - bitmap_zero(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_zero(args) | - bitmap_copy(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_copy(args) | - bitmap_and(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_and(args) | - bitmap_or(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_or(args) | - bitmap_empty(args, ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_empty(args) | - bitmap_andnot(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_andnot(args) | - bitmap_equal(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_equal(args) | - bitmap_intersects(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_intersects(args) | - bitmap_subset(args, __ETHTOOL_LINK_MODE_MASK_NBITS) + linkmode_subset(args) ) Add missing linux/mii.h include to mellanox. -DaveM Signed-off-by: Sean Anderson <sean.anderson@seco.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-22 18:41:04 -04:00
linkmode_and(supported, supported, mask);
linkmode_and(state->advertising, state->advertising, mask);
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
}
net: mscc: ocelot: extend watermark encoding function The ocelot_wm_encode function deals with setting thresholds for pause frame start and stop. In Ocelot and Felix the register layout is the same, but for Seville, it isn't. The easiest way to accommodate Seville hardware configuration is to introduce a function pointer for setting this up. Signed-off-by: Maxim Kochetkov <fido_max@inbox.ru> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:08 +03:00
/* Watermark encode
* Bit 8: Unit; 0:1, 1:16
* Bit 7-0: Value to be multiplied with unit
*/
static u16 vsc9959_wm_enc(u16 value)
{
net: mscc: ocelot: warn when encoding an out-of-bounds watermark value There is an upper bound to the value that a watermark may hold. That upper bound is not immediately obvious during configuration, and it might be possible to have accidental truncation. Actually this has happened already, add a warning to prevent it from happening again. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-10-05 12:09:12 +03:00
WARN_ON(value >= 16 * BIT(8));
net: mscc: ocelot: extend watermark encoding function The ocelot_wm_encode function deals with setting thresholds for pause frame start and stop. In Ocelot and Felix the register layout is the same, but for Seville, it isn't. The easiest way to accommodate Seville hardware configuration is to introduce a function pointer for setting this up. Signed-off-by: Maxim Kochetkov <fido_max@inbox.ru> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:08 +03:00
if (value >= BIT(8))
return BIT(8) | (value / 16);
return value;
}
net: mscc: ocelot: add ops for decoding watermark threshold and occupancy We'll need to read back the watermark thresholds and occupancy from hardware (for devlink-sb integration), not only to write them as we did so far in ocelot_port_set_maxlen. So introduce 2 new functions in struct ocelot_ops, similar to wm_enc, and implement them for the 3 supported mscc_ocelot switches. Remove the INUSE and MAXUSE unpacking helpers for the QSYS_RES_STAT register, because that doesn't scale with the number of switches that mscc_ocelot supports now. They have different bit widths for the watermarks, and we need function pointers to abstract that difference away. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-15 04:11:12 +02:00
static u16 vsc9959_wm_dec(u16 wm)
{
WARN_ON(wm & ~GENMASK(8, 0));
if (wm & BIT(8))
return (wm & GENMASK(7, 0)) * 16;
return wm;
}
static void vsc9959_wm_stat(u32 val, u32 *inuse, u32 *maxuse)
{
*inuse = (val & GENMASK(23, 12)) >> 12;
*maxuse = val & GENMASK(11, 0);
}
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
static int vsc9959_mdio_bus_alloc(struct ocelot *ocelot)
{
struct felix *felix = ocelot_to_felix(ocelot);
struct enetc_mdio_priv *mdio_priv;
struct device *dev = ocelot->dev;
void __iomem *imdio_regs;
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
struct resource res;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
struct enetc_hw *hw;
struct mii_bus *bus;
int port;
int rc;
felix->pcs = devm_kcalloc(dev, felix->info->num_ports,
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
sizeof(struct phylink_pcs *),
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
GFP_KERNEL);
if (!felix->pcs) {
dev_err(dev, "failed to allocate array for PCS PHYs\n");
return -ENOMEM;
}
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
memcpy(&res, felix->info->imdio_res, sizeof(res));
res.flags = IORESOURCE_MEM;
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
res.start += felix->imdio_base;
res.end += felix->imdio_base;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
felix: Fix initialization of ioremap resources The caller of devm_ioremap_resource(), either accidentally or by wrong assumption, is writing back derived resource data to global static resource initialization tables that should have been constant. Meaning that after it computes the final physical start address it saves the address for no reason in the static tables. This doesn't affect the first driver probing after reboot, but it breaks consecutive driver reloads (i.e. driver unbind & bind) because the initialization tables no longer have the correct initial values. So the next probe() will map the device registers to wrong physical addresses, causing ARM SError async exceptions. This patch fixes all of the above. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-22 11:54:34 +03:00
imdio_regs = devm_ioremap_resource(dev, &res);
net: mscc: ocelot: remove redundant dev_err call in vsc9959_mdio_bus_alloc() There is a error message within devm_ioremap_resource already, so remove the dev_err call to avoid redundant error message. Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Guobin Huang <huangguobin4@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-03-29 09:54:05 +08:00
if (IS_ERR(imdio_regs))
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
return PTR_ERR(imdio_regs);
hw = enetc_hw_alloc(dev, imdio_regs);
if (IS_ERR(hw)) {
dev_err(dev, "failed to allocate ENETC HW structure\n");
return PTR_ERR(hw);
}
net: dsa: felix: don't use devres for mdiobus As explained in commits: 74b6d7d13307 ("net: dsa: realtek: register the MDIO bus under devres") 5135e96a3dd2 ("net: dsa: don't allocate the slave_mii_bus using devres") mdiobus_free() will panic when called from devm_mdiobus_free() <- devres_release_all() <- __device_release_driver(), and that mdiobus was not previously unregistered. The Felix VSC9959 switch is a PCI device, so the initial set of constraints that I thought would cause this (I2C or SPI buses which call ->remove on ->shutdown) do not apply. But there is one more which applies here. If the DSA master itself is on a bus that calls ->remove from ->shutdown (like dpaa2-eth, which is on the fsl-mc bus), there is a device link between the switch and the DSA master, and device_links_unbind_consumers() will unbind the felix switch driver on shutdown. So the same treatment must be applied to all DSA switch drivers, which is: either use devres for both the mdiobus allocation and registration, or don't use devres at all. The felix driver has the code structure in place for orderly mdiobus removal, so just replace devm_mdiobus_alloc_size() with the non-devres variant, and add manual free where necessary, to ensure that we don't let devres free a still-registered bus. Fixes: ac3a68d56651 ("net: phy: don't abuse devres in devm_mdiobus_register()") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-07 18:15:50 +02:00
bus = mdiobus_alloc_size(sizeof(*mdio_priv));
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
if (!bus)
return -ENOMEM;
bus->name = "VSC9959 internal MDIO bus";
bus->read = enetc_mdio_read;
bus->write = enetc_mdio_write;
bus->parent = dev;
mdio_priv = bus->priv;
mdio_priv->hw = hw;
/* This gets added to imdio_regs, which already maps addresses
* starting with the proper offset.
*/
mdio_priv->mdio_base = 0;
snprintf(bus->id, MII_BUS_ID_SIZE, "%s-imdio", dev_name(dev));
/* Needed in order to initialize the bus mutex lock */
rc = mdiobus_register(bus);
if (rc < 0) {
dev_err(dev, "failed to register MDIO bus\n");
net: dsa: felix: don't use devres for mdiobus As explained in commits: 74b6d7d13307 ("net: dsa: realtek: register the MDIO bus under devres") 5135e96a3dd2 ("net: dsa: don't allocate the slave_mii_bus using devres") mdiobus_free() will panic when called from devm_mdiobus_free() <- devres_release_all() <- __device_release_driver(), and that mdiobus was not previously unregistered. The Felix VSC9959 switch is a PCI device, so the initial set of constraints that I thought would cause this (I2C or SPI buses which call ->remove on ->shutdown) do not apply. But there is one more which applies here. If the DSA master itself is on a bus that calls ->remove from ->shutdown (like dpaa2-eth, which is on the fsl-mc bus), there is a device link between the switch and the DSA master, and device_links_unbind_consumers() will unbind the felix switch driver on shutdown. So the same treatment must be applied to all DSA switch drivers, which is: either use devres for both the mdiobus allocation and registration, or don't use devres at all. The felix driver has the code structure in place for orderly mdiobus removal, so just replace devm_mdiobus_alloc_size() with the non-devres variant, and add manual free where necessary, to ensure that we don't let devres free a still-registered bus. Fixes: ac3a68d56651 ("net: phy: don't abuse devres in devm_mdiobus_register()") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-07 18:15:50 +02:00
mdiobus_free(bus);
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
return rc;
}
felix->imdio = bus;
for (port = 0; port < felix->info->num_ports; port++) {
struct ocelot_port *ocelot_port = ocelot->ports[port];
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
struct phylink_pcs *phylink_pcs;
net: dsa: felix: name change for clarity from pcs to mdio_device Simple rename of a variable to make things more logical. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:07 -08:00
struct mdio_device *mdio_device;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
net: dsa: ocelot: use the Lynx PCS helpers in Felix and Seville Use the helper functions introduced by the newly added Lynx PCS MDIO module in the Felix VSC9959 and Seville VSC9953. Instead of representing the PCS as a phy_device, a mdio_device structure will be passed to the Lynx module which is now actually implementing all the PCS configuration and status reporting. All code previously used for PCS monitoring and runtime configuration is removed and replaced will calls to the Lynx PCS operations. Tested on the following SERDES protocols of LS1028A: 0x7777 (2500Base-X), 0x85bb (QSGMII), 0x9999 (SGMII) and 0x13bb (USXGMII). Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-08-30 11:34:02 +03:00
if (dsa_is_unused_port(felix->ds, port))
continue;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
net: dsa: ocelot: use the Lynx PCS helpers in Felix and Seville Use the helper functions introduced by the newly added Lynx PCS MDIO module in the Felix VSC9959 and Seville VSC9953. Instead of representing the PCS as a phy_device, a mdio_device structure will be passed to the Lynx module which is now actually implementing all the PCS configuration and status reporting. All code previously used for PCS monitoring and runtime configuration is removed and replaced will calls to the Lynx PCS operations. Tested on the following SERDES protocols of LS1028A: 0x7777 (2500Base-X), 0x85bb (QSGMII), 0x9999 (SGMII) and 0x13bb (USXGMII). Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-08-30 11:34:02 +03:00
if (ocelot_port->phy_mode == PHY_INTERFACE_MODE_INTERNAL)
continue;
net: dsa: felix: name change for clarity from pcs to mdio_device Simple rename of a variable to make things more logical. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:07 -08:00
mdio_device = mdio_device_create(felix->imdio, port);
if (IS_ERR(mdio_device))
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
continue;
net: dsa: felix: name change for clarity from pcs to mdio_device Simple rename of a variable to make things more logical. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:07 -08:00
phylink_pcs = lynx_pcs_create(mdio_device);
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
if (!phylink_pcs) {
net: dsa: felix: name change for clarity from pcs to mdio_device Simple rename of a variable to make things more logical. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:07 -08:00
mdio_device_free(mdio_device);
net: dsa: ocelot: use the Lynx PCS helpers in Felix and Seville Use the helper functions introduced by the newly added Lynx PCS MDIO module in the Felix VSC9959 and Seville VSC9953. Instead of representing the PCS as a phy_device, a mdio_device structure will be passed to the Lynx module which is now actually implementing all the PCS configuration and status reporting. All code previously used for PCS monitoring and runtime configuration is removed and replaced will calls to the Lynx PCS operations. Tested on the following SERDES protocols of LS1028A: 0x7777 (2500Base-X), 0x85bb (QSGMII), 0x9999 (SGMII) and 0x13bb (USXGMII). Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-08-30 11:34:02 +03:00
continue;
}
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
felix->pcs[port] = phylink_pcs;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
dev_info(dev, "Found PCS at internal MDIO address %d\n", port);
}
return 0;
}
net: dsa: seville: duplicate vsc9959_mdio_bus_free While we don't plan on making any changes to this function, currently this is the only remaining dependency between felix and seville, after the PCS has been refactored out into pcs-lynx.c. Duplicate this function in seville to break the dependency completely. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:50 +03:00
static void vsc9959_mdio_bus_free(struct ocelot *ocelot)
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
{
struct felix *felix = ocelot_to_felix(ocelot);
int port;
for (port = 0; port < ocelot->num_phys_ports; port++) {
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
struct phylink_pcs *phylink_pcs = felix->pcs[port];
struct mdio_device *mdio_device;
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
if (!phylink_pcs)
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
continue;
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs Remove references to lynx_pcs structures so drivers like the Felix DSA can reference alternate PCS drivers. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-28 21:03:06 -08:00
mdio_device = lynx_get_mdio_device(phylink_pcs);
mdio_device_free(mdio_device);
lynx_pcs_destroy(phylink_pcs);
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
}
mdiobus_unregister(felix->imdio);
net: dsa: felix: don't use devres for mdiobus As explained in commits: 74b6d7d13307 ("net: dsa: realtek: register the MDIO bus under devres") 5135e96a3dd2 ("net: dsa: don't allocate the slave_mii_bus using devres") mdiobus_free() will panic when called from devm_mdiobus_free() <- devres_release_all() <- __device_release_driver(), and that mdiobus was not previously unregistered. The Felix VSC9959 switch is a PCI device, so the initial set of constraints that I thought would cause this (I2C or SPI buses which call ->remove on ->shutdown) do not apply. But there is one more which applies here. If the DSA master itself is on a bus that calls ->remove from ->shutdown (like dpaa2-eth, which is on the fsl-mc bus), there is a device link between the switch and the DSA master, and device_links_unbind_consumers() will unbind the felix switch driver on shutdown. So the same treatment must be applied to all DSA switch drivers, which is: either use devres for both the mdiobus allocation and registration, or don't use devres at all. The felix driver has the code structure in place for orderly mdiobus removal, so just replace devm_mdiobus_alloc_size() with the non-devres variant, and add manual free where necessary, to ensure that we don't let devres free a still-registered bus. Fixes: ac3a68d56651 ("net: phy: don't abuse devres in devm_mdiobus_register()") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-07 18:15:50 +02:00
mdiobus_free(felix->imdio);
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
}
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
/* Extract shortest continuous gate open intervals in ns for each traffic class
* of a cyclic tc-taprio schedule. If a gate is always open, the duration is
* considered U64_MAX. If the gate is always closed, it is considered 0.
*/
static void vsc9959_tas_min_gate_lengths(struct tc_taprio_qopt_offload *taprio,
u64 min_gate_len[OCELOT_NUM_TC])
{
struct tc_taprio_sched_entry *entry;
u64 gate_len[OCELOT_NUM_TC];
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed min_gate_len[tc] is supposed to track the shortest interval of continuously open gates for a traffic class. For example, in the following case: TC 76543210 t0 00000001b 200000 ns t1 00000010b 200000 ns min_gate_len[0] and min_gate_len[1] should be 200000, while min_gate_len[2-7] should be 0. However what happens is that min_gate_len[0] is 200000, but min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the point where the logic detects the gate close event for TC 1). The problem is that the code considers a "gate close" event whenever it sees that there is a 0 for that TC (essentially it's level rather than edge triggered). By doing that, any time a gate is seen as closed without having been open prior, gate_len, which is 0, will be written into min_gate_len. Once min_gate_len becomes 0, it's impossible for it to track anything higher than that (the length of actually open intervals). To fix this, we make the writing to min_gate_len[tc] be edge-triggered, which avoids writes for gates that are closed in consecutive intervals. However what this does is it makes us need to special-case the permanently closed gates at the end. Fixes: 55a515b1f5a9 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220804202817.1677572-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-04 23:28:17 +03:00
u8 gates_ever_opened = 0;
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
int tc, i, n;
/* Initialize arrays */
for (tc = 0; tc < OCELOT_NUM_TC; tc++) {
min_gate_len[tc] = U64_MAX;
gate_len[tc] = 0;
}
/* If we don't have taprio, consider all gates as permanently open */
if (!taprio)
return;
n = taprio->num_entries;
/* Walk through the gate list twice to determine the length
* of consecutively open gates for a traffic class, including
* open gates that wrap around. We are just interested in the
* minimum window size, and this doesn't change what the
* minimum is (if the gate never closes, min_gate_len will
* remain U64_MAX).
