Merge some powerpc KVM patches from our topic branch.
In particular this brings in Nick's big series rewriting parts of the
guest entry/exit path in C.
Conflicts:
arch/powerpc/kernel/security.c
arch/powerpc/kvm/book3s_hv_rmhandlers.S
All KUAP helpers defined in asm/kup.h are single line functions
that should be inlined. But on book3s/32 build, we get many
instances of <prevent_write_to_user.constprop.0>.
Force inlining of those helpers.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/8479a862e165a57a855292d47e24c259a578f5a0.1622711627.git.christophe.leroy@csgroup.eu
PPC64 uses MMU features to enable/disable KUAP at boot time.
But feature fixups are applied way too early on PPC32.
Now that all KUAP related actions are in C following the
conversion of KUAP initial setup and context switch in C,
static branches can be used to enable/disable KUAP.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
[mpe: Export disable_kuap_key to fix build errors]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/cd79e8008455fba5395d099f9bb1305c039b931c.1622708530.git.christophe.leroy@csgroup.eu
PPC64 uses MMU features to enable/disable KUEP at boot time.
But feature fixups are applied way too early on PPC32.
Now that all KUEP related actions are in C following the
conversion of KUEP initial setup and context switch in C,
static branches can be used to enable/disable KUEP.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/7745a2c3a08ec46302920a3f48d1cb9b5469dbbb.1622708530.git.christophe.leroy@csgroup.eu
segment register has VSID on bits 8-31.
Bits 4-7 are reserved, there is no requirement to set them to 0.
VSIDs are calculated from VSID of SR0 by adding 0x111.
Even with highest possible VSID which would be 0xFFFFF0,
adding 16 times 0x111 results in 0x1001100.
So, the reserved bits are never overflowed, no need to clear
the reserved bits after each calculation.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/ddc1cfd2ec8f3b2395c6a4d7f2b0c1aa1b1e64fb.1622708530.git.christophe.leroy@csgroup.eu
switch_mmu_context() does things that can easily be done in C.
For updating user segments, we have update_user_segments().
As mentionned in commit b5efec00b6 ("powerpc/32s: Move KUEP
locking/unlocking in C"), update_user_segments() has the loop
unrolled which is a significant performance gain.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/05c0875ad8220c03452c3a334946e207c6ca04d6.1622708530.git.christophe.leroy@csgroup.eu
KUEP implements the update of user segment registers.
Move it into mmu-hash.h in order to use it from other places.
And inline kuep_lock() and kuep_unlock(). Inlining kuep_lock() is
important for system_call_exception(), otherwise system_call_exception()
has to save into stack the system call parameters that are used just
after, and doing that takes more instructions than kuep_lock() itself.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/24591ca480d14a62ef910e38a5273d551262c4a2.1622708530.git.christophe.leroy@csgroup.eu
'struct ppc_inst' is an internal representation of an instruction, but
in-memory instructions are and will remain a table of 'u32' forever.
Replace all 'struct ppc_inst *' used for locating an instruction in
memory by 'u32 *'. This removes a lot of undue casts to 'struct
ppc_inst *'.
It also helps locating ab-use of 'struct ppc_inst' dereference.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
[mpe: Fix ppc_inst_next(), use u32 instead of unsigned int]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/7062722b087228e42cbd896e39bfdf526d6a340a.1621516826.git.christophe.leroy@csgroup.eu
Avoid casting/dereferencing ppc_inst() as u64* , check each member
of the struct when relevant.
And remove the 0xff initialisation of the suffix for non
prefixed instruction. An instruction with 0xff as a suffix
might be invalid, but still is a prefixed instruction and
has to be considered as this.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/d8b155e930b7a9708ca110e8ff0ace6713a7af75.1621516826.git.christophe.leroy@csgroup.eu
Start using PPC_RAW_xx() macros where relevant.
PPC_INST_SYNC is used to both represent the 'sync' instruction and
the family of synchronisation instructions. Keep it for the later,
maybe we'll change the name in the future to avoid confusion.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/0945c155d6cb113431185fc1296ac127359fe29b.1621506159.git.christophe.leroy@csgroup.eu
Today we have __REG_Rx macros . They are mainly meant for
internal use by macros __PPC_RA() and friends macros which
allows uses like __PPC_RA(R12).
When used with PPC_RAW_xx() macros, it gives a result which is
not very readable.
Add shorter macros _Rx in order to improve readability when
used with PPC_RAW_xx() macros.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/ec34d92b7c2f810622261acfeeed4b0a0f4d01bd.1621506159.git.christophe.leroy@csgroup.eu
At the time being, we have PPC_RAW_PLXVP() and PPC_RAW_PSTXVP() which
provide a 64 bits value, and then it gets split by open coding to
format it into a 'struct ppc_inst' instruction.
