| Age | Commit message (Collapse) | Author |
|
The platforms that support the interrupt from the I2C
adapter can not handle the amount of interrupts the adapter
generates because of the way the IRQ is routed in the
hardware. The I2C controller driver has to be kept in
polling mode because of that.
The AMC MCU can still generate critical alerts that have to
be handled. The interrupt from SMBus Alert is left enabled
and handled separately in the Xe. The alerts from the AMC
will cause the device to be declared wedged for now.
Fixes: f0e53aadd702 ("drm/xe: Support for I2C attached MCUs")
Cc: stable@vger.kernel.org
Reviewed-by: Raag Jadav <raag.jadav@intel.com>
Co-developed-by: Ramesh Babu B <ramesh.babu.b@intel.com>
Signed-off-by: Ramesh Babu B <ramesh.babu.b@intel.com>
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://patch.msgid.link/20260811121008.1493015-3-heikki.krogerus@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
(cherry picked from commit a55b76b8bc2c49b11d753c1c6d06ec3a2c61c85e)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
The default programming in the driver matches the "coverage" blend mode,
so add the corresponding pixel blend mode property to let userspace know
about it.
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patch.msgid.link/20260826105421.1825331-1-thierry.reding@kernel.org
|
|
[Why]
When should_reset_plane() forces a plane's dc_plane_state to be
recreated (e.g. on every ALLOW_MODESET commit on DCN < 3.2),
fill_dc_plane_attributes() re-derives the plane's color pipeline from
the DRM colorop states. The per-stage helpers look up each colorop
with for_each_new_colorop_in_state(), which only iterates colorops
attached to the current commit. If userspace didn't touch color in
that commit the colorops are absent, so the fixed matrix (YUV->RGB),
shaper, 3D LUT and regamma all silently fall back to bypass. This
corrupts the plane's colors - e.g. an HDR P010 video losing its
color-space conversion - until the pipeline happens to be re-committed.
[How]
When recreating a plane that has an active color_pipeline, call
drm_atomic_add_affected_colorops() so the plane's colorops are pulled
into the commit and the derivation sees the current pipeline instead
of defaulting every stage to bypass. This mirrors what the core
already does on a modeset via drm_atomic_add_affected_planes().
Assisted-by: Copilot:claude-opus-4.8
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Link: https://patch.msgid.link/20260825153539.213495-1-harry.wentland@amd.com
|
|
The ROM early-return checks for PQ, Gamma 2.2, and HLG in
mod_color_calculate_degamma_params() do not consider the map_user_ramp
parameter. When map_user_ramp is true (indicating a subsampled surface
that requires post-scaler degamma via GAMCOR), the function still takes
the ROM path, programming PRE_DEGAM instead.
For subsampled formats the chroma channels must be upsampled by the
scaler before degamma is applied. PRE_DEGAM sits before the scaler in
the pipeline (CNVC domain), while GAMCOR sits after it (CM domain).
Using the ROM for subsampled surfaces means degamma is applied to
subsampled chroma before upsampling, producing incorrect results.
Fix by gating the dcn_arch ROM capability checks on !map_user_ramp,
forcing PQ/Gamma22/HLG through the GAMCOR LUT path when the surface
is subsampled.
Fixes: a8bf71649088 ("drm/amd/display: Internal refactoring to abstract color caps")
Cc: stable@vger.kernel.org
Assisted-by: Copilot:claude-opus-4.6
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-11-harry.wentland@amd.com
|
|
DC looks at plane_info structs.
For legacy drm_plane COLOR_ENCODING and COLOR_RANGE this was
done in fill_dc_plane_info_and_addr's call to
fill_plane_color_attributes.
Assisted-by: Claude:claude-sonnet-4.5
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-10-harry.wentland@amd.com
|
|
Previously we checked the plane_state->state's plane_color_pipeline
client config when determining whether fill_plane_color_attributes
should run, i.e., whether we had a plane color pipeline and should
use that code for handling fixed matrix properties, or whether we
should look at legacy plane encoding and range.