*/
for (i = 0; i < 2 * n; i++) {
entry = &taprio->entries[i % n];
for (tc = 0; tc < OCELOT_NUM_TC; tc++) {
if (entry->gate_mask & BIT(tc)) {
gate_len[tc] += entry->interval;
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed min_gate_len[tc] is supposed to track the shortest interval of continuously open gates for a traffic class. For example, in the following case: TC 76543210 t0 00000001b 200000 ns t1 00000010b 200000 ns min_gate_len[0] and min_gate_len[1] should be 200000, while min_gate_len[2-7] should be 0. However what happens is that min_gate_len[0] is 200000, but min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the point where the logic detects the gate close event for TC 1). The problem is that the code considers a "gate close" event whenever it sees that there is a 0 for that TC (essentially it's level rather than edge triggered). By doing that, any time a gate is seen as closed without having been open prior, gate_len, which is 0, will be written into min_gate_len. Once min_gate_len becomes 0, it's impossible for it to track anything higher than that (the length of actually open intervals). To fix this, we make the writing to min_gate_len[tc] be edge-triggered, which avoids writes for gates that are closed in consecutive intervals. However what this does is it makes us need to special-case the permanently closed gates at the end. Fixes: 55a515b1f5a9 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220804202817.1677572-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-04 23:28:17 +03:00
gates_ever_opened |= BIT(tc);
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
} else {
/* Gate closes now, record a potential new
* minimum and reinitialize length
*/
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed min_gate_len[tc] is supposed to track the shortest interval of continuously open gates for a traffic class. For example, in the following case: TC 76543210 t0 00000001b 200000 ns t1 00000010b 200000 ns min_gate_len[0] and min_gate_len[1] should be 200000, while min_gate_len[2-7] should be 0. However what happens is that min_gate_len[0] is 200000, but min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the point where the logic detects the gate close event for TC 1). The problem is that the code considers a "gate close" event whenever it sees that there is a 0 for that TC (essentially it's level rather than edge triggered). By doing that, any time a gate is seen as closed without having been open prior, gate_len, which is 0, will be written into min_gate_len. Once min_gate_len becomes 0, it's impossible for it to track anything higher than that (the length of actually open intervals). To fix this, we make the writing to min_gate_len[tc] be edge-triggered, which avoids writes for gates that are closed in consecutive intervals. However what this does is it makes us need to special-case the permanently closed gates at the end. Fixes: 55a515b1f5a9 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220804202817.1677572-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-04 23:28:17 +03:00
if (min_gate_len[tc] > gate_len[tc] &&
gate_len[tc])
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
min_gate_len[tc] = gate_len[tc];
gate_len[tc] = 0;
}
}
}
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed min_gate_len[tc] is supposed to track the shortest interval of continuously open gates for a traffic class. For example, in the following case: TC 76543210 t0 00000001b 200000 ns t1 00000010b 200000 ns min_gate_len[0] and min_gate_len[1] should be 200000, while min_gate_len[2-7] should be 0. However what happens is that min_gate_len[0] is 200000, but min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the point where the logic detects the gate close event for TC 1). The problem is that the code considers a "gate close" event whenever it sees that there is a 0 for that TC (essentially it's level rather than edge triggered). By doing that, any time a gate is seen as closed without having been open prior, gate_len, which is 0, will be written into min_gate_len. Once min_gate_len becomes 0, it's impossible for it to track anything higher than that (the length of actually open intervals). To fix this, we make the writing to min_gate_len[tc] be edge-triggered, which avoids writes for gates that are closed in consecutive intervals. However what this does is it makes us need to special-case the permanently closed gates at the end. Fixes: 55a515b1f5a9 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220804202817.1677572-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-04 23:28:17 +03:00
/* min_gate_len[tc] actually tracks minimum *open* gate time, so for
* permanently closed gates, min_gate_len[tc] will still be U64_MAX.
* Therefore they are currently indistinguishable from permanently
* open gates. Overwrite the gate len with 0 when we know they're
* actually permanently closed, i.e. after the loop above.
*/
for (tc = 0; tc < OCELOT_NUM_TC; tc++)
if (!(gates_ever_opened & BIT(tc)))
min_gate_len[tc] = 0;
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
}
/* Update QSYS_PORT_MAX_SDU to make sure the static guard bands added by the
* switch (see the ALWAYS_GUARD_BAND_SCH_Q comment) are correct at all MTU
* values (the default value is 1518). Also, for traffic class windows smaller
* than one MTU sized frame, update QSYS_QMAXSDU_CFG to enable oversized frame
* dropping, such that these won't hang the port, as they will never be sent.
*/
static void vsc9959_tas_guard_bands_update(struct ocelot *ocelot, int port)
{
struct ocelot_port *ocelot_port = ocelot->ports[port];
u64 min_gate_len[OCELOT_NUM_TC];
int speed, picos_per_byte;
u64 needed_bit_time_ps;
u32 val, maxlen;
u8 tas_speed;
int tc;
lockdep_assert_held(&ocelot->tas_lock);
val = ocelot_read_rix(ocelot, QSYS_TAG_CONFIG, port);
tas_speed = QSYS_TAG_CONFIG_LINK_SPEED_X(val);
switch (tas_speed) {
case OCELOT_SPEED_10:
speed = SPEED_10;
break;
case OCELOT_SPEED_100:
speed = SPEED_100;
break;
case OCELOT_SPEED_1000:
speed = SPEED_1000;
break;
case OCELOT_SPEED_2500:
speed = SPEED_2500;
break;
default:
return;
}
picos_per_byte = (USEC_PER_SEC * 8) / speed;
val = ocelot_port_readl(ocelot_port, DEV_MAC_MAXLEN_CFG);
/* MAXLEN_CFG accounts automatically for VLAN. We need to include it
* manually in the bit time calculation, plus the preamble and SFD.
*/
maxlen = val + 2 * VLAN_HLEN;
/* Consider the standard Ethernet overhead of 8 octets preamble+SFD,
* 4 octets FCS, 12 octets IFG.
*/
needed_bit_time_ps = (maxlen + 24) * picos_per_byte;
dev_dbg(ocelot->dev,
"port %d: max frame size %d needs %llu ps at speed %d\n",
port, maxlen, needed_bit_time_ps, speed);
vsc9959_tas_min_gate_lengths(ocelot_port->taprio, min_gate_len);
for (tc = 0; tc < OCELOT_NUM_TC; tc++) {
u32 max_sdu;
if (min_gate_len[tc] == U64_MAX /* Gate always open */ ||
time64.h: consolidate uses of PSEC_PER_NSEC Time-sensitive networking code needs to work with PTP times expressed in nanoseconds, and with packet transmission times expressed in picoseconds, since those would be fractional at higher than gigabit speed when expressed in nanoseconds. Convert the existing uses in tc-taprio and the ocelot/felix DSA driver to a PSEC_PER_NSEC macro. This macro is placed in include/linux/time64.h as opposed to its relatives (PSEC_PER_SEC etc) from include/vdso/time64.h because the vDSO library does not (yet) need/use it. Cc: Andy Lutomirski <luto@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vincenzo Frascino <vincenzo.frascino@arm.com> # for the vDSO parts Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:38 +03:00
min_gate_len[tc] * PSEC_PER_NSEC > needed_bit_time_ps) {
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
/* Setting QMAXSDU_CFG to 0 disables oversized frame
* dropping.
*/
max_sdu = 0;
dev_dbg(ocelot->dev,
"port %d tc %d min gate len %llu"
", sending all frames\n",
port, tc, min_gate_len[tc]);
} else {
/* If traffic class doesn't support a full MTU sized
* frame, make sure to enable oversize frame dropping
* for frames larger than the smallest that would fit.
*/
time64.h: consolidate uses of PSEC_PER_NSEC Time-sensitive networking code needs to work with PTP times expressed in nanoseconds, and with packet transmission times expressed in picoseconds, since those would be fractional at higher than gigabit speed when expressed in nanoseconds. Convert the existing uses in tc-taprio and the ocelot/felix DSA driver to a PSEC_PER_NSEC macro. This macro is placed in include/linux/time64.h as opposed to its relatives (PSEC_PER_SEC etc) from include/vdso/time64.h because the vDSO library does not (yet) need/use it. Cc: Andy Lutomirski <luto@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vincenzo Frascino <vincenzo.frascino@arm.com> # for the vDSO parts Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:38 +03:00
max_sdu = div_u64(min_gate_len[tc] * PSEC_PER_NSEC,
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
picos_per_byte);
/* A TC gate may be completely closed, which is a
* special case where all packets are oversized.
* Any limit smaller than 64 octets accomplishes this
*/
if (!max_sdu)
max_sdu = 1;
/* Take L1 overhead into account, but just don't allow
* max_sdu to go negative or to 0. Here we use 20
* because QSYS_MAXSDU_CFG_* already counts the 4 FCS
* octets as part of packet size.
*/
if (max_sdu > 20)
max_sdu -= 20;
dev_info(ocelot->dev,
"port %d tc %d min gate length %llu"
" ns not enough for max frame size %d at %d"
" Mbps, dropping frames over %d"
" octets including FCS\n",
port, tc, min_gate_len[tc], maxlen, speed,
max_sdu);
}
/* ocelot_write_rix is a macro that concatenates
* QSYS_MAXSDU_CFG_* with _RSZ, so we need to spell out
* the writes to each traffic class
*/
switch (tc) {
case 0:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_0,
port);
break;
case 1:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_1,
port);
break;
case 2:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_2,
port);
break;
case 3:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_3,
port);
break;
case 4:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_4,
port);
break;
case 5:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_5,
port);
break;
case 6:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_6,
port);
break;
case 7:
ocelot_write_rix(ocelot, max_sdu, QSYS_QMAXSDU_CFG_7,
port);
break;
}
}
ocelot_write_rix(ocelot, maxlen, QSYS_PORT_MAX_SDU, port);
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
static void vsc9959_sched_speed_set(struct ocelot *ocelot, int port,
u32 speed)
{
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
struct ocelot_port *ocelot_port = ocelot->ports[port];
net: dsa: felix: convert TAS link speed based on phylink speed state->speed holds a value of 10, 100, 1000 or 2500, but QSYS_TAG_CONFIG_LINK_SPEED expects a value of 0, 1, 2, 3. So convert the speed to a proper value. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-24 09:57:46 +08:00
u8 tas_speed;
switch (speed) {
case SPEED_10:
tas_speed = OCELOT_SPEED_10;
break;
case SPEED_100:
tas_speed = OCELOT_SPEED_100;
break;
case SPEED_1000:
tas_speed = OCELOT_SPEED_1000;
break;
case SPEED_2500:
tas_speed = OCELOT_SPEED_2500;
break;
default:
tas_speed = OCELOT_SPEED_1000;
break;
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
ocelot_rmw_rix(ocelot,
net: dsa: felix: convert TAS link speed based on phylink speed state->speed holds a value of 10, 100, 1000 or 2500, but QSYS_TAG_CONFIG_LINK_SPEED expects a value of 0, 1, 2, 3. So convert the speed to a proper value. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-24 09:57:46 +08:00
QSYS_TAG_CONFIG_LINK_SPEED(tas_speed),
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
QSYS_TAG_CONFIG_LINK_SPEED_M,
QSYS_TAG_CONFIG, port);
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
mutex_lock(&ocelot->tas_lock);
if (ocelot_port->taprio)
vsc9959_tas_guard_bands_update(ocelot, port);
mutex_unlock(&ocelot->tas_lock);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
}
static void vsc9959_new_base_time(struct ocelot *ocelot, ktime_t base_time,
u64 cycle_time,
struct timespec64 *new_base_ts)
{
struct timespec64 ts;
ktime_t new_base_time;
ktime_t current_time;
ocelot_ptp_gettime64(&ocelot->ptp_info, &ts);
current_time = timespec64_to_ktime(ts);
new_base_time = base_time;
if (base_time < current_time) {
u64 nr_of_cycles = current_time - base_time;
do_div(nr_of_cycles, cycle_time);
new_base_time += cycle_time * (nr_of_cycles + 1);
}
*new_base_ts = ktime_to_timespec64(new_base_time);
}
static u32 vsc9959_tas_read_cfg_status(struct ocelot *ocelot)
{
return ocelot_read(ocelot, QSYS_TAS_PARAM_CFG_CTRL);
}
static void vsc9959_tas_gcl_set(struct ocelot *ocelot, const u32 gcl_ix,
struct tc_taprio_sched_entry *entry)
{
ocelot_write(ocelot,
QSYS_GCL_CFG_REG_1_GCL_ENTRY_NUM(gcl_ix) |
QSYS_GCL_CFG_REG_1_GATE_STATE(entry->gate_mask),
QSYS_GCL_CFG_REG_1);
ocelot_write(ocelot, entry->interval, QSYS_GCL_CFG_REG_2);
}
static int vsc9959_qos_port_tas_set(struct ocelot *ocelot, int port,
struct tc_taprio_qopt_offload *taprio)
{
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
struct ocelot_port *ocelot_port = ocelot->ports[port];
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
struct timespec64 base_ts;
int ret, i;
u32 val;
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
mutex_lock(&ocelot->tas_lock);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
if (!taprio->enable) {
net: dsa: felix: keep QSYS_TAG_CONFIG_INIT_GATE_STATE(0xFF) out of rmw In vsc9959_tas_clock_adjust(), the INIT_GATE_STATE field is not changed, only the ENABLE field. Similarly for the disabling of the time-aware shaper in vsc9959_qos_port_tas_set(). To reflect this, keep the QSYS_TAG_CONFIG_INIT_GATE_STATE_M mask out of the read-modify-write procedure to make it clearer what is the intention of the code. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:36 +03:00
ocelot_rmw_rix(ocelot, 0, QSYS_TAG_CONFIG_ENABLE,
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
QSYS_TAG_CONFIG, port);
net: dsa: felix: keep reference on entire tc-taprio config In a future change we will need to remember the entire tc-taprio config on all ports rather than just the base time, so use the taprio_offload_get() helper function to replace ocelot_port->base_time with ocelot_port->taprio. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:35 +03:00
taprio_offload_free(ocelot_port->taprio);
ocelot_port->taprio = NULL;
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
vsc9959_tas_guard_bands_update(ocelot, port);
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
mutex_unlock(&ocelot->tas_lock);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
return 0;
}
if (taprio->cycle_time > NSEC_PER_SEC ||
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
taprio->cycle_time_extension >= NSEC_PER_SEC) {
ret = -EINVAL;
goto err;
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
if (taprio->num_entries > VSC9959_TAS_GCL_ENTRY_MAX) {
ret = -ERANGE;
goto err;
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
net: dsa: felix: re-enable TAS guard band mode Commit 316bcffe4479 ("net: dsa: felix: disable always guard band bit for TAS config") disabled the guard band and broke 802.3Qbv compliance. There are two issues here: (1) Without the guard band the end of the scheduling window could be overrun by a frame in transit. (2) Frames that don't fit into a configured window will still be sent. The reason for both issues is that the switch will schedule the _start_ of a frame transmission inside the predefined window without taking the length of the frame into account. Thus, we'll need the guard band which will close the gate early, so that a complete frame can still be sent. Revert the commit and add a note. For a lengthy discussion see [1]. [1] https://lore.kernel.org/netdev/c7618025da6723418c56a54fe4683bd7@walle.cc/ Fixes: 316bcffe4479 ("net: dsa: felix: disable always guard band bit for TAS config") Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-05-10 13:07:08 +02:00
/* Enable guard band. The switch will schedule frames without taking
* their length into account. Thus we'll always need to enable the
* guard band which reserves the time of a maximum sized frame at the
* end of the time window.