Instead, define a PPC_RAW_xxx_P() and a PPC_RAW_xxx_S() to be used
as is.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/5d146b31b943e7ad674894421db4feef54804b9b.1621506159.git.christophe.leroy@csgroup.eu
Commit b0b3b2c78e ("powerpc: Switch to relative jump labels") switched
us to using relative jump labels. That involves changing the code,
target and key members in struct jump_entry to be relative to the
address of the jump_entry, rather than absolute addresses.
We have two static inlines that create a struct jump_entry,
arch_static_branch() and arch_static_branch_jump(), as well as an asm
macro ARCH_STATIC_BRANCH, which is used by the pseries-only hypervisor
tracing code.
Unfortunately we missed updating the key to be a relative reference in
ARCH_STATIC_BRANCH.
That causes a pseries kernel to have a handful of jump_entry structs
with bad key values. Instead of being a relative reference they instead
hold the full address of the key.
However the code doesn't expect that, it still adds the key value to the
address of the jump_entry (see jump_entry_key()) expecting to get a
pointer to a key somewhere in kernel data.
The table of jump_entry structs sits in rodata, which comes after the
kernel text. In a typical build this will be somewhere around 15MB. The
address of the key will be somewhere in data, typically around 20MB.
Adding the two values together gets us a pointer somewhere around 45MB.
We then call static_key_set_entries() with that bad pointer and modify
some members of the struct static_key we think we are pointing at.
A pseries kernel is typically ~30MB in size, so writing to ~45MB won't
corrupt the kernel itself. However if we're booting with an initrd,
depending on the size and exact location of the initrd, we can corrupt
the initrd. Depending on how exactly we corrupt the initrd it can either
cause the system to not boot, or just corrupt one of the files in the
initrd.
The fix is simply to make the key value relative to the jump_entry
struct in the ARCH_STATIC_BRANCH macro.
Fixes: b0b3b2c78e ("powerpc: Switch to relative jump labels")
Reported-by: Anastasia Kovaleva <a.kovaleva@yadro.com>
Reported-by: Roman Bolshakov <r.bolshakov@yadro.com>
Reported-by: Greg Kurz <groug@kaod.org>
Reported-by: Daniel Axtens <dja@axtens.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Tested-by: Daniel Axtens <dja@axtens.net>
Tested-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210614131440.312360-1-mpe@ellerman.id.au
Implement support for hash guests under hash host. This has to save and
restore the host SLB, and ensure that the MMU is off while switching
into the guest SLB.
POWER9 and later CPUs now always go via the P9 path. The "fast" guest
mode is now renamed to the P9 mode, which is consistent with its
functionality and the rest of the naming.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-32-npiggin@gmail.com
Implement hash guest support. Guest entry/exit has to restore and
save/clear the SLB, plus several other bits to accommodate hash guests
in the P9 path. Radix host, hash guest support is removed from the P7/8
path.
The HPT hcalls and faults are not handled in real mode, which is a
performance regression. A worst-case fork/exit microbenchmark takes 3x
longer after this patch. kbuild benchmark performance is in the noise,
but the slowdown is likely to be noticed somewhere.
For now, accept this penalty for the benefit of simplifying the P7/8
paths and unifying P9 hash with the new code, because hash is a less
important configuration than radix on processors that support it. Hash
will benefit from future optimisations to this path, including possibly
a faster path to handle such hcalls and interrupts without doing a full
exit.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-31-npiggin@gmail.com
Now that the P7/8 path no longer supports radix, real-mode handlers
do not need to deal with being called in virt mode.
This change effectively reverts commit acde25726b ("KVM: PPC: Book3S
HV: Add radix checks in real-mode hypercall handlers").
It removes a few more real-mode tests in rm hcall handlers, which
allows the indirect ops for the xive module to be removed from the
built-in xics rm handlers.
kvmppc_h_random is renamed to kvmppc_rm_h_random to be a bit more
descriptive and consistent with other rm handlers.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-25-npiggin@gmail.com
Dependent-threads mode is the normal KVM mode for pre-POWER9 SMT
processors, where all threads in a core (or subcore) would run the same
partition at the same time, or they would run the host.
This design was mandated by MMU state that is shared between threads in
a processor, so the synchronisation point is in hypervisor real-mode
that has essentially no shared state, so it's safe for multiple threads
to gather and switch to the correct mode.
It is implemented by having the host unplug all secondary threads and
always run in SMT1 mode, and host QEMU threads essentially represent
virtual cores that wake these secondary threads out of unplug when the
ioctl is called to run the guest. This happens via a side-path that is
mostly invisible to the rest of the Linux host and the secondary threads
still appear to be unplugged.
POWER9 / ISA v3.0 has a more flexible MMU design that is independent
per-thread and allows a much simpler KVM implementation. Before the new
"P9 fast path" was added that began to take advantage of this, POWER9
support was implemented in the existing path which has support to run
in the dependent threads mode. So it was not much work to add support to
run POWER9 in this dependent threads mode.