The problem is that we also call this during commit_tail, during
which plane_state doesn't have the state backpointer. This meant
that during commit_tail we'd look at the legacy plane COLOR_RANGE
and COLOR_ENCODING values and overwrite what we pulled from the
fixed matrix colorop in atomic_check previously.
Instead pass the drm_atomic_state in explicitly and check that.
Assisted-by:Claude:claude-sonnet-4.5
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-9-harry.wentland@amd.com
|
|
For subsampled formats we need to use GAMCOR instead of
the DEGAM block. The color module can create a LUT for
that if we set map_user_ramp to true. So do that when
we have subsampled formats.
Assisted-by: Claude:claude-sonnet-4.5
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-8-harry.wentland@amd.com
|
|
Add __set_dm_plane_colorop_fixed_matrix() which maps the fixed matrix
type enum to the appropriate dc_color_space for DC programming:
YCBCR601_FULL_RGB -> COLOR_SPACE_YCBCR601
YCBCR601_LIMITED_RGB -> COLOR_SPACE_YCBCR601_LIMITED
YCBCR709_FULL_RGB -> COLOR_SPACE_YCBCR709
YCBCR709_LIMITED_RGB -> COLOR_SPACE_YCBCR709_LIMITED
YCBCR2020_FULL_RGB_NC -> COLOR_SPACE_2020_YCBCR_FULL
YCBCR2020_LIMITED_RGB_NC -> COLOR_SPACE_2020_YCBCR_LIMITED
When the fixed matrix is bypassed, color_space is set to
COLOR_SPACE_UNKNOWN.
Update amdgpu_dm_plane_set_colorop_properties() to process the
fixed matrix colorop first (before DEGAM), matching the new pipeline
order.
Assisted-by: Claude:claude-opus-4.6
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-7-harry.wentland@amd.com
|
|
Insert a fixed matrix colorop as the first operation in the
amdgpu color pipeline, before the existing DEGAM 1D curve. This
allows userspace to select YUV-to-RGB conversion via the color
pipeline for YCbCr framebuffers.
The fixed matrix colorop advertises support for all six YUV-to-RGB
conversion presets:
- YUV601/709/2020 full-range to RGB
- YUV601/709/2020 limited-range to RGB
Assisted-by: Claude:claude-opus-4.6
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-6-harry.wentland@amd.com
|
|
In vkms_plane_atomic_check(), validate that when a color pipeline
is active and the framebuffer format is YUV:
- The first colorop must be DRM_COLOROP_FIXED_MATRIX
- The fixed matrix colorop must not be bypassed
In vkms_plane_atomic_update(), when a color pipeline is active,
read the fixed_matrix_type from the first colorop state and map
it to the appropriate (color_encoding, color_range) pair for the
existing get_conversion_matrix_to_argb_u16() function.
In apply_colorop(), add a no-op case for DRM_COLOROP_FIXED_MATRIX
since the actual YUV-to-RGB conversion is already performed at
format read time via the conversion matrix.
Assisted-by: Claude:claude-opus-4.6
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Reviewed-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-5-harry.wentland@amd.com
|
|
Insert a fixed matrix colorop as the first operation in the
VKMS color pipeline, before the existing 1D curve. This allows
userspace to select YUV-to-RGB conversion via the color pipeline
for YCbCr framebuffers.
The CSC Fixed Matrix colorop advertises support for all six
YUV-to-RGB conversion presets:
- YUV601/709/2020 full-range to RGB
- YUV601/709/2020 limited-range to RGB
Bump MAX_COLOR_PIPELINE_OPS from 4 to 5 to accommodate the new op.
Assisted-by: Claude:claude-opus-4.6
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Reviewed-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-4-harry.wentland@amd.com
|
|
The limited-/studio-range YCbCr to RGB conversion matrices were generated
with colour.matrix_YCbCr(is_legal=True, bits=8), which normalises the
narrow range by 2^bits (256 for 8-bit) rather than by the full-range
maximum 2^bits - 1 (255). As a result the luma scale was 256/219 and the
chroma scale 256/224 instead of the correct 255/219 and 255/224.