*
* Although the ALWAYS_GUARD_BAND_SCH_Q bit is global for all ports, we
* need to set PORT_NUM, because subsequent writes to PARAM_CFG_REG_n
* operate on the port number.
net: dsa: felix: disable always guard band bit for TAS config ALWAYS_GUARD_BAND_SCH_Q bit in TAS config register is descripted as this: 0: Guard band is implemented for nonschedule queues to schedule queues transition. 1: Guard band is implemented for any queue to schedule queue transition. The driver set guard band be implemented for any queue to schedule queue transition before, which will make each GCL time slot reserve a guard band time that can pass the max SDU frame. Because guard band time could not be set in tc-taprio now, it will use about 12000ns to pass 1500B max SDU. This limits each GCL time interval to be more than 12000ns. This patch change the guard band to be only implemented for nonschedule queues to schedule queues transition, so that there is no need to reserve guard band on each GCL. Users can manually add guard band time for each schedule queues in their configuration if they want. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-04-19 18:25:30 +08:00
*/
net: dsa: felix: re-enable TAS guard band mode Commit 316bcffe4479 ("net: dsa: felix: disable always guard band bit for TAS config") disabled the guard band and broke 802.3Qbv compliance. There are two issues here: (1) Without the guard band the end of the scheduling window could be overrun by a frame in transit. (2) Frames that don't fit into a configured window will still be sent. The reason for both issues is that the switch will schedule the _start_ of a frame transmission inside the predefined window without taking the length of the frame into account. Thus, we'll need the guard band which will close the gate early, so that a complete frame can still be sent. Revert the commit and add a note. For a lengthy discussion see [1]. [1] https://lore.kernel.org/netdev/c7618025da6723418c56a54fe4683bd7@walle.cc/ Fixes: 316bcffe4479 ("net: dsa: felix: disable always guard band bit for TAS config") Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-05-10 13:07:08 +02:00
ocelot_rmw(ocelot, QSYS_TAS_PARAM_CFG_CTRL_PORT_NUM(port) |
QSYS_TAS_PARAM_CFG_CTRL_ALWAYS_GUARD_BAND_SCH_Q,
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
QSYS_TAS_PARAM_CFG_CTRL_PORT_NUM_M |
QSYS_TAS_PARAM_CFG_CTRL_ALWAYS_GUARD_BAND_SCH_Q,
QSYS_TAS_PARAM_CFG_CTRL);
/* Hardware errata - Admin config could not be overwritten if
* config is pending, need reset the TAS module
*/
val = ocelot_read(ocelot, QSYS_PARAM_STATUS_REG_8);
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
if (val & QSYS_PARAM_STATUS_REG_8_CONFIG_PENDING) {
ret = -EBUSY;
goto err;
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
ocelot_rmw_rix(ocelot,
QSYS_TAG_CONFIG_ENABLE |
QSYS_TAG_CONFIG_INIT_GATE_STATE(0xFF) |
QSYS_TAG_CONFIG_SCH_TRAFFIC_QUEUES(0xFF),
QSYS_TAG_CONFIG_ENABLE |
QSYS_TAG_CONFIG_INIT_GATE_STATE_M |
QSYS_TAG_CONFIG_SCH_TRAFFIC_QUEUES_M,
QSYS_TAG_CONFIG, port);
vsc9959_new_base_time(ocelot, taprio->base_time,
taprio->cycle_time, &base_ts);
ocelot_write(ocelot, base_ts.tv_nsec, QSYS_PARAM_CFG_REG_1);
ocelot_write(ocelot, lower_32_bits(base_ts.tv_sec), QSYS_PARAM_CFG_REG_2);
val = upper_32_bits(base_ts.tv_sec);
ocelot_write(ocelot,
QSYS_PARAM_CFG_REG_3_BASE_TIME_SEC_MSB(val) |
QSYS_PARAM_CFG_REG_3_LIST_LENGTH(taprio->num_entries),
QSYS_PARAM_CFG_REG_3);
ocelot_write(ocelot, taprio->cycle_time, QSYS_PARAM_CFG_REG_4);
ocelot_write(ocelot, taprio->cycle_time_extension, QSYS_PARAM_CFG_REG_5);
for (i = 0; i < taprio->num_entries; i++)
vsc9959_tas_gcl_set(ocelot, i, &taprio->entries[i]);
ocelot_rmw(ocelot, QSYS_TAS_PARAM_CFG_CTRL_CONFIG_CHANGE,
QSYS_TAS_PARAM_CFG_CTRL_CONFIG_CHANGE,
QSYS_TAS_PARAM_CFG_CTRL);
ret = readx_poll_timeout(vsc9959_tas_read_cfg_status, ocelot, val,
!(val & QSYS_TAS_PARAM_CFG_CTRL_CONFIG_CHANGE),
10, 100000);
net: dsa: felix: keep reference on entire tc-taprio config In a future change we will need to remember the entire tc-taprio config on all ports rather than just the base time, so use the taprio_offload_get() helper function to replace ocelot_port->base_time with ocelot_port->taprio. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:35 +03:00
if (ret)
goto err;
ocelot_port->taprio = taprio_offload_get(taprio);
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
vsc9959_tas_guard_bands_update(ocelot, port);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
err:
mutex_unlock(&ocelot->tas_lock);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
return ret;
}
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
static void vsc9959_tas_clock_adjust(struct ocelot *ocelot)
{
net: dsa: felix: keep reference on entire tc-taprio config In a future change we will need to remember the entire tc-taprio config on all ports rather than just the base time, so use the taprio_offload_get() helper function to replace ocelot_port->base_time with ocelot_port->taprio. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:35 +03:00
struct tc_taprio_qopt_offload *taprio;
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
struct ocelot_port *ocelot_port;
struct timespec64 base_ts;
int port;
u32 val;
mutex_lock(&ocelot->tas_lock);
for (port = 0; port < ocelot->num_phys_ports; port++) {
net: dsa: felix: keep reference on entire tc-taprio config In a future change we will need to remember the entire tc-taprio config on all ports rather than just the base time, so use the taprio_offload_get() helper function to replace ocelot_port->base_time with ocelot_port->taprio. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:35 +03:00
ocelot_port = ocelot->ports[port];
taprio = ocelot_port->taprio;
if (!taprio)
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
continue;
ocelot_rmw(ocelot,
QSYS_TAS_PARAM_CFG_CTRL_PORT_NUM(port),
QSYS_TAS_PARAM_CFG_CTRL_PORT_NUM_M,
QSYS_TAS_PARAM_CFG_CTRL);
net: dsa: felix: keep QSYS_TAG_CONFIG_INIT_GATE_STATE(0xFF) out of rmw In vsc9959_tas_clock_adjust(), the INIT_GATE_STATE field is not changed, only the ENABLE field. Similarly for the disabling of the time-aware shaper in vsc9959_qos_port_tas_set(). To reflect this, keep the QSYS_TAG_CONFIG_INIT_GATE_STATE_M mask out of the read-modify-write procedure to make it clearer what is the intention of the code. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:36 +03:00
/* Disable time-aware shaper */
ocelot_rmw_rix(ocelot, 0, QSYS_TAG_CONFIG_ENABLE,
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
QSYS_TAG_CONFIG, port);
net: dsa: felix: keep reference on entire tc-taprio config In a future change we will need to remember the entire tc-taprio config on all ports rather than just the base time, so use the taprio_offload_get() helper function to replace ocelot_port->base_time with ocelot_port->taprio. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:35 +03:00
vsc9959_new_base_time(ocelot, taprio->base_time,
taprio->cycle_time, &base_ts);
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
ocelot_write(ocelot, base_ts.tv_nsec, QSYS_PARAM_CFG_REG_1);
ocelot_write(ocelot, lower_32_bits(base_ts.tv_sec),
QSYS_PARAM_CFG_REG_2);
val = upper_32_bits(base_ts.tv_sec);
ocelot_rmw(ocelot,
QSYS_PARAM_CFG_REG_3_BASE_TIME_SEC_MSB(val),
QSYS_PARAM_CFG_REG_3_BASE_TIME_SEC_MSB_M,
QSYS_PARAM_CFG_REG_3);
ocelot_rmw(ocelot, QSYS_TAS_PARAM_CFG_CTRL_CONFIG_CHANGE,
QSYS_TAS_PARAM_CFG_CTRL_CONFIG_CHANGE,
QSYS_TAS_PARAM_CFG_CTRL);
net: dsa: felix: keep QSYS_TAG_CONFIG_INIT_GATE_STATE(0xFF) out of rmw In vsc9959_tas_clock_adjust(), the INIT_GATE_STATE field is not changed, only the ENABLE field. Similarly for the disabling of the time-aware shaper in vsc9959_qos_port_tas_set(). To reflect this, keep the QSYS_TAG_CONFIG_INIT_GATE_STATE_M mask out of the read-modify-write procedure to make it clearer what is the intention of the code. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:36 +03:00
/* Re-enable time-aware shaper */
ocelot_rmw_rix(ocelot, QSYS_TAG_CONFIG_ENABLE,
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
QSYS_TAG_CONFIG_ENABLE,
QSYS_TAG_CONFIG, port);
}
mutex_unlock(&ocelot->tas_lock);
}
net: dsa: felix: add support Credit Based Shaper(CBS) for hardware offload VSC9959 hardware support the Credit Based Shaper(CBS) which part of the IEEE-802.1Qav. This patch support sch_cbs set for VSC9959. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:10 +08:00
static int vsc9959_qos_port_cbs_set(struct dsa_switch *ds, int port,
struct tc_cbs_qopt_offload *cbs_qopt)
{
struct ocelot *ocelot = ds->priv;
int port_ix = port * 8 + cbs_qopt->queue;
u32 rate, burst;
if (cbs_qopt->queue >= ds->num_tx_queues)
return -EINVAL;
if (!cbs_qopt->enable) {
ocelot_write_gix(ocelot, QSYS_CIR_CFG_CIR_RATE(0) |
QSYS_CIR_CFG_CIR_BURST(0),
QSYS_CIR_CFG, port_ix);
ocelot_rmw_gix(ocelot, 0, QSYS_SE_CFG_SE_AVB_ENA,
QSYS_SE_CFG, port_ix);
return 0;
}
/* Rate unit is 100 kbps */
rate = DIV_ROUND_UP(cbs_qopt->idleslope, 100);
/* Avoid using zero rate */
rate = clamp_t(u32, rate, 1, GENMASK(14, 0));
/* Burst unit is 4kB */
burst = DIV_ROUND_UP(cbs_qopt->hicredit, 4096);
/* Avoid using zero burst size */
net: dsa: felix: fix incorrect clamp calculation for burst Currently burst is clamping on rate and not burst, the assignment of burst from the clamping discards the previous assignment of burst. This looks like a cut-n-paste error from the previous clamping calculation on ramp. Fix this by replacing ramp with burst. Addresses-Coverity: ("Unused value") Fixes: 0fbabf875d18 ("net: dsa: felix: add support Credit Based Shaper(CBS) for hardware offload") Signed-off-by: Colin Ian King <colin.king@canonical.com> Acked-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-14 19:33:02 +01:00
burst = clamp_t(u32, burst, 1, GENMASK(5, 0));
net: dsa: felix: add support Credit Based Shaper(CBS) for hardware offload VSC9959 hardware support the Credit Based Shaper(CBS) which part of the IEEE-802.1Qav. This patch support sch_cbs set for VSC9959. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:10 +08:00
ocelot_write_gix(ocelot,
QSYS_CIR_CFG_CIR_RATE(rate) |
QSYS_CIR_CFG_CIR_BURST(burst),
QSYS_CIR_CFG,
port_ix);
ocelot_rmw_gix(ocelot,
QSYS_SE_CFG_SE_FRM_MODE(0) |
QSYS_SE_CFG_SE_AVB_ENA,
QSYS_SE_CFG_SE_AVB_ENA |
QSYS_SE_CFG_SE_FRM_MODE_M,
QSYS_SE_CFG,
port_ix);
return 0;
}
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
static int vsc9959_port_setup_tc(struct dsa_switch *ds, int port,
enum tc_setup_type type,
void *type_data)
{
struct ocelot *ocelot = ds->priv;
switch (type) {
case TC_SETUP_QDISC_TAPRIO:
return vsc9959_qos_port_tas_set(ocelot, port, type_data);
net: dsa: felix: add support Credit Based Shaper(CBS) for hardware offload VSC9959 hardware support the Credit Based Shaper(CBS) which part of the IEEE-802.1Qav. This patch support sch_cbs set for VSC9959. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:10 +08:00
case TC_SETUP_QDISC_CBS:
return vsc9959_qos_port_cbs_set(ds, port, type_data);
net: dsa: felix: Configure Time-Aware Scheduler via taprio offload Ocelot VSC9959 switch supports time-based egress shaping in hardware according to IEEE 802.1Qbv. This patch add support for TAS configuration on egress port of VSC9959 switch. Felix driver is an instance of Ocelot family, with a DSA front-end. The patch uses tc taprio hardware offload to setup TAS set function on felix driver. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-13 10:25:09 +08:00
default:
return -EOPNOTSUPP;
}
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
#define VSC9959_PSFP_SFID_MAX 175
#define VSC9959_PSFP_GATE_ID_MAX 183
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
#define VSC9959_PSFP_POLICER_BASE 63
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
#define VSC9959_PSFP_POLICER_MAX 383
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
#define VSC9959_PSFP_GATE_LIST_NUM 4
#define VSC9959_PSFP_GATE_CYCLETIME_MIN 5000
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct felix_stream {
struct list_head list;
unsigned long id;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
bool dummy;
int ports;
int port;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
u8 dmac[ETH_ALEN];
u16 vid;
s8 prio;
u8 sfid_valid;
u8 ssid_valid;
u32 sfid;
u32 ssid;
};
struct felix_stream_filter {
struct list_head list;
refcount_t refcount;
u32 index;
u8 enable;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
int portmask;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
u8 sg_valid;
u32 sgid;
u8 fm_valid;
u32 fmid;
u8 prio_valid;
u8 prio;
u32 maxsdu;
};
struct felix_stream_filter_counters {
u32 match;
u32 not_pass_gate;
u32 not_pass_sdu;
u32 red;
};
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
struct felix_stream_gate {
u32 index;
u8 enable;
u8 ipv_valid;
u8 init_ipv;
u64 basetime;
u64 cycletime;
u64 cycletime_ext;
u32 num_entries;
net: dsa: felix: fix flexible_array.cocci warnings Zero-length and one-element arrays are deprecated, see Documentation/process/deprecated.rst Flexible-array members should be used instead. Generated by: scripts/coccinelle/misc/flexible_array.cocci Fixes: 23ae3a787771 ("net: dsa: felix: add stream gate settings for psfp") CC: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: kernel test robot <lkp@intel.com> Signed-off-by: Julia Lawall <julia.lawall@inria.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-27 19:03:20 +01:00
struct action_gate_entry entries[];
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
};
struct felix_stream_gate_entry {
struct list_head list;
refcount_t refcount;
u32 index;
};
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
static int vsc9959_stream_identify(struct flow_cls_offload *f,
struct felix_stream *stream)
{
struct flow_rule *rule = flow_cls_offload_flow_rule(f);
struct flow_dissector *dissector = rule->match.dissector;
if (dissector->used_keys &
~(BIT(FLOW_DISSECTOR_KEY_CONTROL) |
BIT(FLOW_DISSECTOR_KEY_BASIC) |
BIT(FLOW_DISSECTOR_KEY_VLAN) |
BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS)))
return -EOPNOTSUPP;
if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
struct flow_match_eth_addrs match;
flow_rule_match_eth_addrs(rule, &match);
ether_addr_copy(stream->dmac, match.key->dst);
if (!is_zero_ether_addr(match.mask->src))
return -EOPNOTSUPP;
} else {
return -EOPNOTSUPP;
}
if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN)) {
struct flow_match_vlan match;
flow_rule_match_vlan(rule, &match);
if (match.mask->vlan_priority)
stream->prio = match.key->vlan_priority;
else
stream->prio = -1;
if (!match.mask->vlan_id)
return -EOPNOTSUPP;
stream->vid = match.key->vlan_id;
} else {
return -EOPNOTSUPP;
}
stream->id = f->cookie;
return 0;
}
static int vsc9959_mact_stream_set(struct ocelot *ocelot,
struct felix_stream *stream,
struct netlink_ext_ack *extack)
{
enum macaccess_entry_type type;
int ret, sfid, ssid;
u32 vid, dst_idx;
u8 mac[ETH_ALEN];
ether_addr_copy(mac, stream->dmac);
vid = stream->vid;
/* Stream identification desn't support to add a stream with non
* existent MAC (The MAC entry has not been learned in MAC table).
