Support commits via the flip queue (as opposed to DSB or MMIO).
As it's somewhat unknown if we can actually use it is currently
gated behind the new use_flipq modparam, which defaults to disabled.
The implementation has a bunch of limitations that would need
real though to solve:
- disabled when PSR is used
- disabled when VRR is used
- color management updates not performed via the flip queue
v2: Don't use flip queue if there is no dmc
v3: Use intel_flipq_supported()
v3: Configure PKG_C_LATENCY appropriately
Ignore INT_VECTOR if there is a real PIPEDMC interrupt
(nothing in the hw appears to clear INT_VECTOR)
v4: Leave added_wake_time=0 when flip queue isn't used, to
avoid needleslly increasing pkg_c_latency on lnl/ptl due
to Wa_22020432604. This is a bit racy though...
Use IS_DISPLAY_VER()
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250624170049.27284-6-ville.syrjala@linux.intel.com
The current PKG_C_LATENCY stuff looks busted in several ways:
- doesn't account for multiple pipes from different commits
correctly
- WM_LINETIME is in units of 0.125usec, PKG_C_LATENCY wants
units on 1 usec
- weird VRR state stuff being checked
- use of pointless RMW
Fix it all up. Note that it's still a bit unclear how all this
works, especially how the added_wake_time ties into the flipq
triggers in DMC, and how we need to sequence updates to
PKG_C_LATENCY when enabling/disabling pipes/etc. We may also
need to think what to about the WM1+ disabling and the related
PSR chicken bits when we can use PKG_C_LATENCY for early wake...
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250624170049.27284-3-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Add intel_dbuf_pmdemand_needs_update() helper to avoid looking at struct
intel_dbuf_state internals outside of skl_watermark.c.
With this, we can also move to_intel_dbuf_state(),
intel_atomic_get_old_dbuf_state(), and intel_atomic_get_new_dbuf_state()
inside skl_watermark.c.
Reviewed-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/b493f259d0d3db047151fee18d7e801ad469fa88.1750847509.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
When KMSAN is enabled, this function causes has a rather excessive stack usage:
drivers/gpu/drm/i915/display/skl_watermark.c:2977:1: error: stack frame size (1432) exceeds limit (1408) in 'skl_compute_wm' [-Werror,-Wframe-larger-than]
This is apparently all caused by the varargs calls to drm_dbg_kms(). Inlining
this into skl_compute_wm() means that any function called by skl_compute_wm()
has its own stack on top of that.
Move the worst bit into a separate function marked as noinline_for_stack to
limit that to the one code path that actually needs it.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20250620113748.3869160-1-arnd@kernel.org
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With the struct drm_device based pcode interface in place in both i915
and xe, we can switch display code to use that, and ditch a number of
struct drm_i915_private uses. Also drop the dependency on i915_drv.h
from a couple of files.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/f948fad1b8208522e15140692c17cf493ef305d9.1750678991.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
This is a scripted split of the display related register macros from
i915_reg.h to display/intel_display_regs.h. As a starting point, move
all the macros that are only used in display code (or GVT). If there are
users in core i915 code or soc/, or no users anywhere, keep the macros
in i915_reg.h. This is done in groups of macros separated by blank
lines, moving the comments along with the groups.
Some manually picked macro groups are kept/moved regardless of the
heuristics above.
This is obviously a very crude approach. It's not perfect. But there are
4.2k lines in i915_reg.h, and its refactoring has ground to a halt. This
is the big hammer that splits the file to two, and enables further
cleanup.
Cc: Suraj Kandpal <suraj.kandpal@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com> # v2
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250606102256.2080073-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Add a function to get the (const) pointer to struct dram_info, and use
that to obtain the pointer instead of poking at i915->dram_info
directly.
Clean up a couple of local variables while at it.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Link: https://lore.kernel.org/r/4174edf649e2f6805dab6fd6ce2ec10f4e5f2498.1748337870.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
If there are no changes to intel_crtc_can_enable_sagv() there
is no need to do all the sagv bw_state recomputation.