This over-scales every limited-range conversion by a factor of 256/255,
producing an error of up to ~3/255 (8-bit) and causing IGT tests that
compare software-computed surfaces against VKMS-processed ones to fail.
Recompute the three limited-range matrices from first principles rather
than relying on colour's is_legal scaling:
1. Start from the standard ITU-R YCbCr -> RGB relations for the luma
weights Kr, Kb (Kg = 1 - Kr - Kb) of each encoding (BT.601, BT.709
and BT.2020 non-constant luminance). These are the same relations
that produce the existing full-range matrices.
2. Expand the studio input range to full range relative to a full-range
maximum of 2^n - 1: the luma coefficient by 255/(235 - 16) and the
chroma coefficients by 255/(240 - 16). This matches the DRM UAPI
definition and IGT's igt_ycbcr_to_rgb_matrix().
3. Convert each coefficient to S31.32 fixed point (round(coeff * 2^32)).
Update the limited-range reference values in the vkms-format KUnit test
accordingly. Provide a python script to show how the kunit test value
are calculated.
Fixes: fe22d21e9342 ("drm/vkms: Add YUV support")
Assisted-by: Copilot:claude-opus-4.8
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Tested-by: Robert Mader <robert.mader@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-3-harry.wentland@amd.com
|
|
Introduce DRM_COLOROP_FIXED_MATRIX, a new colorop type representing a
hardware that performs a fixed matrix operation.
Unlike CTM-based colorops, this block does not expose programmable
coefficients. Instead, userspace selects one of the predefined
hardware modes via a new FIXED_MATRIX_TYPE enum property. Supported modes
include common YCbCr->RGB and RGB709->RGB2020 conversions.
v5:
- Drop YCbCr limited to full and RGB709 to RGB2020 matrices
until we have non-IGT userspace that shows their use (Robert)
v4:
- Add limited-range YCbCr-RGB conversion matrix enums (Robert)
- Document matrix values (Pekka)
- Clarify RGB709 to RGB2020 is a full-range matrix (Pekka)
- Fix confusing "CSC preset" doc (Pekka)
v2:
- Naming changes (Pekka)
v3:
- Fix NC matrix enum name and string (Melissa)
- Rebase
Co-developed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Robert Mader <robert.mader@collabora.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patch.msgid.link/20260814211816.1219597-2-harry.wentland@amd.com
|
|
vecs configuration for newer platforms (media version >= 35) is
different. Fix powergating info for these platforms.
Bspec: 67103, 77977
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Badal Nilawar <badal.nilawar@intel.com>
Link: https://patch.msgid.link/20260818050041.448014-1-ashutosh.dixit@intel.com
|
|
Having scratch enabled does not make a VM capable of handling recoverable
page faults. Allowing scratch VMs through the ASID lookup also admits
dma-fence mode VMs.
If such a VM faults on an already valid VMA, the handler reports success
without fixing the fault, causing the GPU to retry indefinitely.
Only allow fault-mode VMs through the ASID lookup. Fault-mode VMs using
scratch remain supported, while faults from 3D VMs are rejected.
Fixes: ad9843aac91a ("drm/xe/madvise: Implement purgeable buffer object support")
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Suggested-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260820065445.567228-1-arvind.yadav@intel.com
(cherry picked from commit bfb24a06405b652d37831f3fb66b71d33a6605de)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
While reading the main GT powergating info from debugfs, include both
RCS and CCS engine masks.
Fixes: 0914c1e45d3a1 ("drm/xe/xe_gt_idle: add debugfs entry for powergating info")
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Link: https://patch.msgid.link/20260819073457.1812722-2-balasubramani.vivekanandan@intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
(cherry picked from commit 8899e413c5ab85443ec9bbc50cffe924c6b596de)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
The database was updated and the WA is no longer listed as applicable
to media 3503, so don't enable it there.