*/
ret = ocelot_mact_lookup(ocelot, &dst_idx, mac, vid, &type);
if (ret) {
if (extack)
NL_SET_ERR_MSG_MOD(extack, "Stream is not learned in MAC table");
return -EOPNOTSUPP;
}
if ((stream->sfid_valid || stream->ssid_valid) &&
type == ENTRYTYPE_NORMAL)
type = ENTRYTYPE_LOCKED;
sfid = stream->sfid_valid ? stream->sfid : -1;
ssid = stream->ssid_valid ? stream->ssid : -1;
ret = ocelot_mact_learn_streamdata(ocelot, dst_idx, mac, vid, type,
sfid, ssid);
return ret;
}
static struct felix_stream *
vsc9959_stream_table_lookup(struct list_head *stream_list,
struct felix_stream *stream)
{
struct felix_stream *tmp;
list_for_each_entry(tmp, stream_list, list)
if (ether_addr_equal(tmp->dmac, stream->dmac) &&
tmp->vid == stream->vid)
return tmp;
return NULL;
}
static int vsc9959_stream_table_add(struct ocelot *ocelot,
struct list_head *stream_list,
struct felix_stream *stream,
struct netlink_ext_ack *extack)
{
struct felix_stream *stream_entry;
int ret;
net: dsa: felix: use kmemdup() to replace kmalloc + memcpy Fix following coccicheck warning: /drivers/net/dsa/ocelot/felix_vsc9959.c:1627:13-20: WARNING opportunity for kmemdup /drivers/net/dsa/ocelot/felix_vsc9959.c:1506:16-23: WARNING opportunity for kmemdup Signed-off-by: Yihao Han <hanyihao@vivo.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20211207064419.38632-1-hanyihao@vivo.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-06 22:44:18 -08:00
stream_entry = kmemdup(stream, sizeof(*stream_entry), GFP_KERNEL);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
if (!stream_entry)
return -ENOMEM;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
if (!stream->dummy) {
ret = vsc9959_mact_stream_set(ocelot, stream_entry, extack);
if (ret) {
kfree(stream_entry);
return ret;
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
}
list_add_tail(&stream_entry->list, stream_list);
return 0;
}
static struct felix_stream *
vsc9959_stream_table_get(struct list_head *stream_list, unsigned long id)
{
struct felix_stream *tmp;
list_for_each_entry(tmp, stream_list, list)
if (tmp->id == id)
return tmp;
return NULL;
}
static void vsc9959_stream_table_del(struct ocelot *ocelot,
struct felix_stream *stream)
{
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
if (!stream->dummy)
vsc9959_mact_stream_set(ocelot, stream, NULL);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
list_del(&stream->list);
kfree(stream);
}
static u32 vsc9959_sfi_access_status(struct ocelot *ocelot)
{
return ocelot_read(ocelot, ANA_TABLES_SFIDACCESS);
}
static int vsc9959_psfp_sfi_set(struct ocelot *ocelot,
struct felix_stream_filter *sfi)
{
u32 val;
if (sfi->index > VSC9959_PSFP_SFID_MAX)
return -EINVAL;
if (!sfi->enable) {
ocelot_write(ocelot, ANA_TABLES_SFIDTIDX_SFID_INDEX(sfi->index),
ANA_TABLES_SFIDTIDX);
val = ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(SFIDACCESS_CMD_WRITE);
ocelot_write(ocelot, val, ANA_TABLES_SFIDACCESS);
return readx_poll_timeout(vsc9959_sfi_access_status, ocelot, val,
(!ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(val)),
10, 100000);
}
if (sfi->sgid > VSC9959_PSFP_GATE_ID_MAX ||
sfi->fmid > VSC9959_PSFP_POLICER_MAX)
return -EINVAL;
ocelot_write(ocelot,
(sfi->sg_valid ? ANA_TABLES_SFIDTIDX_SGID_VALID : 0) |
ANA_TABLES_SFIDTIDX_SGID(sfi->sgid) |
(sfi->fm_valid ? ANA_TABLES_SFIDTIDX_POL_ENA : 0) |
ANA_TABLES_SFIDTIDX_POL_IDX(sfi->fmid) |
ANA_TABLES_SFIDTIDX_SFID_INDEX(sfi->index),
ANA_TABLES_SFIDTIDX);
ocelot_write(ocelot,
(sfi->prio_valid ? ANA_TABLES_SFIDACCESS_IGR_PRIO_MATCH_ENA : 0) |
ANA_TABLES_SFIDACCESS_IGR_PRIO(sfi->prio) |
ANA_TABLES_SFIDACCESS_MAX_SDU_LEN(sfi->maxsdu) |
ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(SFIDACCESS_CMD_WRITE),
ANA_TABLES_SFIDACCESS);
return readx_poll_timeout(vsc9959_sfi_access_status, ocelot, val,
(!ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(val)),
10, 100000);
}
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
static int vsc9959_psfp_sfidmask_set(struct ocelot *ocelot, u32 sfid, int ports)
{
u32 val;
ocelot_rmw(ocelot,
ANA_TABLES_SFIDTIDX_SFID_INDEX(sfid),
ANA_TABLES_SFIDTIDX_SFID_INDEX_M,
ANA_TABLES_SFIDTIDX);
ocelot_write(ocelot,
ANA_TABLES_SFID_MASK_IGR_PORT_MASK(ports) |
ANA_TABLES_SFID_MASK_IGR_SRCPORT_MATCH_ENA,
ANA_TABLES_SFID_MASK);
ocelot_rmw(ocelot,
ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(SFIDACCESS_CMD_WRITE),
ANA_TABLES_SFIDACCESS_SFID_TBL_CMD_M,
ANA_TABLES_SFIDACCESS);
return readx_poll_timeout(vsc9959_sfi_access_status, ocelot, val,
(!ANA_TABLES_SFIDACCESS_SFID_TBL_CMD(val)),
10, 100000);
}
static int vsc9959_psfp_sfi_list_add(struct ocelot *ocelot,
struct felix_stream_filter *sfi,
struct list_head *pos)
{
struct felix_stream_filter *sfi_entry;
int ret;
net: dsa: felix: use kmemdup() to replace kmalloc + memcpy Fix following coccicheck warning: /drivers/net/dsa/ocelot/felix_vsc9959.c:1627:13-20: WARNING opportunity for kmemdup /drivers/net/dsa/ocelot/felix_vsc9959.c:1506:16-23: WARNING opportunity for kmemdup Signed-off-by: Yihao Han <hanyihao@vivo.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20211207064419.38632-1-hanyihao@vivo.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-06 22:44:18 -08:00
sfi_entry = kmemdup(sfi, sizeof(*sfi_entry), GFP_KERNEL);
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
if (!sfi_entry)
return -ENOMEM;
refcount_set(&sfi_entry->refcount, 1);
ret = vsc9959_psfp_sfi_set(ocelot, sfi_entry);
if (ret) {
kfree(sfi_entry);
return ret;
}
vsc9959_psfp_sfidmask_set(ocelot, sfi->index, sfi->portmask);
list_add(&sfi_entry->list, pos);
return 0;
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
static int vsc9959_psfp_sfi_table_add(struct ocelot *ocelot,
struct felix_stream_filter *sfi)
{
struct list_head *pos, *q, *last;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
struct felix_stream_filter *tmp;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct ocelot_psfp_list *psfp;
u32 insert = 0;
psfp = &ocelot->psfp;
last = &psfp->sfi_list;
list_for_each_safe(pos, q, &psfp->sfi_list) {
tmp = list_entry(pos, struct felix_stream_filter, list);
if (sfi->sg_valid == tmp->sg_valid &&
sfi->fm_valid == tmp->fm_valid &&
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
sfi->portmask == tmp->portmask &&
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
tmp->sgid == sfi->sgid &&
tmp->fmid == sfi->fmid) {
sfi->index = tmp->index;
refcount_inc(&tmp->refcount);
return 0;
}
/* Make sure that the index is increasing in order. */
if (tmp->index == insert) {
last = pos;
insert++;
}
}
sfi->index = insert;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
return vsc9959_psfp_sfi_list_add(ocelot, sfi, last);
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
static int vsc9959_psfp_sfi_table_add2(struct ocelot *ocelot,
struct felix_stream_filter *sfi,
struct felix_stream_filter *sfi2)
{
struct felix_stream_filter *tmp;
struct list_head *pos, *q, *last;
struct ocelot_psfp_list *psfp;
u32 insert = 0;
int ret;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
psfp = &ocelot->psfp;
last = &psfp->sfi_list;
list_for_each_safe(pos, q, &psfp->sfi_list) {
tmp = list_entry(pos, struct felix_stream_filter, list);
/* Make sure that the index is increasing in order. */
if (tmp->index >= insert + 2)
break;
insert = tmp->index + 1;
last = pos;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
}
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
sfi->index = insert;
ret = vsc9959_psfp_sfi_list_add(ocelot, sfi, last);
if (ret)
return ret;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
sfi2->index = insert + 1;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
return vsc9959_psfp_sfi_list_add(ocelot, sfi2, last->next);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
}
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
static struct felix_stream_filter *
vsc9959_psfp_sfi_table_get(struct list_head *sfi_list, u32 index)
{
struct felix_stream_filter *tmp;
list_for_each_entry(tmp, sfi_list, list)
if (tmp->index == index)
return tmp;
return NULL;
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
static void vsc9959_psfp_sfi_table_del(struct ocelot *ocelot, u32 index)
{
struct felix_stream_filter *tmp, *n;
struct ocelot_psfp_list *psfp;
u8 z;
psfp = &ocelot->psfp;
list_for_each_entry_safe(tmp, n, &psfp->sfi_list, list)
if (tmp->index == index) {
z = refcount_dec_and_test(&tmp->refcount);
if (z) {
tmp->enable = 0;
vsc9959_psfp_sfi_set(ocelot, tmp);
list_del(&tmp->list);
kfree(tmp);
}
break;
}
}
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
static void vsc9959_psfp_parse_gate(const struct flow_action_entry *entry,
struct felix_stream_gate *sgi)
{
flow_offload: add index to flow_action_entry structure Add index to flow_action_entry structure and delete index from police and gate child structure. We make this change to offload tc action for driver to identify a tc action. Signed-off-by: Baowen Zheng <baowen.zheng@corigine.com> Signed-off-by: Simon Horman <simon.horman@corigine.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-17 19:16:19 +01:00
sgi->index = entry->hw_index;
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
sgi->ipv_valid = (entry->gate.prio < 0) ? 0 : 1;
sgi->init_ipv = (sgi->ipv_valid) ? entry->gate.prio : 0;
sgi->basetime = entry->gate.basetime;
sgi->cycletime = entry->gate.cycletime;
sgi->num_entries = entry->gate.num_entries;
sgi->enable = 1;
memcpy(sgi->entries, entry->gate.entries,
entry->gate.num_entries * sizeof(struct action_gate_entry));
}
static u32 vsc9959_sgi_cfg_status(struct ocelot *ocelot)
{
return ocelot_read(ocelot, ANA_SG_ACCESS_CTRL);
}
static int vsc9959_psfp_sgi_set(struct ocelot *ocelot,
struct felix_stream_gate *sgi)
{
struct action_gate_entry *e;
struct timespec64 base_ts;
u32 interval_sum = 0;
u32 val;
int i;
if (sgi->index > VSC9959_PSFP_GATE_ID_MAX)
return -EINVAL;
ocelot_write(ocelot, ANA_SG_ACCESS_CTRL_SGID(sgi->index),
ANA_SG_ACCESS_CTRL);
if (!sgi->enable) {
ocelot_rmw(ocelot, ANA_SG_CONFIG_REG_3_INIT_GATE_STATE,
ANA_SG_CONFIG_REG_3_INIT_GATE_STATE |
ANA_SG_CONFIG_REG_3_GATE_ENABLE,
ANA_SG_CONFIG_REG_3);
return 0;
}
if (sgi->cycletime < VSC9959_PSFP_GATE_CYCLETIME_MIN ||
sgi->cycletime > NSEC_PER_SEC)
return -EINVAL;
if (sgi->num_entries > VSC9959_PSFP_GATE_LIST_NUM)
return -EINVAL;
vsc9959_new_base_time(ocelot, sgi->basetime, sgi->cycletime, &base_ts);
ocelot_write(ocelot, base_ts.tv_nsec, ANA_SG_CONFIG_REG_1);
val = lower_32_bits(base_ts.tv_sec);
ocelot_write(ocelot, val, ANA_SG_CONFIG_REG_2);
val = upper_32_bits(base_ts.tv_sec);
ocelot_write(ocelot,
(sgi->ipv_valid ? ANA_SG_CONFIG_REG_3_IPV_VALID : 0) |
ANA_SG_CONFIG_REG_3_INIT_IPV(sgi->init_ipv) |
ANA_SG_CONFIG_REG_3_GATE_ENABLE |
ANA_SG_CONFIG_REG_3_LIST_LENGTH(sgi->num_entries) |
ANA_SG_CONFIG_REG_3_INIT_GATE_STATE |
ANA_SG_CONFIG_REG_3_BASE_TIME_SEC_MSB(val),
ANA_SG_CONFIG_REG_3);
ocelot_write(ocelot, sgi->cycletime, ANA_SG_CONFIG_REG_4);
e = sgi->entries;
for (i = 0; i < sgi->num_entries; i++) {
u32 ips = (e[i].ipv < 0) ? 0 : (e[i].ipv + 8);
ocelot_write_rix(ocelot, ANA_SG_GCL_GS_CONFIG_IPS(ips) |
(e[i].gate_state ?