The only slight caveat here is hw state takeover where we
initially disable SAGV, and want it to get re-enabled once
we've determined that it's safe to do so. That can now be
achieved by having intel_crtc_can_enable_sagv() reject SAGV
as long as the crtc_state->inherited flag is set. Once the
flag gets cleared (during initial commit for inactive pipes,
during the first userspace commit for active pipes), we
will naturally recompute all the sagv related state.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-14-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
The only thing between the current intel_bw_check_sagv_mask() call
site and intel_bw_atomic_check() is skl_wm_add_affected_planes()
which no longer depends on the sagv mask, so we can make life
a lot less confusing by calling intel_bw_check_sagv_mask() from
intel_bw_atomic_check() instead.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-12-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Move the bw_state->pipe_sagv_reject computation into intel_bw.c
where it belongs.
Previously we had a complicated dance between watermarks and
sagv which required this to be computed earlier, but that was
changed in commit 5e8146251f7b ("extract intel_bw_check_sagv_mask()")
which allows the whole thing to be cleaned up quite a bit.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-11-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Currently, during the computation of global watermarks, the latency for
each scaler user is calculated to compute the DSC prefill latency.
At this point, the number of scaler users can exceed the number of
supported scalers, which is checked later in intel_atomic_setup_scalers().
This can cause issues when the number of scaler users exceeds the number
of supported scalers.
While checking for DSC prefill, ensure that the number of scaler users does
not exceed the number of supported scalers.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4341
Fixes: a9b14af999 ("drm/i915/dsc: Check if vblank is sufficient for dsc prefill")
Cc: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250227034106.1638203-1-ankit.k.nautiyal@intel.com
Move the dbuf_state readout parts into skl_wm_get_hw_state()
so that the details are better hidden from sight.
This will stop updating this on pre-skl, but that's what we want
since the dbuf state is only used on skl+.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306163420.3961-18-ville.syrjala@linux.intel.com
I'll need to move the wm readout to an earlier point in the
sequence (since the bw state readout will need ddb information
from the wm readout). But (at least for now) the wm sanitation
will need to stay put as it needs to also sanitize things for
any pipes/planes we disable later during the hw state takeover.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306163420.3961-16-ville.syrjala@linux.intel.com
Add skl_wm_plane_disable_noatomic() which will clear out all
the ddb and wm state for the plane. And let's do this _before_
we call plane->disable_arm() so that it'll actually clear out
the state in the hardware as well.
Currently this won't do anything new for most of the
intel_plane_disable_noatomic() calls since those are done before
wm readout, and thus everything wm/ddb related in the state
will still be zeroed anyway. The only difference will be for
skl_dbuf_sanitize() is happens after wm readout. But I'll be
reordering thigns so that wm readout happens earlier and at that
point this will guarantee that we still clear out the old
wm/ddb junk from the state.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306163420.3961-8-ville.syrjala@linux.intel.com
Update the ddb tracking information when we disable a pipe
during sanitization. Avoids leaving stale junk in the states.
Currently this doesn't do anything as we haven't read out this
state yet when we do the sanitization, but that will change soon.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306163420.3961-7-ville.syrjala@linux.intel.com
Hoist the dbuf stuff into a separate function from
intel_crtc_disable_noatomic_complete() so that the details
are better hidden inside skl_watermark.c.
We can also skip the whole thing on pre-skl since the dbuf state
isn't actually used on those platforms. The readout path does
still fill dbuf_state->active_pipes but we'll remedy that later.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306163420.3961-6-ville.syrjala@linux.intel.com
Convert the remaining intel_display_power.h interfaces to
take struct intel_display instead of struct drm_i915_private.
intel_display_power.c still has some internal uses due to
i915->runtime_pm.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250211000135.6096-3-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Pass intel_display to the display power stuff. These are spread
all over the place so tend to hinder clean conversions of whole
files.