Fixes: c57db41b8d2c ("drm/xe/guc: Add Wa_14025883347 for GuC DMA failure on reset")
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Sk Anirban <sk.anirban@intel.com>
Cc: Badal Nilawar <badal.nilawar@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/20260818213520.283063-1-daniele.ceraolospurio@intel.com
(cherry picked from commit fae59d5de5de39bc51ac2839f74970312e0c8905)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
Add gpu_test_buddy_dirty_tracker_performance to demonstrate the key
advantage of the decoupled dirty-tracker design over the old dual-tree
/ force_merge approach.
The test runs two scenarios on a 4 GiB pool after alternating
clear/dirty fragmentation at 4 KiB granularity:
1. Contiguous 2 GiB alloc: smaller than the 4 GiB span so it routes
through the contiguous allocator rather than the exact-range fast
path. The old design requires __force_merge() to rebuild the
block; the new design coalesces during free() so the alloc is
O(log N).
old (force_merge) - 45 ms
dirty tracker design - 11 ms
2. Repeated 256 KiB alloc throughput: the old design pays
__force_merge() on every alloc; the new design does not.
old (force_merge) - 86 ms
dirty tracker design - 20 ms
v2:
- Force a contiguous 256 KiB allocation in the repeated-alloc loop so
the baseline actually exercises __force_merge(). (sashiko)
v3:
- Make the first test allocate 2 GiB instead of the full 4 GiB pool,
so it really uses the contiguous path; update the numbers. (sashiko)
Assisted-by: Claude:claude-opus-4-8
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patch.msgid.link/20260825172432.80355-2-Arunpravin.PaneerSelvam@amd.com
|
|
The current buddy allocator maintains separate clear_tree[] and
dirty_tree[] rbtrees per order, preventing coalescing between cleared
and dirty buddies. Under mixed workloads, this creates a merge barrier:
adjacent buddies frequently end up split across trees, forcing reliance
on __force_merge() during allocation.
__force_merge() performs an O(N x max_order) scan under the VRAM manager
lock, leading to allocation stalls and failures for large contiguous
requests even when sufficient total free memory is available.
Solution
Replace the dual-tree design with:
- A single free_tree[order] rbtree for dirty and mixed free blocks
(fully cleared free blocks float outside this tree)
- A lightweight out-of-band dirty tracker (gpu_dirty_tracker)
Fully cleared free blocks are tracked outside the buddy trees using an
augmented interval rbtree, enabling O(log E) lookup of the largest
cleared extents.
Buddy coalescing is now unconditional in __gpu_buddy_free(), regardless
of clear/dirty state. This removes the merge barrier and eliminates the
need for __force_merge().
Benefits
- Correct high-order allocations after mixed clear/dirty workloads
- Elimination of O(N x max_order) merge cost from the allocation path
- O(log E) cleared-extent lookup replacing O(N) scans
- Predictable allocation latency under fragmentation
- Reduced complexity with a single tree per order
Test:
dEQP-VK.memory.allocation.basic.size_8KiB.reverse.count_4000
Below data is from /sys/kernel/debug/dri/1/amdgpu_vram_mm:
Base (dual-tree), before VKCTS test:
order- 6 free: 6 MiB, blocks: 26
order- 5 free: 1 MiB, blocks: 15
order- 4 free: 960 KiB, blocks: 15
order- 3 free: 5 MiB, blocks: 171
order- 2 free: 2 MiB, blocks: 176
order- 1 free: 1 MiB, blocks: 165
order- 0 free: 16 KiB, blocks: 4
Base (dual-tree), after VKCTS test:
order- 6 free: 768 KiB, blocks: 3
order- 5 free: 499 MiB, blocks: 3999
order- 4 free: 250 MiB, blocks: 4001
order- 3 free: 129 MiB, blocks: 4157
order- 2 free: 65 MiB, blocks: 4161
order- 1 free: 63 MiB, blocks: 8138
order- 0 free: 20 KiB, blocks: 5
Dirty tracker, before VKCTS test:
order- 6 free: 4 MiB, blocks: 19
order- 5 free: 2 MiB, blocks: 18
order- 4 free: 704 KiB, blocks: 11
order- 3 free: 5 MiB, blocks: 168
order- 2 free: 2 MiB, blocks: 174
order- 1 free: 1 MiB, blocks: 167
order- 0 free: 32 KiB, blocks: 8
Dirty tracker, after VKCTS test:
order- 6 free: 4 MiB, blocks: 19
order- 5 free: 2 MiB, blocks: 18
order- 4 free: 704 KiB, blocks: 11
order- 3 free: 5 MiB, blocks: 168
order- 2 free: 2 MiB, blocks: 174
order- 1 free: 1 MiB, blocks: 167
order- 0 free: 28 KiB, blocks: 7
v2:
- Code-style cleanup and minor refactoring
- Renamed locals for clarity
v3:
- Keep cleared blocks inside free_tree[] instead of floating them.