ANA_SG_GCL_GS_CONFIG_GATE_STATE : 0),
ANA_SG_GCL_GS_CONFIG, i);
interval_sum += e[i].interval;
ocelot_write_rix(ocelot, interval_sum, ANA_SG_GCL_TI_CONFIG, i);
}
ocelot_rmw(ocelot, ANA_SG_ACCESS_CTRL_CONFIG_CHANGE,
ANA_SG_ACCESS_CTRL_CONFIG_CHANGE,
ANA_SG_ACCESS_CTRL);
return readx_poll_timeout(vsc9959_sgi_cfg_status, ocelot, val,
(!(ANA_SG_ACCESS_CTRL_CONFIG_CHANGE & val)),
10, 100000);
}
static int vsc9959_psfp_sgi_table_add(struct ocelot *ocelot,
struct felix_stream_gate *sgi)
{
struct felix_stream_gate_entry *tmp;
struct ocelot_psfp_list *psfp;
int ret;
psfp = &ocelot->psfp;
list_for_each_entry(tmp, &psfp->sgi_list, list)
if (tmp->index == sgi->index) {
refcount_inc(&tmp->refcount);
return 0;
}
tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
if (!tmp)
return -ENOMEM;
ret = vsc9959_psfp_sgi_set(ocelot, sgi);
if (ret) {
kfree(tmp);
return ret;
}
tmp->index = sgi->index;
refcount_set(&tmp->refcount, 1);
list_add_tail(&tmp->list, &psfp->sgi_list);
return 0;
}
static void vsc9959_psfp_sgi_table_del(struct ocelot *ocelot,
u32 index)
{
struct felix_stream_gate_entry *tmp, *n;
struct felix_stream_gate sgi = {0};
struct ocelot_psfp_list *psfp;
u8 z;
psfp = &ocelot->psfp;
list_for_each_entry_safe(tmp, n, &psfp->sgi_list, list)
if (tmp->index == index) {
z = refcount_dec_and_test(&tmp->refcount);
if (z) {
sgi.index = index;
sgi.enable = 0;
vsc9959_psfp_sgi_set(ocelot, &sgi);
list_del(&tmp->list);
kfree(tmp);
}
break;
}
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
static void vsc9959_psfp_counters_get(struct ocelot *ocelot, u32 index,
struct felix_stream_filter_counters *counters)
{
net: mscc: ocelot: turn stats_lock into a spinlock ocelot_get_stats64() currently runs unlocked and therefore may collide with ocelot_port_update_stats() which indirectly accesses the same counters. However, ocelot_get_stats64() runs in atomic context, and we cannot simply take the sleepable ocelot->stats_lock mutex. We need to convert it to an atomic spinlock first. Do that as a preparatory change. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:48 +03:00
spin_lock(&ocelot->stats_lock);
net: dsa: felix: fix race between reading PSFP stats and port stats Both PSFP stats and the port stats read by ocelot_check_stats_work() are indirectly read through the same mechanism - write to STAT_CFG:STAT_VIEW, read from SYS:STAT:CNT[n]. It's just that for port stats, we write STAT_VIEW with the index of the port, and for PSFP stats, we write STAT_VIEW with the filter index. So if we allow them to run concurrently, ocelot_check_stats_work() may change the view from vsc9959_psfp_counters_get(), and vice versa. Fixes: 7d4b564d6add ("net: dsa: felix: support psfp filter on vsc9959") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220629183007.3808130-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-29 21:30:07 +03:00
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
ocelot_rmw(ocelot, SYS_STAT_CFG_STAT_VIEW(index),
SYS_STAT_CFG_STAT_VIEW_M,
SYS_STAT_CFG);
counters->match = ocelot_read_gix(ocelot, SYS_CNT, 0x200);
counters->not_pass_gate = ocelot_read_gix(ocelot, SYS_CNT, 0x201);
counters->not_pass_sdu = ocelot_read_gix(ocelot, SYS_CNT, 0x202);
counters->red = ocelot_read_gix(ocelot, SYS_CNT, 0x203);
/* Clear the PSFP counter. */
ocelot_write(ocelot,
SYS_STAT_CFG_STAT_VIEW(index) |
SYS_STAT_CFG_STAT_CLEAR_SHOT(0x10),
SYS_STAT_CFG);
net: dsa: felix: fix race between reading PSFP stats and port stats Both PSFP stats and the port stats read by ocelot_check_stats_work() are indirectly read through the same mechanism - write to STAT_CFG:STAT_VIEW, read from SYS:STAT:CNT[n]. It's just that for port stats, we write STAT_VIEW with the index of the port, and for PSFP stats, we write STAT_VIEW with the filter index. So if we allow them to run concurrently, ocelot_check_stats_work() may change the view from vsc9959_psfp_counters_get(), and vice versa. Fixes: 7d4b564d6add ("net: dsa: felix: support psfp filter on vsc9959") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220629183007.3808130-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-29 21:30:07 +03:00
net: mscc: ocelot: turn stats_lock into a spinlock ocelot_get_stats64() currently runs unlocked and therefore may collide with ocelot_port_update_stats() which indirectly accesses the same counters. However, ocelot_get_stats64() runs in atomic context, and we cannot simply take the sleepable ocelot->stats_lock mutex. We need to convert it to an atomic spinlock first. Do that as a preparatory change. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-16 16:53:48 +03:00
spin_unlock(&ocelot->stats_lock);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
}
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
static int vsc9959_psfp_filter_add(struct ocelot *ocelot, int port,
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct flow_cls_offload *f)
{
struct netlink_ext_ack *extack = f->common.extack;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
struct felix_stream_filter old_sfi, *sfi_entry;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct felix_stream_filter sfi = {0};
const struct flow_action_entry *a;
struct felix_stream *stream_entry;
struct felix_stream stream = {0};
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
struct felix_stream_gate *sgi;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct ocelot_psfp_list *psfp;
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
struct ocelot_policer pol;
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
int ret, i, size;
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
u64 rate, burst;
u32 index;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
psfp = &ocelot->psfp;
ret = vsc9959_stream_identify(f, &stream);
if (ret) {
NL_SET_ERR_MSG_MOD(extack, "Only can match on VID, PCP, and dest MAC");
return ret;
}
flow_action_for_each(i, a, &f->rule->action) {
switch (a->id) {
case FLOW_ACTION_GATE:
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
size = struct_size(sgi, entries, a->gate.num_entries);
sgi = kzalloc(size, GFP_KERNEL);
net: dsa: felix: fix possible NULL pointer dereference As the possible failure of the allocation, kzalloc() may return NULL pointer. Therefore, it should be better to check the 'sgi' in order to prevent the dereference of NULL pointer. Fixes: 23ae3a7877718 ("net: dsa: felix: add stream gate settings for psfp"). Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220329090800.130106-1-zhengyongjun3@huawei.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-03-29 09:08:00 +00:00
if (!sgi) {
ret = -ENOMEM;
goto err;
}
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
vsc9959_psfp_parse_gate(a, sgi);
ret = vsc9959_psfp_sgi_table_add(ocelot, sgi);
if (ret) {
kfree(sgi);
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
goto err;
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
}
sfi.sg_valid = 1;
sfi.sgid = sgi->index;
kfree(sgi);
break;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
case FLOW_ACTION_POLICE:
flow_offload: add index to flow_action_entry structure Add index to flow_action_entry structure and delete index from police and gate child structure. We make this change to offload tc action for driver to identify a tc action. Signed-off-by: Baowen Zheng <baowen.zheng@corigine.com> Signed-off-by: Simon Horman <simon.horman@corigine.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-17 19:16:19 +01:00
index = a->hw_index + VSC9959_PSFP_POLICER_BASE;
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
if (index > VSC9959_PSFP_POLICER_MAX) {
ret = -EINVAL;
goto err;
}
rate = a->police.rate_bytes_ps;
burst = rate * PSCHED_NS2TICKS(a->police.burst);
pol = (struct ocelot_policer) {
.burst = div_u64(burst, PSCHED_TICKS_PER_SEC),
.rate = div_u64(rate, 1000) * 8,
};
ret = ocelot_vcap_policer_add(ocelot, index, &pol);
if (ret)
goto err;
sfi.fm_valid = 1;
sfi.fmid = index;
sfi.maxsdu = a->police.mtu;
break;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
default:
return -EOPNOTSUPP;
}
}
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
stream.ports = BIT(port);
stream.port = port;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
sfi.portmask = stream.ports;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
sfi.prio_valid = (stream.prio < 0 ? 0 : 1);
sfi.prio = (sfi.prio_valid ? stream.prio : 0);
sfi.enable = 1;
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
/* Check if stream is set. */
stream_entry = vsc9959_stream_table_lookup(&psfp->stream_list, &stream);
if (stream_entry) {
if (stream_entry->ports & BIT(port)) {
NL_SET_ERR_MSG_MOD(extack,
"The stream is added on this port");
ret = -EEXIST;
goto err;
}
if (stream_entry->ports != BIT(stream_entry->port)) {
NL_SET_ERR_MSG_MOD(extack,
"The stream is added on two ports");
ret = -EEXIST;
goto err;
}
stream_entry->ports |= BIT(port);
stream.ports = stream_entry->ports;
sfi_entry = vsc9959_psfp_sfi_table_get(&psfp->sfi_list,
stream_entry->sfid);
memcpy(&old_sfi, sfi_entry, sizeof(old_sfi));
vsc9959_psfp_sfi_table_del(ocelot, stream_entry->sfid);
old_sfi.portmask = stream_entry->ports;
sfi.portmask = stream.ports;
if (stream_entry->port > port) {
ret = vsc9959_psfp_sfi_table_add2(ocelot, &sfi,
&old_sfi);
stream_entry->dummy = true;
} else {
ret = vsc9959_psfp_sfi_table_add2(ocelot, &old_sfi,
&sfi);
stream.dummy = true;
}
if (ret)
goto err;
stream_entry->sfid = old_sfi.index;
} else {
ret = vsc9959_psfp_sfi_table_add(ocelot, &sfi);
if (ret)
goto err;
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
stream.sfid = sfi.index;
stream.sfid_valid = 1;
ret = vsc9959_stream_table_add(ocelot, &psfp->stream_list,
&stream, extack);
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
if (ret) {
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
vsc9959_psfp_sfi_table_del(ocelot, stream.sfid);
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
goto err;
}
return 0;
err:
if (sfi.sg_valid)
vsc9959_psfp_sgi_table_del(ocelot, sfi.sgid);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
if (sfi.fm_valid)
ocelot_vcap_policer_del(ocelot, sfi.fmid);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
return ret;
}
static int vsc9959_psfp_filter_del(struct ocelot *ocelot,
struct flow_cls_offload *f)
{
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
struct felix_stream *stream, tmp, *stream_entry;
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
static struct felix_stream_filter *sfi;
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
struct ocelot_psfp_list *psfp;
psfp = &ocelot->psfp;
stream = vsc9959_stream_table_get(&psfp->stream_list, f->cookie);
if (!stream)
return -ENOMEM;
net: dsa: felix: add stream gate settings for psfp This patch adds stream gate settings for PSFP. Use SGI table to store stream gate entries. Disable the gate entry when it is not used by any stream. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:01 +08:00
sfi = vsc9959_psfp_sfi_table_get(&psfp->sfi_list, stream->sfid);
if (!sfi)
return -ENOMEM;
if (sfi->sg_valid)
vsc9959_psfp_sgi_table_del(ocelot, sfi->sgid);
net: dsa: felix: use vcap policer to set flow meter for psfp This patch add police action to set flow meter table which is defined in IEEE802.1Qci. Flow metering is two rates two buckets and three color marker to policing the frames, we only enable one rate one bucket in this patch. Flow metering shares a same policer pool with VCAP policers, so the PSFP policer calls ocelot_vcap_policer_add() and ocelot_vcap_policer_del() to set flow meter police. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:03 +08:00
if (sfi->fm_valid)
ocelot_vcap_policer_del(ocelot, sfi->fmid);
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
vsc9959_psfp_sfi_table_del(ocelot, stream->sfid);
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
memcpy(&tmp, stream, sizeof(tmp));
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
stream->sfid_valid = 0;
vsc9959_stream_table_del(ocelot, stream);
net: dsa: felix: restrict psfp rules on ingress port PSFP rules take effect on the streams from any port of VSC9959 switch. This patch use ingress port to limit the rule only active on this port. Each stream can only match two ingress source ports in VSC9959. Streams from lowest port gets the configuration of SFID pointed by MAC Table lookup and streams from highest port gets the configuration of (SFID+1) pointed by MAC Table lookup. This patch defines the PSFP rule on highest port as dummy rule, which means that it does not modify the MAC table. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:04 +08:00
stream_entry = vsc9959_stream_table_lookup(&psfp->stream_list, &tmp);
if (stream_entry) {
stream_entry->ports = BIT(stream_entry->port);
if (stream_entry->dummy) {
stream_entry->dummy = false;
vsc9959_mact_stream_set(ocelot, stream_entry, NULL);
}
vsc9959_psfp_sfidmask_set(ocelot, stream_entry->sfid,
stream_entry->ports);
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
return 0;
}
static int vsc9959_psfp_stats_get(struct ocelot *ocelot,
struct flow_cls_offload *f,
struct flow_stats *stats)
{
struct felix_stream_filter_counters counters;
struct ocelot_psfp_list *psfp;
struct felix_stream *stream;
psfp = &ocelot->psfp;
stream = vsc9959_stream_table_get(&psfp->stream_list, f->cookie);
if (!stream)
return -ENOMEM;
vsc9959_psfp_counters_get(ocelot, stream->sfid, &counters);
stats->pkts = counters.match;
stats->drops = counters.not_pass_gate + counters.not_pass_sdu +
counters.red;
return 0;
}
static void vsc9959_psfp_init(struct ocelot *ocelot)
{
struct ocelot_psfp_list *psfp = &ocelot->psfp;
INIT_LIST_HEAD(&psfp->stream_list);
INIT_LIST_HEAD(&psfp->sfi_list);
INIT_LIST_HEAD(&psfp->sgi_list);
}
net: dsa: felix: enable cut-through forwarding between ports by default The VSC9959 switch embedded within NXP LS1028A (and that version of Ocelot switches only) supports cut-through forwarding - meaning it can start the process of looking up the destination ports for a packet, and forward towards those ports, before the entire packet has been received (as opposed to the store-and-forward mode). The up side is having lower forwarding latency for large packets. The down side is that frames with FCS errors are forwarded instead of being dropped. However, erroneous frames do not result in incorrect updates of the FDB or incorrect policer updates, since these processes are deferred inside the switch to the end of frame. Since the switch starts the cut-through forwarding process after all packet headers (including IP, if any) have been processed, packets with large headers and small payload do not see the benefit of lower forwarding latency. There are two cases that need special attention. The first is when a packet is multicast (or flooded) to multiple destinations, one of which doesn't have cut-through forwarding enabled. The switch deals with this automatically by disabling cut-through forwarding for the frame towards all destination ports. The second is when a packet is forwarded from a port of lower link speed towards a port of higher link speed. This is not handled by the hardware and needs software intervention. Since we practically need to update the cut-through forwarding domain from paths that aren't serialized by the rtnl_mutex (phylink mac_link_down/mac_link_up ops), this means we need to serialize physical link events with user space updates of bonding/bridging domains. Enabling cut-through forwarding is done per {egress port, traffic class}. I don't see any reason why this would be a configurable option as long as it works without issues, and there doesn't appear to be any user space configuration tool to toggle this on/off, so this patch enables cut-through forwarding on all eligible ports and traffic classes. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20211125125808.2383984-2-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 14:58:08 +02:00
/* When using cut-through forwarding and the egress port runs at a higher data
* rate than the ingress port, the packet currently under transmission would
* suffer an underrun since it would be transmitted faster than it is received.