TODO: The gt part/unpark power domain shenanigans need some
kind of more abstract interface...
v2: Deal with cmtg
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250206185533.32306-7-ville.syrjala@linux.intel.com
High refresh rate panels which may have small line times
and vblank sizes, Check if vblank size is sufficient for
dsc prefill latency.
--v2:
- Consider chroma downscaling factor in latency calculation. [Ankit]
- Replace with appropriate function name.
--v3:
- Remove FIXME tag.[Ankit]
- Replace Ycbcr444 to Ycbcr420.[Ankit]
- Correct precision. [Ankit]
- Use some local valiables like linetime_factor and latency to
adjust precision.
- Declare latency to 0 initially to avoid returning any garbage values.
- Account for second scaler downscaling factor as well. [Ankit]
--v4:
- Improvise hscale and vscale calculation. [Ankit]
- Use appropriate name for number of scaler users. [Ankit]
- Update commit message and rebase.
- Add linetime and cdclk prefill adjustment calculation. [Ankit]
--v5:
- Update bspec link in trailer. [Ankit]
- Correct hscale, vscale datatype. [Ankit]
- Use intel_crtc_compute_min_cdclk. [Ankit]
--v6:
- Use cdclk_state->logical.cdclk instead of
intel_crtc_compute_min_cdclk. [Ankit]
--v7:
- Fix linetime calculation. [Ankit]
- Reduce redandancy use of variables. [Ankit]
- Fix typos. [Ankit]
- Update calculation for precision. [Ankit]
--v8:
- Initialise variable to return garbage later. [Ankit]
- Initialise few variables to use at local loop, where
it is used. [Ankit]
Bspec: 70151
Signed-off-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250120172209.188488-8-mitulkumar.ajitkumar.golani@intel.com
High refresh rate panels which may have small line times
and vblank sizes, Check if vblank size is sufficient for
enabled scaler users.
--v2:
- Use hweight* family of functions for counting bits. [Jani]
- Update precision handling for hscale and vscale. [Ankit]
- Consider chroma downscaling factor during latency
calculation. [Ankit]
- Replace function name from scaler_prefill_time to
scaler_prefill_latency.
--v3:
- hscale_k and vscale_k values are already left shifted
by 16, after multiplying by 1000, those need to be right
shifted to 16. [Ankit]
- Replace YCBCR444 to YCBCR420. [Ankit]
- Divide by 1000 * 1000 in end to get correct precision. [Ankit]
- Initialise latency to 0 to avoid any garbage.
--v4:
- Elaborate commit message and add Bspec number. [Ankit]
- Improvise latency calculation. [Ankit]
- Use ceiling value for down scaling factor when less than 1
as per bspec. [Ankit]
- Correct linetime calculation. [Ankit]
- Consider cdclk prefill adjustment while prefill
computation.[Ankit]
--v5:
- Add Bspec link in commit message trailer. [Ankit]
- Correct hscale, vscale data type.
- Use intel_crtc_compute_min_cdclk. [Ankit]
--v6:
- Update FIXME comment.
- Use cdclk_state->logical.cdclk instead of
intel_crtc_compute_min_cdclk. [Ankit]
--v7:
- Handle error return from cdclk_prefill_adjustment. [Ankit]
- Avoid incorrect round off for linetime. [Ankit]
- Correct precision. [Ankit]
--v8:
- Remove redundancy calculation added from previous patch. [Ankit]
Bspec: 70151
Signed-off-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250120172209.188488-7-mitulkumar.ajitkumar.golani@intel.com
Fix all typos in files under drm/i915/display reported by codespell tool.
v2:
- Include british and american spelling, as those are
not typos.