- Add subtree_has_dirty rbtree augment for O(log N) dirty-first walk.
v4:
- Fixed checkpatch warnings.
- Optimized gpu_buddy_reset_clear() to a single post-order walk that
flips block headers and recomputes the rbtree augment in one pass.
- Propagate subtree_max_size top-down in insert_extent() so ancestors
are not left with stale values on no-rotation inserts. (sashiko)
- Drop the whole extent in gpu_dirty_tracker_mark_dirty() when the
inside-split allocation fails, avoiding a stale clear claim. (sashiko)
- Make gpu_dirty_tracker_find() alignment-aware and fall back to the
dirty tree on steered failure to avoid spurious -ENOSPC. (sashiko)
v5:
- Track dirty extents instead of cleared ones: steer dirty allocs onto
tracked dirty windows and pick clear allocs via a free-tree augment,
avoiding clear-memory wastage by keeping cleared free blocks untouched
during dirty allocation.
v6:
- Make __alloc_range_bias() return the highest/right-most address by
default, establishing top-down as the intended placement for
range-biased allocations.
- Honour GPU_BUDDY_CLEAR_ALLOCATION in __alloc_range_bias() by steering
the descent towards clear subtrees for non-top-down clear
requests. (sashiko)
- Skip dirty-tracker steering for offset-aligned requests so they keep
their min_block_size alignment. (sashiko)
- sashiko reported that the __GFP_NOFAIL dirty-extent allocations on
the free path could deadlock during memory reclaim, since that is a
GFP_KERNEL allocation on the free path; move to a per-tracker
mempool so extent nodes are guaranteed without __GFP_NOFAIL.
(sashiko)
- Derive each free block's clear/dirty class from the blocks already
in hand on split, free, alloc, trim and init instead of querying the
dirty tracker, removing the tracker lookups from the hot paths.
v7:
- Preserve mixed-block clear state in __gpu_buddy_free() when a mixed
split child is re-merged after an undone split. (sashiko)
- Prefer a fully-clear block over a mixed one of the same order via a
single ordered clear-state max augment on free_tree[].
v8:
- Coalesce contiguous dirty blocks in __gpu_buddy_free_list() into one
dirty extent update instead of one mark_dirty() per block. (Matthew)
v9:
- Reset has_clear on allocation so a mixed block taken whole and later
freed fully dirty is not re-tracked as mixed. (sashiko)
v10:
- Use a plain slab allocation for dirty extents; skip and log once on
failure. (Matthew)
- Assert a non-zero size in the dirty-tracker range helpers. (Matthew)
- Drop the cached clear_avail member; derive it on demand. (Matthew)
- Collapse the two dirty branches of gpu_buddy_reset_clear(). (Matthew)
- Move the gpu_block_state enum above the gpu_buddy_block kernel-doc
so the doc directly precedes its struct. (Matthew)
- Mark the gpu_dirty_tracker struct private. (Matthew)
- Preserve a block's clear state on non-clear allocation instead of
force-dirtying it. (Matthew)
- Drop the redundant header clear in gpu_buddy_block_trim(). (Matthew)
v11:
- Assert the manager lock is held in gpu_buddy_clear_avail(). (Matthew)
Assisted-by: Claude:claude-opus-4-8
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patch.msgid.link/20260825172432.80355-1-Arunpravin.PaneerSelvam@amd.com
|
|
Now that intel_cdclk_{pre,post}_notify() are implemented via vfuncs
there is no need to keep them inside the .set_cdclk() hooks. Move
the calls one level up to intel_cdclk_set_cdclk().