The mode is not required by the POWER9 MMU (although "mixed-mode" hash /
radix MMU limitations of early processors were worked around using this
mode). But it is one way to run SMT guests without running different
guests or guest and host on different threads of the same core, so it
could avoid or reduce some SMT attack surfaces without turning off SMT
entirely.
This security feature has some real, if indeterminate, value. However
the old path is lagging in features (nested HV), and with this series
the new P9 path adds remaining missing features (radix prefetch bug
and hash support, in later patches), so POWER9 dependent threads mode
support would be the only remaining reason to keep that code in and keep
supporting POWER9/POWER10 in the old path. So here we make the call to
drop this feature.
Remove dependent threads mode support for POWER9 and above processors.
Systems can still achieve this security by disabling SMT entirely, but
that would generally come at a larger performance cost for guests.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-23-npiggin@gmail.com
Rather than partition the guest PID space + flush a rogue guest PID to
work around this problem, instead fix it by always disabling the MMU when
switching in or out of guest MMU context in HV mode.
This may be a bit less efficient, but it is a lot less complicated and
allows the P9 path to trivally implement the workaround too. Newer CPUs
are not subject to this issue.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-22-npiggin@gmail.com
Now the initial C implementation is done, inline more HV code to make
rearranging things easier.
And rename __kvmhv_vcpu_entry_p9 to drop the leading underscores as it's
now C, and is now a more complete vcpu entry.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-16-npiggin@gmail.com
Almost all logic is moved to C, by introducing a new in_guest mode for
the P9 path that branches very early in the KVM interrupt handler to P9
exit code.
The main P9 entry and exit assembly is now only about 160 lines of low
level stack setup and register save/restore, plus a bad-interrupt
handler.
There are two motivations for this, the first is just make the code more
maintainable being in C. The second is to reduce the amount of code
running in a special KVM mode, "realmode". In quotes because with radix
it is no longer necessarily real-mode in the MMU, but it still has to be
treated specially because it may be in real-mode, and has various
important registers like PID, DEC, TB, etc set to guest. This is hostile
to the rest of Linux and can't use arbitrary kernel functionality or be
instrumented well.
This initial patch is a reasonably faithful conversion of the asm code,
but it does lack any loop to return quickly back into the guest without
switching out of realmode in the case of unimportant or easily handled
interrupts. As explained in previous changes, handling HV interrupts
very quickly in this low level realmode is not so important for P9
performance, and are important to avoid for security, observability,
debugability reasons.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-15-npiggin@gmail.com
In the interest of minimising the amount of code that is run in
"real-mode", don't handle hcalls in real mode in the P9 path. This
requires some new handlers for H_CEDE and xics-on-xive to be added
before xive is pulled or cede logic is checked.
This introduces a change in radix guest behaviour where radix guests
that execute 'sc 1' in userspace now get a privilege fault whereas
previously the 'sc 1' would be reflected as a syscall interrupt to the
guest kernel. That reflection is only required for hash guests that run
PR KVM.
Background:
In POWER8 and earlier processors, it is very expensive to exit from the
HV real mode context of a guest hypervisor interrupt, and switch to host
virtual mode. On those processors, guest->HV interrupts reach the
hypervisor with the MMU off because the MMU is loaded with guest context
(LPCR, SDR1, SLB), and the other threads in the sub-core need to be
pulled out of the guest too. Then the primary must save off guest state,
invalidate SLB and ERAT, and load up host state before the MMU can be
enabled to run in host virtual mode (~= regular Linux mode).
Hash guests also require a lot of hcalls to run due to the nature of the
MMU architecture and paravirtualisation design. The XICS interrupt
controller requires hcalls to run.
So KVM traditionally tries hard to avoid the full exit, by handling
hcalls and other interrupts in real mode as much as possible.
By contrast, POWER9 has independent MMU context per-thread, and in radix
mode the hypervisor is in host virtual memory mode when the HV interrupt
is taken. Radix guests do not require significant hcalls to manage their
translations, and xive guests don't need hcalls to handle interrupts. So
it's much less important for performance to handle hcalls in real mode on
POWER9.
One caveat is that the TCE hcalls are performance critical, real-mode
variants introduced for POWER8 in order to achieve 10GbE performance.
Real mode TCE hcalls were found to be less important on POWER9, which
was able to drive 40GBe networking without them (using the virt mode
hcalls) but performance is still important. These hcalls will benefit
from subsequent guest entry/exit optimisation including possibly a
faster "partial exit" that does not entirely switch to host context to
handle the hcall.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-14-npiggin@gmail.com
irq_work's use of the DEC SPR is racy with guest<->host switch and guest
entry which flips the DEC interrupt to guest, which could lose a host
work interrupt.
This patch closes one race, and attempts to comment another class of
races.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-11-npiggin@gmail.com
This is more symmetric with kvmppc_xive_push_vcpu, and has the advantage
that it runs with the MMU on.
The extra test added to the asm will go away with a future change.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210528090752.3542186-9-npiggin@gmail.com