* The Felix switch implementation of cut-through forwarding does not check in
* hardware whether this condition is satisfied or not, so we must restrict the
* list of ports that have cut-through forwarding enabled on egress to only be
* the ports operating at the lowest link speed within their respective
* forwarding domain.
*/
static void vsc9959_cut_through_fwd(struct ocelot *ocelot)
{
struct felix *felix = ocelot_to_felix(ocelot);
struct dsa_switch *ds = felix->ds;
int port, other_port;
lockdep_assert_held(&ocelot->fwd_domain_lock);
for (port = 0; port < ocelot->num_phys_ports; port++) {
struct ocelot_port *ocelot_port = ocelot->ports[port];
int min_speed = ocelot_port->speed;
unsigned long mask = 0;
u32 tmp, val = 0;
/* Disable cut-through on ports that are down */
if (ocelot_port->speed <= 0)
goto set;
if (dsa_is_cpu_port(ds, port)) {
/* Ocelot switches forward from the NPI port towards
* any port, regardless of it being in the NPI port's
* forwarding domain or not.
*/
mask = dsa_user_ports(ds);
} else {
mask = ocelot_get_bridge_fwd_mask(ocelot, port);
mask &= ~BIT(port);
if (ocelot->npi >= 0)
mask |= BIT(ocelot->npi);
else
net: mscc: ocelot: switch from {,un}set to {,un}assign for tag_8021q CPU ports There is a desire for the felix driver to gain support for multiple tag_8021q CPU ports, but the current model prevents it. This is because ocelot_apply_bridge_fwd_mask() only takes into consideration whether a port is a tag_8021q CPU port, but not whose CPU port it is. We need a model where we can have a direct affinity between an ocelot port and a tag_8021q CPU port. This serves as the basis for multiple CPU ports. Declare a "dsa_8021q_cpu" backpointer in struct ocelot_port which encodes that affinity. Repurpose the "ocelot_set_dsa_8021q_cpu" API to "ocelot_assign_dsa_8021q_cpu" to express the change of paradigm. Note that this change makes the first practical use of the new ocelot_port->index field in ocelot_port_unassign_dsa_8021q_cpu(), where we need to remove the old tag_8021q CPU port from the reserved VLAN range. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-05-22 00:37:42 +03:00
mask |= ocelot_port_assigned_dsa_8021q_cpu_mask(ocelot,
port);
net: dsa: felix: enable cut-through forwarding between ports by default The VSC9959 switch embedded within NXP LS1028A (and that version of Ocelot switches only) supports cut-through forwarding - meaning it can start the process of looking up the destination ports for a packet, and forward towards those ports, before the entire packet has been received (as opposed to the store-and-forward mode). The up side is having lower forwarding latency for large packets. The down side is that frames with FCS errors are forwarded instead of being dropped. However, erroneous frames do not result in incorrect updates of the FDB or incorrect policer updates, since these processes are deferred inside the switch to the end of frame. Since the switch starts the cut-through forwarding process after all packet headers (including IP, if any) have been processed, packets with large headers and small payload do not see the benefit of lower forwarding latency. There are two cases that need special attention. The first is when a packet is multicast (or flooded) to multiple destinations, one of which doesn't have cut-through forwarding enabled. The switch deals with this automatically by disabling cut-through forwarding for the frame towards all destination ports. The second is when a packet is forwarded from a port of lower link speed towards a port of higher link speed. This is not handled by the hardware and needs software intervention. Since we practically need to update the cut-through forwarding domain from paths that aren't serialized by the rtnl_mutex (phylink mac_link_down/mac_link_up ops), this means we need to serialize physical link events with user space updates of bonding/bridging domains. Enabling cut-through forwarding is done per {egress port, traffic class}. I don't see any reason why this would be a configurable option as long as it works without issues, and there doesn't appear to be any user space configuration tool to toggle this on/off, so this patch enables cut-through forwarding on all eligible ports and traffic classes. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20211125125808.2383984-2-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 14:58:08 +02:00
}
/* Calculate the minimum link speed, among the ports that are
* up, of this source port's forwarding domain.
*/
for_each_set_bit(other_port, &mask, ocelot->num_phys_ports) {
struct ocelot_port *other_ocelot_port;
other_ocelot_port = ocelot->ports[other_port];
if (other_ocelot_port->speed <= 0)
continue;
if (min_speed > other_ocelot_port->speed)
min_speed = other_ocelot_port->speed;
}
/* Enable cut-through forwarding for all traffic classes. */
if (ocelot_port->speed == min_speed)
val = GENMASK(7, 0);
set:
tmp = ocelot_read_rix(ocelot, ANA_CUT_THRU_CFG, port);
if (tmp == val)
continue;
dev_dbg(ocelot->dev,
"port %d fwd mask 0x%lx speed %d min_speed %d, %s cut-through forwarding\n",
port, mask, ocelot_port->speed, min_speed,
val ? "enabling" : "disabling");
ocelot_write_rix(ocelot, val, ANA_CUT_THRU_CFG, port);
}
}
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
static const struct ocelot_ops vsc9959_ops = {
.reset = vsc9959_reset,
.wm_enc = vsc9959_wm_enc,
.wm_dec = vsc9959_wm_dec,
.wm_stat = vsc9959_wm_stat,
.port_to_netdev = felix_port_to_netdev,
.netdev_to_port = felix_netdev_to_port,
.psfp_init = vsc9959_psfp_init,
.psfp_filter_add = vsc9959_psfp_filter_add,
.psfp_filter_del = vsc9959_psfp_filter_del,
.psfp_stats_get = vsc9959_psfp_stats_get,
net: dsa: felix: enable cut-through forwarding between ports by default The VSC9959 switch embedded within NXP LS1028A (and that version of Ocelot switches only) supports cut-through forwarding - meaning it can start the process of looking up the destination ports for a packet, and forward towards those ports, before the entire packet has been received (as opposed to the store-and-forward mode). The up side is having lower forwarding latency for large packets. The down side is that frames with FCS errors are forwarded instead of being dropped. However, erroneous frames do not result in incorrect updates of the FDB or incorrect policer updates, since these processes are deferred inside the switch to the end of frame. Since the switch starts the cut-through forwarding process after all packet headers (including IP, if any) have been processed, packets with large headers and small payload do not see the benefit of lower forwarding latency. There are two cases that need special attention. The first is when a packet is multicast (or flooded) to multiple destinations, one of which doesn't have cut-through forwarding enabled. The switch deals with this automatically by disabling cut-through forwarding for the frame towards all destination ports. The second is when a packet is forwarded from a port of lower link speed towards a port of higher link speed. This is not handled by the hardware and needs software intervention. Since we practically need to update the cut-through forwarding domain from paths that aren't serialized by the rtnl_mutex (phylink mac_link_down/mac_link_up ops), this means we need to serialize physical link events with user space updates of bonding/bridging domains. Enabling cut-through forwarding is done per {egress port, traffic class}. I don't see any reason why this would be a configurable option as long as it works without issues, and there doesn't appear to be any user space configuration tool to toggle this on/off, so this patch enables cut-through forwarding on all eligible ports and traffic classes. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20211125125808.2383984-2-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 14:58:08 +02:00
.cut_through_fwd = vsc9959_cut_through_fwd,
net: dsa: felix: update base time of time-aware shaper when adjusting PTP time When adjusting the PTP clock, the base time of the TAS configuration will become unreliable. We need reset the TAS configuration by using a new base time. For example, if the driver gets a base time 0 of Qbv configuration from user, and current time is 20000. The driver will set the TAS base time to be 20000. After the PTP clock adjustment, the current time becomes 10000. If the TAS base time is still 20000, it will be a future time, and TAS entry list will stop running. Another example, if the current time becomes to be 10000000 after PTP clock adjust, a large time offset can cause the hardware to hang. This patch introduces a tas_clock_adjust() function to reset the TAS module by using a new base time after the PTP clock adjustment. This can avoid issues above. Due to PTP clock adjustment can occur at any time, it may conflict with the TAS configuration. We introduce a new TAS lock to serialize the access to the TAS registers. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-06-17 11:24:23 +08:00
.tas_clock_adjust = vsc9959_tas_clock_adjust,
net: dsa: felix: support psfp filter on vsc9959 VSC9959 supports Per-Stream Filtering and Policing(PSFP) that complies with the IEEE 802.1Qci standard. The stream is identified by Null stream identification(DMAC and VLAN ID) defined in IEEE802.1CB. For PSFP, four tables need to be set up: stream table, stream filter table, stream gate table, and flow meter table. Identify the stream by parsing the tc flower keys and add it to the stream table. The stream filter table is automatically maintained, and its index is determined by SGID(flow gate index) and FMID(flow meter index). Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:00 +08:00
};
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
static const struct felix_info felix_info_vsc9959 = {
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
.target_io_res = vsc9959_target_io_res,
.port_io_res = vsc9959_port_io_res,
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
.imdio_res = &vsc9959_imdio_res,
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
.regfields = vsc9959_regfields,
.map = vsc9959_regmap,
.ops = &vsc9959_ops,
.stats_layout = vsc9959_stats_layout,
net: dsa: felix: Wire up the ocelot cls_flower methods Export the cls_flower methods from the ocelot driver and hook them up to the DSA passthrough layer. Tables for the VCAP IS2 parameters, as well as half key packing (field offsets and lengths) need to be defined for the VSC9959 core, as they are different from Ocelot, mainly due to the different port count. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-02-29 16:31:14 +02:00
.vcap = vsc9959_vcap_props,
net: mscc: ocelot: use index to set vcap policer Policer was previously automatically assigned from the highest index to the lowest index from policer pool. But police action of tc flower now uses index to set an police entry. This patch uses the police index to set vcap policers, so that one policer can be shared by multiple rules. Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-18 18:12:02 +08:00
.vcap_pol_base = VSC9959_VCAP_POLICER_BASE,
.vcap_pol_max = VSC9959_VCAP_POLICER_MAX,
.vcap_pol_base2 = 0,
.vcap_pol_max2 = 0,
net: dsa: ocelot: the MAC table on Felix is twice as large When running 'bridge fdb dump' on Felix, sometimes learnt and static MAC addresses would appear, sometimes they wouldn't. Turns out, the MAC table has 4096 entries on VSC7514 (Ocelot) and 8192 entries on VSC9959 (Felix), so the existing code from the Ocelot common library only dumped half of Felix's MAC table. They are both organized as a 4-way set-associative TCAM, so we just need a single variable indicating the correct number of rows. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-04 01:20:26 +03:00
.num_mact_rows = 2048,
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
.num_ports = VSC9959_NUM_PORTS,
net: mscc: ocelot: export NUM_TC constant from felix to common switch lib We should be moving anything that isn't DSA-specific or SoC-specific out of the felix DSA driver, and into the common mscc_ocelot switch library. The number of traffic classes is one of the aspects that is common between all ocelot switches, so it belongs in the library. This patch also makes seville use 8 TX queues, and therefore enables prioritization via the QOS_CLASS field in the NPI injection header. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-15 04:11:16 +02:00
.num_tx_queues = OCELOT_NUM_TC,
net: dsa: felix: setup MMIO filtering rules for PTP when using tag_8021q Since the tag_8021q tagger is software-defined, it has no means by itself for retrieving hardware timestamps of PTP event messages. Because we do want to support PTP on ocelot even with tag_8021q, we need to use the CPU port module for that. The RX timestamp is present in the Extraction Frame Header. And because we can't use NPI mode which redirects the CPU queues to an "external CPU" (meaning the ARM CPU running Linux), then we need to poll the CPU port module through the MMIO registers to retrieve TX and RX timestamps. Sadly, on NXP LS1028A, the Felix switch was integrated into the SoC without wiring the extraction IRQ line to the ARM GIC. So, if we want to be notified of any PTP packets received on the CPU port module, we have a problem. There is a possible workaround, which is to use the Ethernet CPU port as a notification channel that packets are available on the CPU port module as well. When a PTP packet is received by the DSA tagger (without timestamp, of course), we go to the CPU extraction queues, poll for it there, then we drop the original Ethernet packet and masquerade the packet retrieved over MMIO (plus the timestamp) as the original when we inject it up the stack. Create a quirk in struct felix is selected by the Felix driver (but not by Seville, since that doesn't support PTP at all). We want to do this such that the workaround is minimally invasive for future switches that don't require this workaround. The only traffic for which we need timestamps is PTP traffic, so add a redirection rule to the CPU port module for this. Currently we only have the need for PTP over L2, so redirection rules for UDP ports 319 and 320 are TBD for now. Note that for the workaround of matching of PTP-over-Ethernet-port with PTP-over-MMIO queues to work properly, both channels need to be absolutely lossless. There are two parts to achieving that: - We keep flow control enabled on the tag_8021q CPU port - We put the DSA master interface in promiscuous mode, so it will never drop a PTP frame (for the profiles we are interested in, these are sent to the multicast MAC addresses of 01-80-c2-00-00-0e and 01-1b-19-00-00-00). Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-14 00:38:00 +02:00
.quirk_no_xtr_irq = true,
net: dsa: felix: move the PTP clock structure to felix_vsc9959.c Not only does Sevile not have a PTP clock, but with separate modules, this structure cannot even live in felix.c, due to the .owner = THIS_MODULE assignment causing this link time error: drivers/net/dsa/ocelot/felix.o:(.data+0x0): undefined reference to `__this_module' Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:52 +03:00
.ptp_caps = &vsc9959_ptp_caps,
net: dsa: felix: Add PCS operations for PHYLINK Layerscape SoCs traditionally expose the SerDes configuration/status for Ethernet protocols (PCS for SGMII/USXGMII/10GBase-R etc etc) in a register format that is compatible with clause 22 or clause 45 (depending on SerDes protocol). Each MAC has its own internal MDIO bus on which there is one or more of these PCS's, responding to commands at a configurable PHY address. The per-port internal MDIO bus (which is just for PCSs) is totally separate and has nothing to do with the dedicated external MDIO controller (which is just for PHYs), but the register map for the MDIO controller is the same. The VSC9959 (Felix) switch instantiated in the LS1028A is integrated in hardware with the ENETC PCS of its DSA master, and reuses its MDIO controller driver, so Felix has been made to depend on it in Kconfig. +------------------------------------------------------------------------+ | +--------+ GMII (typically disabled via RCW) | | ENETC PCI | ENETC |--------------------------+ | | Root Complex | port 3 |-----------------------+ | | | Integrated +--------+ | | | | Endpoint | | | | +--------+ 2.5G GMII | | | | | ENETC |--------------+ | | | | | port 2 |-----------+ | | | | | +--------+ | | | | | | +--------+ +--------+ | | | Felix | | Felix | | | | port 4 | | port 5 | | | +--------+ +--------+ | | | | +--------+ +--------+ +--------+ +--------+ +--------+ +--------+ | | | ENETC | | ENETC | | Felix | | Felix | | Felix | | Felix | | | | port 0 | | port 1 | | port 0 | | port 1 | | port 2 | | port 3 | | +------------------------------------------------------------------------+ | |||| SerDes | |||| |||| |||| |||| | | +--------+block | +--------------------------------------------+ | | | ENETC | | | ENETC port 2 internal MDIO bus | | | | port 0 | | | PCS PCS PCS PCS | | | | PCS | | | 0 1 2 3 | | +-----------------|------------------------------------------------------+ v v v v v v SGMII/ RGMII QSGMII/QSXGMII/4xSGMII/4x1000Base-X/4x2500Base-X USXGMII/ (bypasses 1000Base-X/ SerDes) 2500Base-X In the LS1028A SoC described above, the VSC9959 Felix switch is PF5 of the ENETC root complex, and has 2 BARs: - BAR 4: the switch's effective registers - BAR 0: the MDIO controller register map lended from ENETC port 2 (PF2), for accessing its associated PCS's. This explanation is necessary because the patch does some renaming "pci_bar" -> "switch_pci_bar" for clarity, which would otherwise appear a bit obtuse. The fact that the internal MDIO bus is "borrowed" is relevant because the register map is found in PF5 (the switch) but it triggers an access fault if PF2 (the ENETC DSA master) is not enabled. This is not treated in any way (and I don't think it can be treated). All of this is so SoC-specific, that it was contained as much as possible in the platform-integration file felix_vsc9959.c. We need to parse and pre-validate the device tree because of 2 reasons: - The PHY mode (SerDes protocol) cannot change at runtime due to SoC design. - There is a circular dependency in that we need to know what clause the PCS speaks in order to find it on the internal MDIO bus. But the clause of the PCS depends on what phy-mode it is configured for. The goal of this patch is to make steps towards removing the bootloader dependency for SGMII PCS pre-configuration, as well as to add support for monitoring the in-band SGMII AN between the PCS and the system-side link partner (PHY or other MAC). In practice the bootloader dependency is not completely removed. U-Boot pre-programs the PHY address at which each PCS can be found on the internal MDIO bus (MDEV_PORT). This is needed because the PCS of each port has the same out-of-reset PHY address of zero. The SerDes register for changing MDEV_PORT is pretty deep in the SoC (outside the addresses of the ENETC PCI BARs) and therefore inaccessible to us from here. Felix VSC9959 and Ocelot VSC7514 are integrated very differently in their respective SoCs, and for that reason Felix does not use the Ocelot core library for PHYLINK. On one hand we don't want to impose the fixed phy-mode limitation to Ocelot, and on the other hand Felix doesn't need to force the MAC link speed the way Ocelot does, since the MAC is connected to the PCS through a fixed GMII, and the PCS is the one who does the rate adaptation at lower link speeds, which the MAC does not even need to know about. In fact changing the GMII speed for Felix irrecoverably breaks transmission through that port until a reset. The pair with ENETC port 3 and Felix port 5 is optional and doesn't support tagging. When we enable it, swp5 is a regular slave port, albeit an internal one. The trouble is that it doesn't work, and that is because the DSA PHYLIB adaptation layer doesn't treat fixed-link slave ports. So that is yet another reason for wanting to convert Felix to the native PHYLINK API. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-01-06 03:34:17 +02:00