We do need to adjust {skl,bxt}_cdclk_(un)init_hw() to call the wrapper
rather than the low level implementation directly, or else they would
not do the pcode notification anymore.
The two slight functional changes here are:
- bdw_set_cdclk() might theoretically bail out after doing the
pre notification, but that codepath would only come into play
if the hardware is seriously misprogrammed, so should never happen
- cdclk hw readout is still done from .set_cdclk(), so that now
happens before the post notify vs. previously the readout happened
before it. This should not matter as the readout is not affected
by the post notify (since we can't actually read out anything from
pcode).
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-15-ville.syrjala@linux.intel.com
|
|
Turn the cdclk pcode pre/post notify functions into vfuncs.
Mainly to get rid of the hideous if-ladders in bxt_set_cdclk().
DG2 is currently doing its own thing with its pcode notify funcs so
can't be converted yet. And MTL+ go via the pmdemand stuff so this
is all supposedly handled elsewhere.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-14-ville.syrjala@linux.intel.com
|
|
Extract the BXT/GLK pcode notify stuff to a few small helpers.
The plan is to unify these between all the platforms and turn
them into vfuncs.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-13-ville.syrjala@linux.intel.com
|
|
Extract the SKL/ICL+ pcode notify stuff to a few small helpers.
The plan is to unify these between all the platforms and turn
them into vfuncs.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-12-ville.syrjala@linux.intel.com
|
|
Extract the BDW pcode notify stuff to a few small helpers. The
plan is to unify these between all the platforms and turn them
into vfuncs.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-11-ville.syrjala@linux.intel.com
|
|
The debug spew for the cdclk pcode per/post notify is very
inconsistent between different platforms. Unify it all to
the same form.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-10-ville.syrjala@linux.intel.com
|
|
The control flow between the pcode pre and post notifications in
bxt_set_cdclk() is written in two different ways, even though
they end up doing the same thing. Unify the code.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-9-ville.syrjala@linux.intel.com
|
|
intel_pcode_*notify() are all DG2 specific code. Rename them
to have a dg2_ namespace.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-8-ville.syrjala@linux.intel.com
|
|
Overwrite cdclk.hw.voltage_level from intel_update_cdclk() at the
end on bxt_set_cdclk() also on bxt/glk. While this isn't actually
necessary due to bxt/glk not having any extra DDI based voltage
level requirements, it does avoid one less silly 'if' in the code.
On icl+ the value derived by bxt_get_cdclk() may not be correct
if the voltage level was bumped up due to DDI requirements, thus
overwriting the assumed value is crucial there.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-7-ville.syrjala@linux.intel.com
|
|
SKL_CDCLK_PREPARE_FOR_CHANGE == DISPLAY_TO_PCODE_VOLTAGE(3) so
we are currently forcing the voltage level to 3 all the time on
DG2. Remove SKL_CDCLK_PREPARE_FOR_CHANGE from the mask to avoid
this.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-6-ville.syrjala@linux.intel.com
|
|
We're currently skipping the pcode notifies on DG2 if the CDCLK isn't
changing while the power well counts would still need updating.
Do the pcode notifications also for pure pipe power well changes.