- Fix commenting style. <Jani>
v3: Fix "In case" wrongly capitalized and
also fix comment style. <Krzysztof Niemiec>
Signed-off-by: Nitin Gote <nitin.r.gote@intel.com>
Reviewed-by: Krzysztof Niemiec <krzysztof.niemiec@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250120081517.3237326-8-nitin.r.gote@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Increase the PKG_C_LATENCY Pkg C Latency field by the added wake time.
v1: Initial version.
v2: Rebase and cosmetic changes.
v3:
- Place latency adjustment early to accommodate round-up. [Suraj]
- Modify commit description and cosmetic change. [Suraj]
WA: 22020432604
Cc: Suraj Kandpal <suraj.kandpal@intel.com>
Signed-off-by: Animesh Manna <animesh.manna@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250106094408.1011063-1-animesh.manna@intel.com
Cursor planes do not have the minimum dbuf configuration. The
ddb allocation and registration routines handle this correctly.
But the plane wm handling for cursor need to be differentiated
for this case. Auto minimum buf enable bit should not be
enabled for cursor wm registers in xe3.
Fixes: a831920c37 ("drm/i915/xe3: Use hw support for min/interim ddb allocations for async flip")
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241213120357.300584-1-vinod.govindapillai@intel.com
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Xe3 is capable of switching automatically to min ddb allocation
(not using any extra blocks) or interim SAGV-adjusted allocation
in case if async flip is used. Introduce the minimum and interim
ddb allocation configuration for that purpose. Also i915 is
replaced with intel_display within the patch's context
v2: update min/interim ddb declarations and handling (Ville)
update to register definitions styling
consolidation of minimal wm0 check with min DDB check
Bspec: 69880, 72053
Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241121112726.510220-4-vinod.govindapillai@intel.com
Add a helper to the relative data rate handling in skl_watermarks.c
where other similar functions are implemented. Also get rid of
use_min_ddb() and use use_minimal_wm0() instead to decide whether
the relative data rate can be returned as 0
v2: re-phrase the commit description (uma)
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241121112726.510220-3-vinod.govindapillai@intel.com
Future platforms can have new additions in the plane_wm
registers. So update skl_wm_level_from_reg_val() to have
possiblity for such platform differentiations. This is in
preparation for the rest of the patches in this series where
hw support for the minimum and interim ddb allocations for
async flip is added. Replace all the i915 uses to intel_display
in this function while updating this function
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241121112726.510220-2-vinod.govindapillai@intel.com
Refactor program_dpkgc function so that all initialized variables
of same type are clubbed together.
--v2
-Modify commit message to reflect what is being done in patch [Mitul]
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-6-suraj.kandpal@intel.com
Increase the latency programmed into PKG_C_LATENCY latency to be
a multiple of line time which is written into WM_LINETIME.
--v2
-Fix commit subject line [Sai Teja]
-Use individual DISPLAY_VER checks instead of range [Sai Teja]
-Initialize max_linetime [Sai Teja]
--v3
-take into account the scenario when adjusted_latency is 0 [Vinod]
--v4
-rename adjusted_latency to latency [Mitul]
-fix the condition in which dpkgc is disabled [Vinod]
--v5
-Add check to see if max_linetime is 0 [Vinod]
--v6
-Avoid nested if statements [Mitul]
WA: 22020299601
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-5-suraj.kandpal@intel.com
Refactor the code to check the fixed refresh rate condition in the dpkgc
function itself and call it from intel_atomic_commit_tail so that we
have all the required values specially linetime which is computed after
intel_wm_compute, this will also help implement some WA's which requires
linetime. This also avoid writing into any of the registers while we are
in compute_config phase.
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-4-suraj.kandpal@intel.com
Use intel_display for DPKGC code wherever we can. While we are
at it also use intel_de_rmw instead of intel_uncore_rmw as we
really don't need the internal uncore_rmw_function.
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-3-suraj.kandpal@intel.com
Refactor the value getting prepped to be written into the PKG_C_LATENCY
register by ORing the REG_FIELD_PREP values instead of having val
getting operated on twice.
We dont need the clear and val variables to be initialized.
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-2-suraj.kandpal@intel.com
Initialize max_latency variable to LNL_PKG_C_LATENCY_MASK which helps
to eliminate the else block and make the whole code a lot cleaner.
--v2
-Seprate patch to club variables together [Mitul]
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241203084706.2126189-1-suraj.kandpal@intel.com