.mdio_bus_alloc = vsc9959_mdio_bus_alloc,
.mdio_bus_free = vsc9959_mdio_bus_free,
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
.phylink_validate = vsc9959_phylink_validate,
net: dsa: felix: remove prevalidate_phy_mode interface All users of the felix driver were creating their own prevalidate_phy_mode function. The same logic can be performed in a more general way by using a simple array of bit fields. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-26 14:36:50 -08:00
.port_modes = vsc9959_port_modes,
net: dsa: felix: replace tabs with spaces Over the time, some patches have introduced structures aligned with spaces, near structures aligned with tabs. Fix the inconsistencies. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:49 +03:00
.port_setup_tc = vsc9959_port_setup_tc,
.port_sched_speed_set = vsc9959_sched_speed_set,
net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port Currently, sending a packet into a time gate too small for it (or always closed) causes the queue system to hold the frame forever. Even worse, this frame isn't subject to aging either, because for that to happen, it needs to be scheduled for transmission in the first place. But the frame will consume buffer memory and frame references while it is forever held in the queue system. Before commit a4ae997adcbd ("net: mscc: ocelot: initialize watermarks to sane defaults"), this behavior was somewhat subtle, as the switch had a more intricately tuned default watermark configuration out of reset, which did not allow any single port and tc to consume the entire switch buffer space. Nonetheless, the held frames are still there, and they reduce the total backplane capacity of the switch. However, after the aforementioned commit, the behavior can be very clearly seen, since we deliberately allow each {port, tc} to consume the entire shared buffer of the switch minus the reservations (and we disable all reservations by default). That is to say, we allow a permanently closed tc-taprio gate to hang the entire switch. A careful inspection of the documentation shows that the QSYS:Q_MAX_SDU per-port-tc registers serve 2 purposes: one is for guard band calculation (when zero, this falls back to QSYS:PORT_MAX_SDU), and the other is to enable oversized frame dropping (when non-zero). Currently the QSYS:Q_MAX_SDU registers are all zero, so oversized frame dropping is disabled. The goal of the change is to enable it seamlessly. For that, we need to hook into the MTU change, tc-taprio change, and port link speed change procedures, since we depend on these variables. Frames are not dropped on egress due to a queue system oversize condition, instead that egress port is simply excluded from the mask of valid destination ports for the packet. If there are no destination ports at all, the ingress counter that increments is the generic "drop_tail" in ethtool -S. The issue exists in various forms since the tc-taprio offload was introduced. Fixes: de143c0e274b ("net: dsa: felix: Configure Time-Aware Scheduler via taprio offload") Reported-by: Richie Pearn <richard.pearn@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-06-28 17:52:37 +03:00
.tas_guard_bands_update = vsc9959_tas_guard_bands_update,
net: dsa: ocelot: felix: add interface for custom regmaps Add an interface so that non-mmio regmaps can be used Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 09:00:29 -08:00
.init_regmap = ocelot_regmap_init,
net: dsa: ocelot: add driver for Felix switch family This supports an Ethernet switching core from Vitesse / Microsemi / Microchip (VSC9959) which is part of the Ocelot family (a brand name), and whose code name is Felix. The switch can be (and is) integrated on different SoCs as a PCIe endpoint device. The functionality is provided by the core of the Ocelot switch driver (drivers/net/ethernet/mscc). In this regard, the current driver is an instance of Microsemi's Ocelot core driver, with a DSA front-end. It inherits its name from VSC9959's code name, to distinguish itself from the switchdev ocelot driver. The patch adds the logic for probing a PCI device and defines the register map for the VSC9959 switch core, since it has some differences in register addresses and bitfield mappings compared to the other Ocelot switches (VSC7511, VSC7512, VSC7513, VSC7514). The Felix driver declares the register map as part of the "instance table". Currently the VSC9959 inside NXP LS1028A is the only instance, but presumably it can support other switches in the Ocelot family, when used in DSA mode (Linux running on the external CPU, and not on the embedded MIPS). In a few cases, some h/w operations have to be done differently on VSC9959 due to missing bitfields. This is the case for the switch core reset and init. Because for this operation Ocelot uses some bits that are not present on Felix, the latter has to use a register from the global registers block (GCB) instead. Although it is a PCI driver, it relies on DT bindings for compatibility with DSA (CPU port link, PHY library). It does not have any custom device tree bindings, since we would like to minimize its dependency on device tree though. Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-14 17:03:30 +02:00
};
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
static irqreturn_t felix_irq_handler(int irq, void *data)
{
struct ocelot *ocelot = (struct ocelot *)data;
/* The INTB interrupt is used for both PTP TX timestamp interrupt
* and preemption status change interrupt on each port.
*
* - Get txtstamp if have
* - TODO: handle preemption. Without handling it, driver may get
* interrupt storm.
*/
ocelot_get_txtstamp(ocelot);
return IRQ_HANDLED;
}
static int felix_pci_probe(struct pci_dev *pdev,
const struct pci_device_id *id)
{
struct dsa_switch *ds;
struct ocelot *ocelot;
struct felix *felix;
int err;
if (pdev->dev.of_node && !of_device_is_available(pdev->dev.of_node)) {
dev_info(&pdev->dev, "device is disabled, skipping\n");
return -ENODEV;
}
err = pci_enable_device(pdev);
if (err) {
dev_err(&pdev->dev, "device enable failed\n");
goto err_pci_enable;
}
felix = kzalloc(sizeof(struct felix), GFP_KERNEL);
if (!felix) {
err = -ENOMEM;
dev_err(&pdev->dev, "Failed to allocate driver memory\n");
goto err_alloc_felix;
}
pci_set_drvdata(pdev, felix);
ocelot = &felix->ocelot;
ocelot->dev = &pdev->dev;
net: mscc: ocelot: export NUM_TC constant from felix to common switch lib We should be moving anything that isn't DSA-specific or SoC-specific out of the felix DSA driver, and into the common mscc_ocelot switch library. The number of traffic classes is one of the aspects that is common between all ocelot switches, so it belongs in the library. This patch also makes seville use 8 TX queues, and therefore enables prioritization via the QOS_CLASS field in the NPI injection header. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-15 04:11:16 +02:00
ocelot->num_flooding_pgids = OCELOT_NUM_TC;
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
felix->info = &felix_info_vsc9959;
net: dsa: ocelot: remove unnecessary pci_bar variables The pci_bar variables for the switch and imdio don't make sense for the generic felix driver. Moving them to felix_vsc9959 to limit scope and simplify the felix_info struct. Signed-off-by: Colin Foster <colin.foster@in-advantage.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 09:00:27 -08:00
felix->switch_base = pci_resource_start(pdev, VSC9959_SWITCH_PCI_BAR);
felix->imdio_base = pci_resource_start(pdev, VSC9959_IMDIO_PCI_BAR);
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
pci_set_master(pdev);
err = devm_request_threaded_irq(&pdev->dev, pdev->irq, NULL,
&felix_irq_handler, IRQF_ONESHOT,
"felix-intb", ocelot);
if (err) {
dev_err(&pdev->dev, "Failed to request irq\n");
goto err_alloc_irq;
}
ocelot->ptp = 1;
ds = kzalloc(sizeof(struct dsa_switch), GFP_KERNEL);
if (!ds) {
err = -ENOMEM;
dev_err(&pdev->dev, "Failed to allocate DSA switch\n");
goto err_alloc_ds;
}
ds->dev = &pdev->dev;
ds->num_ports = felix->info->num_ports;
ds->num_tx_queues = felix->info->num_tx_queues;
ds->ops = &felix_switch_ops;
ds->priv = ocelot;
felix->ds = ds;
net: dsa: felix: convert to the new .change_tag_protocol DSA API In expectation of the new tag_ocelot_8021q tagger implementation, we need to be able to do runtime switchover between one tagger and another. So we must structure the existing code for the current NPI-based tagger in a certain way. We move the felix_npi_port_init function in expectation of the future driver configuration necessary for tag_ocelot_8021q: we would like to not have the NPI-related bits interspersed with the tag_8021q bits. The conversion from this: ocelot_write_rix(ocelot, ANA_PGID_PGID_PGID(GENMASK(ocelot->num_phys_ports, 0)), ANA_PGID_PGID, PGID_UC); to this: cpu_flood = ANA_PGID_PGID_PGID(BIT(ocelot->num_phys_ports)); ocelot_rmw_rix(ocelot, cpu_flood, cpu_flood, ANA_PGID_PGID, PGID_UC); is perhaps non-trivial, but is nonetheless non-functional. The PGID_UC (replicator for unknown unicast) is already configured out of hardware reset to flood to all ports except ocelot->num_phys_ports (the CPU port module). All we change is that we use a read-modify-write to only add the CPU port module to the unknown unicast replicator, as opposed to doing a full write to the register. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-29 03:00:07 +02:00
felix->tag_proto = DSA_TAG_PROTO_OCELOT;
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
err = dsa_register_switch(ds);
if (err) {
net: dsa: felix: suppress -EPROBE_DEFER errors The DSA master might not have been probed yet in which case the probe of the felix switch fails with -EPROBE_DEFER: [ 4.435305] mscc_felix 0000:00:00.5: Failed to register DSA switch: -517 It is not an error. Use dev_err_probe() to demote this particular error to a debug message. Fixes: 56051948773e ("net: dsa: ocelot: add driver for Felix switch family") Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Link: https://lore.kernel.org/r/20220408101521.281886-1-michael@walle.cc Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-08 12:15:21 +02:00
dev_err_probe(&pdev->dev, err, "Failed to register DSA switch\n");
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
goto err_register_ds;
}
return 0;
err_register_ds:
kfree(ds);
err_alloc_ds:
err_alloc_irq:
kfree(felix);
net: dsa: felix: the switch does not support DMA The code that sets the DMA mask to 64 bits is bogus, it is taken from the enetc driver together with the rest of the PCI probing boilerplate. Since this patch is touching the error path to delete err_dma, let's also change the err_alloc_felix label which was incorrect. The kzalloc failure does not need a kfree, but it doesn't hurt either, since kfree works with NULL pointers. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Link: https://lore.kernel.org/r/20210109203415.2120142-1-olteanv@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 22:34:15 +02:00
err_alloc_felix:
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
pci_disable_device(pdev);
err_pci_enable:
return err;
}
static void felix_pci_remove(struct pci_dev *pdev)
{
net: dsa: be compatible with masters which unregister on shutdown Lino reports that on his system with bcmgenet as DSA master and KSZ9897 as a switch, rebooting or shutting down never works properly. What does the bcmgenet driver have special to trigger this, that other DSA masters do not? It has an implementation of ->shutdown which simply calls its ->remove implementation. Otherwise said, it unregisters its network interface on shutdown. This message can be seen in a loop, and it hangs the reboot process there: unregister_netdevice: waiting for eth0 to become free. Usage count = 3 So why 3? A usage count of 1 is normal for a registered network interface, and any virtual interface which links itself as an upper of that will increment it via dev_hold. In the case of DSA, this is the call path: dsa_slave_create -> netdev_upper_dev_link -> __netdev_upper_dev_link -> __netdev_adjacent_dev_insert -> dev_hold So a DSA switch with 3 interfaces will result in a usage count elevated by two, and netdev_wait_allrefs will wait until they have gone away. Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and delete themselves, but DSA cannot just vanish and go poof, at most it can unbind itself from the switch devices, but that must happen strictly earlier compared to when the DSA master unregisters its net_device, so reacting on the NETDEV_UNREGISTER event is way too late. It seems that it is a pretty established pattern to have a driver's ->shutdown hook redirect to its ->remove hook, so the same code is executed regardless of whether the driver is unbound from the device, or the system is just shutting down. As Florian puts it, it is quite a big hammer for bcmgenet to unregister its net_device during shutdown, but having a common code path with the driver unbind helps ensure it is well tested. So DSA, for better or for worse, has to live with that and engage in an arms race of implementing the ->shutdown hook too, from all individual drivers, and do something sane when paired with masters that unregister their net_device there. The only sane thing to do, of course, is to unlink from the master. However, complications arise really quickly. The pattern of redirecting ->shutdown to ->remove is not unique to bcmgenet or even to net_device drivers. In fact, SPI controllers do it too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers and MDIO controllers do it too (this is something I have not researched too deeply, but even if this is not the case today, it is certainly plausible to happen in the future, and must be taken into consideration). Since DSA switches might be SPI devices, I2C devices, MDIO devices, the insane implication is that for the exact same DSA switch device, we might have both ->shutdown and ->remove getting called. So we need to do something with that insane environment. The pattern I've come up with is "if this, then not that", so if either ->shutdown or ->remove gets called, we set the device's drvdata to NULL, and in the other hook, we check whether the drvdata is NULL and just do nothing. This is probably not necessary for platform devices, just for devices on buses, but I would really insist for consistency among drivers, because when code is copy-pasted, it is not always copy-pasted from the best sources. So depending on whether the DSA switch's ->remove or ->shutdown will get called first, we cannot really guarantee even for the same driver if rebooting will result in the same code path on all platforms. But nonetheless, we need to do something minimally reasonable on ->shutdown too to fix the bug. Of course, the ->remove will do more (a full teardown of the tree, with all data structures freed, and this is why the bug was not caught for so long). The new ->shutdown method is kept separate from dsa_unregister_switch not because we couldn't have unregistered the switch, but simply in the interest of doing something quick and to the point. The big question is: does the DSA switch's ->shutdown get called earlier than the DSA master's ->shutdown? If not, there is still a risk that we might still trigger the WARN_ON in unregister_netdevice that says we are attempting to unregister a net_device which has uppers. That's no good. Although the reference to the master net_device won't physically go away even if DSA's ->shutdown comes afterwards, remember we have a dev_hold on it. The answer to that question lies in this comment above device_link_add: * A side effect of the link creation is re-ordering of dpm_list and the * devices_kset list by moving the consumer device and all devices depending * on it to the ends of these lists (that does not happen to devices that have * not been registered when this function is called). so the fact that DSA uses device_link_add towards its master is not exactly for nothing. device_shutdown() walks devices_kset from the back, so this is our guarantee that DSA's shutdown happens before the master's shutdown. Fixes: 2f1e8ea726e9 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings") Link: https://lore.kernel.org/netdev/20210909095324.12978-1-LinoSanfilippo@gmx.de/ Reported-by: Lino Sanfilippo <LinoSanfilippo@gmx.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-17 16:34:33 +03:00