The lower level intel_set_cdclk() will skip doing any real work if the
actual CDCLK stuff doesn't need to change, so it's fine to call it even
if only the power wells need updating.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-5-ville.syrjala@linux.intel.com
|
|
The pcode post notify needs to happen after the CDCLK has been
changed, not before. Also move the pre_notify call a bit for the
sake of symmetry.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-4-ville.syrjala@linux.intel.com
|
|
We are currently trying to pass the CDCLK in kHz to the pcode
on DG2, while the pcode expects a value in MHz units. Adjust
the units appropriately.
v2: Call it 'cdclk_mhz' (Jani)
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-3-ville.syrjala@linux.intel.com
|
|
We already changed the actual cdclk frequency by the time we do
the pcode post notify. So skipping the subsequent readout is plain
wrong.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826134033.13976-2-ville.syrjala@linux.intel.com
|
|
dg2_power_well_count() is supposed to return an integer,
not a boolean. Make it so.
Fixes: 9112ce99c1d7 ("drm/i915/cdclk: Extract dg2_power_well_count()")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260826143100.19401-1-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
|
|
Instead of harder to read min_array() and prone to mistakes min_t()
use min3() and min(). No functional changes intended. The types of
the values are unsigned, no side-effects are expected.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patch.msgid.link/20260825093552.4088094-1-andriy.shevchenko@linux.intel.com
|
|
Make the sched_process_csg_irq_locked() call part of
group_unbind_locked() so we don't have to manually call it in
tick_ctx_apply()/panthor_sched_suspend().
This implies moving group_[un]bind_locked() around to avoid a
forward declaration.
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-9-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
Avoid a workqueue roundtrip and process things immediately from
panthor_sched_report_fw_events().
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-8-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
Call group_check_job_completion() directly from
csg_slot_sync_update_locked() instead of deferring it.
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-7-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
Add a specific spinlock for events processing so we can selectively
move some event processing to the threaded IRQ handler. For events to be
processed, we need to have access to the group attached to the CSG slot
which also forces us to protect the csg_slots[] updates with this
lock.
Note that fatal_queues/timedout are turned into atomics to avoid having
to take the events_lock every time those are checked or updated.
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-6-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
This lock is never used, and we're about to make fatal_queues an
atomic to cope with concurrent updates.
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-5-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
By scheduling an immediate tick, we already force idleness re-evaluation,
which gives the scheduler the opportunity to evict idle groups
and schedule onces that have jobs pending.
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-4-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
Now that panthor_irq contains the iomem region, there's no real need
for the macro-based panthor_irq helper generation logic. We can just
provide inline helpers that do the same and let the compiler optimize
indirect function calls. The only extra annoyance is the fact we have
to open-code the panthor_xxx_irq_threaded_handler() implementation, but
those are single-line functions, so it's acceptable.
While at it, we changed the prototype of the IRQ handlers to take
a panthor_irq instead of panthor_device, since that's the thing
that's passed around when it comes to panthor_irq, and the
panthor_device can be directly extracted from there.
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-3-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
We're about to add an IRQ inline helper using gpu_read(). Move things
around to avoid forward declarations.
No functional changes.
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-2-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
The only place where panthor_irq::state is accessed without
panthor_irq::mask_lock held is in the prologue of _irq_suspend(),
which is not really a fast-path. So let's simplify things by assuming
panthor_irq::state must always be accessed with the mask_lock held,
and add a scoped_guard() in _irq_suspend().
While at it, rename the lock so it's clear it doesn't just protect
access to the panthor_irq::mask or the INT_MASK register.
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Link: https://patch.msgid.link/20260825-panthor-signal-from-irq-v7-1-ebb5b4c891a4@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux
Pull another power sequencing update from Bartosz Golaszewski:
"A single tree-wide rename of two of the public functions to better
reflect their actual semantics:
- rename pwrseq_power_on/off() to pwrseq_enable/disable() tree-wide"
* tag 'pwrseq-updates-for-v7.3-rc1-part2' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
power: sequencing: rename pwrseq_power_on/off() to pwrseq_enable/disable()
|
|
The mailbox PHASE bit in SYSCTRL_MB_CTRL is toggled per-message and
was tracked in software as sc->phase_bit, reset to 0 on error paths.
If the cached value ever drifts from what the hardware last saw, all
following messages carry the wrong phase and get silently misread by
firmware.
Drop the cache and read PHASE directly from SYSCTRL_MB_CTRL before
each frame instead, removing xe_sysctrl_mailbox_init() and its call
sites along with it.
Fixes: 1f95f618182b ("drm/xe/xe_sysctrl: Add System Controller mailbox communication support")
Signed-off-by: Anoop Vijay <anoop.c.vijay@intel.com>
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patch.msgid.link/20260825172827.3801591-2-anoop.c.vijay@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
Commit 1c6ceeee6ebb ("drm/atomic: Fix memleak on ERESTARTSYS during
non-blocking commits") fixed a very similar issue when the event was
allocated by drm_atomic_helper_setup_commit() itself.
However, if the event is allocated in prepare_signaling(), it will also be
set to NULL in complete_signaling(), which prevents drm_crtc_commit from
being put in __drm_atomic_helper_crtc_destroy_state().
Dropping the reference when the event is set to NULL at
complete_signaling() fixes the leak.
The leak can be reproduced by sending a signal to the thread using
DRM_MODE_PAGE_FLIP_EVENT and using a sw_sync fence to cause the atomic
ioctl to block at drm_atomic_helper_wait_for_fences(). It happened both
with amdgpu and vkms.
Fixes: 24835e442f28 ("drm: reference count event->completion")
Cc: stable@vger.kernel.org
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Melissa Wen <mwen@igalia.com>
Link: https://patch.msgid.link/20260727-drm_crtc_atomic_commit_leak-v1-1-23d9948a9d7c@igalia.com
|
|
Since file_priv can never be NULL at prepare_signaling() as it is only
called by drm_mode_atomic_ioctl(), remove the check.
If that was not the case, skipping the rest of the block here would cause
the drm_pending_vblank_event object to leak and fail to set up the fence in
case out_fence_ptr is set.
Since the check is unreachable, there is no possible leak.
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Melissa Wen <mwen@igalia.com>
Link: https://patch.msgid.link/20260817-drm_atomic_bogus_check-v2-1-2b9e60f32a7e@igalia.com
|
|
The way power sequencing works means that a call to pwrseq_power_on()
does not necessarily result in the pwrseq target being powered-on at
that time: it may have already been powered on before. Similarly: a call
to pwrseq_power_off() does not have to result in an actual powering off
of resources: there may still be other users that requested a power-on
before.
We will also introduce the concept of "non-controllable" pwrseq targets
soon which further increases the disconnect between the naming
convention and the actual semantics.
What consumers of pwrseq descriptors actually do is: they *vote* for a
powering on of a given target or retract that vote. These operations
could be called get/put in line with runtime PM but this could become
confusing since we already provide pwrseq_get/put() for a different
purpose. pwrseq_vote_on/off() also have been rejected as unusual in
the tree.
Change the name of the two functions to pwrseq_enable/disable() which
better reflects their purpose and semantics and also mirrors other
enable-counted resources like regulators and clocks. No functional change
intended.
If at any point users need to know *when* the exact power event happens,
we can provide that information in the form of a notifier.
Acked-by: Jeff Johnson <jeff.johnson@oss.qualcomm.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>
Acked-by: Alessio Belle <alessio.belle@imgtec.com> # imagination
Link: https://patch.msgid.link/20260731-pwrseq-vote-rename-v3-1-44e60b8be053@oss.qualcomm.com
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
|
|
The simple KMS helpers are deprecated because they only add an
intermediate layer between drivers and atomic modesetting.
Open-code drm_simple_encoder_init() by calling drm_encoder_init()
directly and providing driver-local drm_encoder_funcs.
Signed-off-by: Diogo Silva <diogompaissilva@gmail.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patch.msgid.link/20260720-drm_simple_encoder_init-v2-13-5020b630668a@gmail.com
|