struct felix *felix = pci_get_drvdata(pdev);
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
net: dsa: be compatible with masters which unregister on shutdown Lino reports that on his system with bcmgenet as DSA master and KSZ9897 as a switch, rebooting or shutting down never works properly. What does the bcmgenet driver have special to trigger this, that other DSA masters do not? It has an implementation of ->shutdown which simply calls its ->remove implementation. Otherwise said, it unregisters its network interface on shutdown. This message can be seen in a loop, and it hangs the reboot process there: unregister_netdevice: waiting for eth0 to become free. Usage count = 3 So why 3? A usage count of 1 is normal for a registered network interface, and any virtual interface which links itself as an upper of that will increment it via dev_hold. In the case of DSA, this is the call path: dsa_slave_create -> netdev_upper_dev_link -> __netdev_upper_dev_link -> __netdev_adjacent_dev_insert -> dev_hold So a DSA switch with 3 interfaces will result in a usage count elevated by two, and netdev_wait_allrefs will wait until they have gone away. Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and delete themselves, but DSA cannot just vanish and go poof, at most it can unbind itself from the switch devices, but that must happen strictly earlier compared to when the DSA master unregisters its net_device, so reacting on the NETDEV_UNREGISTER event is way too late. It seems that it is a pretty established pattern to have a driver's ->shutdown hook redirect to its ->remove hook, so the same code is executed regardless of whether the driver is unbound from the device, or the system is just shutting down. As Florian puts it, it is quite a big hammer for bcmgenet to unregister its net_device during shutdown, but having a common code path with the driver unbind helps ensure it is well tested. So DSA, for better or for worse, has to live with that and engage in an arms race of implementing the ->shutdown hook too, from all individual drivers, and do something sane when paired with masters that unregister their net_device there. The only sane thing to do, of course, is to unlink from the master. However, complications arise really quickly. The pattern of redirecting ->shutdown to ->remove is not unique to bcmgenet or even to net_device drivers. In fact, SPI controllers do it too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers and MDIO controllers do it too (this is something I have not researched too deeply, but even if this is not the case today, it is certainly plausible to happen in the future, and must be taken into consideration). Since DSA switches might be SPI devices, I2C devices, MDIO devices, the insane implication is that for the exact same DSA switch device, we might have both ->shutdown and ->remove getting called. So we need to do something with that insane environment. The pattern I've come up with is "if this, then not that", so if either ->shutdown or ->remove gets called, we set the device's drvdata to NULL, and in the other hook, we check whether the drvdata is NULL and just do nothing. This is probably not necessary for platform devices, just for devices on buses, but I would really insist for consistency among drivers, because when code is copy-pasted, it is not always copy-pasted from the best sources. So depending on whether the DSA switch's ->remove or ->shutdown will get called first, we cannot really guarantee even for the same driver if rebooting will result in the same code path on all platforms. But nonetheless, we need to do something minimally reasonable on ->shutdown too to fix the bug. Of course, the ->remove will do more (a full teardown of the tree, with all data structures freed, and this is why the bug was not caught for so long). The new ->shutdown method is kept separate from dsa_unregister_switch not because we couldn't have unregistered the switch, but simply in the interest of doing something quick and to the point. The big question is: does the DSA switch's ->shutdown get called earlier than the DSA master's ->shutdown? If not, there is still a risk that we might still trigger the WARN_ON in unregister_netdevice that says we are attempting to unregister a net_device which has uppers. That's no good. Although the reference to the master net_device won't physically go away even if DSA's ->shutdown comes afterwards, remember we have a dev_hold on it. The answer to that question lies in this comment above device_link_add: * A side effect of the link creation is re-ordering of dpm_list and the * devices_kset list by moving the consumer device and all devices depending * on it to the ends of these lists (that does not happen to devices that have * not been registered when this function is called). so the fact that DSA uses device_link_add towards its master is not exactly for nothing. device_shutdown() walks devices_kset from the back, so this is our guarantee that DSA's shutdown happens before the master's shutdown. Fixes: 2f1e8ea726e9 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings") Link: https://lore.kernel.org/netdev/20210909095324.12978-1-LinoSanfilippo@gmx.de/ Reported-by: Lino Sanfilippo <LinoSanfilippo@gmx.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-17 16:34:33 +03:00
if (!felix)
return;
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
dsa_unregister_switch(felix->ds);
kfree(felix->ds);
kfree(felix);
pci_disable_device(pdev);
net: dsa: be compatible with masters which unregister on shutdown Lino reports that on his system with bcmgenet as DSA master and KSZ9897 as a switch, rebooting or shutting down never works properly. What does the bcmgenet driver have special to trigger this, that other DSA masters do not? It has an implementation of ->shutdown which simply calls its ->remove implementation. Otherwise said, it unregisters its network interface on shutdown. This message can be seen in a loop, and it hangs the reboot process there: unregister_netdevice: waiting for eth0 to become free. Usage count = 3 So why 3? A usage count of 1 is normal for a registered network interface, and any virtual interface which links itself as an upper of that will increment it via dev_hold. In the case of DSA, this is the call path: dsa_slave_create -> netdev_upper_dev_link -> __netdev_upper_dev_link -> __netdev_adjacent_dev_insert -> dev_hold So a DSA switch with 3 interfaces will result in a usage count elevated by two, and netdev_wait_allrefs will wait until they have gone away. Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and delete themselves, but DSA cannot just vanish and go poof, at most it can unbind itself from the switch devices, but that must happen strictly earlier compared to when the DSA master unregisters its net_device, so reacting on the NETDEV_UNREGISTER event is way too late. It seems that it is a pretty established pattern to have a driver's ->shutdown hook redirect to its ->remove hook, so the same code is executed regardless of whether the driver is unbound from the device, or the system is just shutting down. As Florian puts it, it is quite a big hammer for bcmgenet to unregister its net_device during shutdown, but having a common code path with the driver unbind helps ensure it is well tested. So DSA, for better or for worse, has to live with that and engage in an arms race of implementing the ->shutdown hook too, from all individual drivers, and do something sane when paired with masters that unregister their net_device there. The only sane thing to do, of course, is to unlink from the master. However, complications arise really quickly. The pattern of redirecting ->shutdown to ->remove is not unique to bcmgenet or even to net_device drivers. In fact, SPI controllers do it too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers and MDIO controllers do it too (this is something I have not researched too deeply, but even if this is not the case today, it is certainly plausible to happen in the future, and must be taken into consideration). Since DSA switches might be SPI devices, I2C devices, MDIO devices, the insane implication is that for the exact same DSA switch device, we might have both ->shutdown and ->remove getting called. So we need to do something with that insane environment. The pattern I've come up with is "if this, then not that", so if either ->shutdown or ->remove gets called, we set the device's drvdata to NULL, and in the other hook, we check whether the drvdata is NULL and just do nothing. This is probably not necessary for platform devices, just for devices on buses, but I would really insist for consistency among drivers, because when code is copy-pasted, it is not always copy-pasted from the best sources. So depending on whether the DSA switch's ->remove or ->shutdown will get called first, we cannot really guarantee even for the same driver if rebooting will result in the same code path on all platforms. But nonetheless, we need to do something minimally reasonable on ->shutdown too to fix the bug. Of course, the ->remove will do more (a full teardown of the tree, with all data structures freed, and this is why the bug was not caught for so long). The new ->shutdown method is kept separate from dsa_unregister_switch not because we couldn't have unregistered the switch, but simply in the interest of doing something quick and to the point. The big question is: does the DSA switch's ->shutdown get called earlier than the DSA master's ->shutdown? If not, there is still a risk that we might still trigger the WARN_ON in unregister_netdevice that says we are attempting to unregister a net_device which has uppers. That's no good. Although the reference to the master net_device won't physically go away even if DSA's ->shutdown comes afterwards, remember we have a dev_hold on it. The answer to that question lies in this comment above device_link_add: * A side effect of the link creation is re-ordering of dpm_list and the * devices_kset list by moving the consumer device and all devices depending * on it to the ends of these lists (that does not happen to devices that have * not been registered when this function is called). so the fact that DSA uses device_link_add towards its master is not exactly for nothing. device_shutdown() walks devices_kset from the back, so this is our guarantee that DSA's shutdown happens before the master's shutdown. Fixes: 2f1e8ea726e9 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings") Link: https://lore.kernel.org/netdev/20210909095324.12978-1-LinoSanfilippo@gmx.de/ Reported-by: Lino Sanfilippo <LinoSanfilippo@gmx.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-17 16:34:33 +03:00
pci_set_drvdata(pdev, NULL);
}
static void felix_pci_shutdown(struct pci_dev *pdev)
{
struct felix *felix = pci_get_drvdata(pdev);
if (!felix)
return;
dsa_switch_shutdown(felix->ds);
pci_set_drvdata(pdev, NULL);
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
}
static struct pci_device_id felix_ids[] = {
{
/* NXP LS1028A */
PCI_DEVICE(PCI_VENDOR_ID_FREESCALE, 0xEEF0),
},
{ 0, }
};
MODULE_DEVICE_TABLE(pci, felix_ids);
net: dsa: seville: build as separate module Seville does not need to depend on PCI or on the ENETC MDIO controller. There will also be other compile-time differences in the future. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:53 +03:00
static struct pci_driver felix_vsc9959_pci_driver = {
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
.name = "mscc_felix",
.id_table = felix_ids,
.probe = felix_pci_probe,
.remove = felix_pci_remove,
net: dsa: be compatible with masters which unregister on shutdown Lino reports that on his system with bcmgenet as DSA master and KSZ9897 as a switch, rebooting or shutting down never works properly. What does the bcmgenet driver have special to trigger this, that other DSA masters do not? It has an implementation of ->shutdown which simply calls its ->remove implementation. Otherwise said, it unregisters its network interface on shutdown. This message can be seen in a loop, and it hangs the reboot process there: unregister_netdevice: waiting for eth0 to become free. Usage count = 3 So why 3? A usage count of 1 is normal for a registered network interface, and any virtual interface which links itself as an upper of that will increment it via dev_hold. In the case of DSA, this is the call path: dsa_slave_create -> netdev_upper_dev_link -> __netdev_upper_dev_link -> __netdev_adjacent_dev_insert -> dev_hold So a DSA switch with 3 interfaces will result in a usage count elevated by two, and netdev_wait_allrefs will wait until they have gone away. Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and delete themselves, but DSA cannot just vanish and go poof, at most it can unbind itself from the switch devices, but that must happen strictly earlier compared to when the DSA master unregisters its net_device, so reacting on the NETDEV_UNREGISTER event is way too late. It seems that it is a pretty established pattern to have a driver's ->shutdown hook redirect to its ->remove hook, so the same code is executed regardless of whether the driver is unbound from the device, or the system is just shutting down. As Florian puts it, it is quite a big hammer for bcmgenet to unregister its net_device during shutdown, but having a common code path with the driver unbind helps ensure it is well tested. So DSA, for better or for worse, has to live with that and engage in an arms race of implementing the ->shutdown hook too, from all individual drivers, and do something sane when paired with masters that unregister their net_device there. The only sane thing to do, of course, is to unlink from the master. However, complications arise really quickly. The pattern of redirecting ->shutdown to ->remove is not unique to bcmgenet or even to net_device drivers. In fact, SPI controllers do it too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers and MDIO controllers do it too (this is something I have not researched too deeply, but even if this is not the case today, it is certainly plausible to happen in the future, and must be taken into consideration). Since DSA switches might be SPI devices, I2C devices, MDIO devices, the insane implication is that for the exact same DSA switch device, we might have both ->shutdown and ->remove getting called. So we need to do something with that insane environment. The pattern I've come up with is "if this, then not that", so if either ->shutdown or ->remove gets called, we set the device's drvdata to NULL, and in the other hook, we check whether the drvdata is NULL and just do nothing. This is probably not necessary for platform devices, just for devices on buses, but I would really insist for consistency among drivers, because when code is copy-pasted, it is not always copy-pasted from the best sources. So depending on whether the DSA switch's ->remove or ->shutdown will get called first, we cannot really guarantee even for the same driver if rebooting will result in the same code path on all platforms. But nonetheless, we need to do something minimally reasonable on ->shutdown too to fix the bug. Of course, the ->remove will do more (a full teardown of the tree, with all data structures freed, and this is why the bug was not caught for so long). The new ->shutdown method is kept separate from dsa_unregister_switch not because we couldn't have unregistered the switch, but simply in the interest of doing something quick and to the point. The big question is: does the DSA switch's ->shutdown get called earlier than the DSA master's ->shutdown? If not, there is still a risk that we might still trigger the WARN_ON in unregister_netdevice that says we are attempting to unregister a net_device which has uppers. That's no good. Although the reference to the master net_device won't physically go away even if DSA's ->shutdown comes afterwards, remember we have a dev_hold on it. The answer to that question lies in this comment above device_link_add: * A side effect of the link creation is re-ordering of dpm_list and the * devices_kset list by moving the consumer device and all devices depending * on it to the ends of these lists (that does not happen to devices that have * not been registered when this function is called). so the fact that DSA uses device_link_add towards its master is not exactly for nothing. device_shutdown() walks devices_kset from the back, so this is our guarantee that DSA's shutdown happens before the master's shutdown. Fixes: 2f1e8ea726e9 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings") Link: https://lore.kernel.org/netdev/20210909095324.12978-1-LinoSanfilippo@gmx.de/ Reported-by: Lino Sanfilippo <LinoSanfilippo@gmx.de> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-17 16:34:33 +03:00
.shutdown = felix_pci_shutdown,
net: dsa: felix: move probing to felix_vsc9959.c Felix is not actually meant to be a DSA driver only for the switch inside NXP LS1028A, but an umbrella for all Vitesse / Microsemi / Microchip switches that are register-compatible with Ocelot and that are using in DSA mode (with an NPI Ethernet port). For the dsa_switch_ops exported by the felix driver to be generic enough to be used by other non-PCI switches, we need to move the PCI-specific probing to the low-level translation module felix_vsc9959.c. This way, other switches can have their own probing functions, as platform devices or otherwise. This patch also removes the "Felix instance table", which did not stand the test of time and is unnecessary at this point. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-13 19:57:09 +03:00
};
net: dsa: seville: build as separate module Seville does not need to depend on PCI or on the ENETC MDIO controller. There will also be other compile-time differences in the future. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-18 13:57:53 +03:00
module_pci_driver(felix_vsc9959_pci_driver);
MODULE_DESCRIPTION("Felix Switch driver");
MODULE_LICENSE("GPL v2");
Reference in a new issue Copy permalink