If we don't end up initializing the fences, free them when
we free the job. We can't set the hw_fence to NULL after
emitting it because we need it in the cleanup path for the
submit direct case.
v2: take a reference to the fences if we emit them
v3: handle non-job fence in error paths
Fixes: db36632ea5 ("drm/amdgpu: clean up and unify hw fence handling")
Reviewed-by: Jesse Zhang <Jesse.Zhang@amd.com> (v1)
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
During gpu reset, all GPU-related resources are
inaccessible. To avoid affecting ras functionality,
suspend ras module before gpu reset and resume
it after gpu reset is complete.
V2:
Rename functions to avoid misunderstanding.
V3:
Move flush_delayed_work to amdgpu_ras_process_pause,
Move schedule_delayed_work to amdgpu_ras_process_unpause.
V4:
Rename functions.
V5:
Move the function to amdgpu_ras.c.
Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
add wrapper functions for pmfw eeprom interface, for these interfaces
to be easily and safely called
Signed-off-by: Gangliang Xie <ganglxie@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
add function to check if pmfw is supported, skip eeprom
check and recover when pmfw eeprom is supported
Signed-off-by: Gangliang Xie <ganglxie@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
To fix regression report on gfx8, which requires the exhaustive search
path for signaled event.
The high CPU usage of KFD interrupt wq issue is gone after HIP/ROCr add
option to reduce HW event interrupts, safe to revert this optimization
patch now.
This reverts commit de844846f7.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
If process is killed. the vm entity is stopped, submit pt update job
will trigger the error message "*ERROR* Trying to push to a killed
entity", job will not execute.
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Use a schema similar to gpu metrics v1.9 for partition metrics also. It
will have field type encoded followed by the field value(s). The
attribute ids used will be shared with gpu metrics. The structure
definition is only to distinguish between gpu metrics and partition
metrics though both gpu metrics v1.9 and partition metrics v1.1 follow
the same definition.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
BO's reservation object must be locked before using
amdgpu_vm_bo_update_shared otherwise dma_resv_assert_held will
complain in amdgpu_vm_update_shared.
Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
../ras/rascore/ras_cper.c: In function ‘cper_generate_fatal_record.isra’:
../ras/rascore/ras_cper.c:75:36: error: ‘%llX’ directive output may be truncated writing between 1 and 14 bytes into a region of size between 0 and 7 [-Werror=format-truncation=]
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~
../ras/rascore/ras_cper.c:75:32: note: directive argument in the range [0, 72057594037927935]
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~~~~~~
../ras/rascore/ras_cper.c:75:9: note: ‘snprintf’ output between 4 and 27 bytes into a destination of size 9
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
76 | RAS_LOG_SEQNO_TO_BATCH_IDX(trace->seqno));
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../ras/rascore/ras_cper.c: In function ‘cper_generate_runtime_record.isra’:
../ras/rascore/ras_cper.c:75:36: error: ‘%llX’ directive output may be truncated writing between 1 and 14 bytes into a region of size between 0 and 7 [-Werror=format-truncation=]
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~
../ras/rascore/ras_cper.c:75:32: note: directive argument in the range [0, 72057594037927935]
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~~~~~~
../ras/rascore/ras_cper.c:75:9: note: ‘snprintf’ output between 4 and 27 bytes into a destination of size 9
75 | snprintf(record_id, 9, "%d:%llX", dev_info.socket_id,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
76 | RAS_LOG_SEQNO_TO_BATCH_IDX(trace->seqno));
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
Signed-off-by: Xiang Liu <xiang.liu@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This version brings along following update:
- HDCP2 FW locality check refactors
- Fix black screen issue with HDMI output
- Increase IB mem size
- Revert max buffered cursor size to 64
- Extend inbox0 lock to run Replay / PSR
- Refactor VActive implementation
- Add Pstate viewport reduction
- Persist stream refcount through restore
Acked-by: Wayne Lin <Wayne.Lin@amd.com>
Signed-off-by: Taimur Hassan <Syed.Hassan@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
This fixes the black screen issue on certain APUs with HDMI,
accompanied by the following messages:
amdgpu 0000:c4:00.0: amdgpu: [drm] Failed to setup vendor info
frame on connector DP-1: -22
amdgpu 0000:c4:00.0: [drm] Cannot find any crtc or sizes [drm]
Cannot find any crtc or sizes
Fixes: 489f0f600c ("drm/amd/display: Fix DVI-D/HDMI adapters")
Suggested-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
The buffered cursor cap is expressed assuming a square cursor, and usage
of the cursor buffer is limited by the request size. For greater than 32
pixels, the request size is fixed at 256 bytes, so the maximum width
must be floored to the nearest 256th byte. At 4bpp this means even with
24kB DCN4 can only hold a 64x64 cursor in the buffer as even 65 pixels
would require 512 bytes per line instead of 256.
Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
Overwriting the refcount on stream restore can lead to double-free errors
or memory leaks if an unbalanced number of retains and releases occurs
between a backup and restore.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
Refactors VActive accounting in PMO, and breaks down fill time
requirement by P-State type as it can result in drasitcally different
bandwidth requirements depending on the blackout length.
Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Austin Zheng <Austin.Zheng@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
The inbox1 infrastructure is deprecated, so to support display
power features requiring a DMUB interlock moving forward extend
the inbox0 locking conditions to also include Replay or PSR.
[How]
Implemented a series of changes to improve HW lock handling:
- Deprecated should_use_dmub_inbox1_lock() and guarded it with
DCN401 flag.
- Migrated lock checks into inbox0 helpers and added PSR/Replay
enablement checks to ensure correct behavior.
- Updated HWSS fast update path to acquire HW lock as needed
using the new helpers.
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Andrew Mazour <Andrew.Mazour@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
There are some new changes for HDCP2 firmware locality check. The
implementation doesn't perfectly fit the intended design and clarity.
1. Clarify and consolidate variable responsibilities.
The previous implementation introduced the following variables:
- config.ddc.funcs.atomic_write_poll_read_i2c (optional pointer)
- hdcp->config.ddc.funcs.atomic_write_poll_read_aux (optional pointer)
- hdcp->connection.link.adjust.hdcp2.force_sw_locality_check (bool)
- hdcp->config.debug.lc_enable_sw_fallback (bool)
- use_fw (bool)
They will be used together to determine two operations:
- Whether to use FW locality check
- Whether to use SW fallback on FW locality check failure
The refactor streamlines this by introducing two variables in the hdcp2
link adjustment, while ensuring function pointers are always assigned
and remain independent from policy decisions:
- use_fw_locality_check (bool) -> true if fw locality should be used.
- use_sw_locality_fallback (bool) -> true to reset use_fw_locality_check
back to false and retry on fw locality check failure.
2. Mixed meanings of l_prime_read transition input
l_prime_read originally means if l_prime is read when sw locality check
is used. When FW locality check is used, l_prime_read means if lc init
write, l prime poll and l_prime read combo operation is successful. The
mix of meanings is confusing. The refactor introduces a new variable
l_prime_combo_read to isolate the second meaning into its own variable.
3. Missing specific error code on firmware locality error.
The original change reuses the generic DDC failure error code when
firmware fails to return locality check result. This is not ideal as
DDC failure indicates an error occurred during an I2C/AUX transaction.
FW locality failure could be caused by polling timeout in firmware or
failure to acquire firmware access. Which sits at a higher level of
abstraction above DDC hardware. An incorrect error code could mislead
the debug into a wrong direction.
4. Correcting misplaced comments. The previous implementation of the
firmware locality check resulted in some comments in hdcp2_transition
being incorrectly positioned. This refactor relocates those comments to
their appropriate locations for better clarity.
Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Signed-off-by: Ray Wu <ray.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch adds robust reset handling for user queues (userq) to improve
recovery from queue failures. The key components include:
1. Queue detection and reset logic:
- amdgpu_userq_detect_and_reset_queues() identifies failed queues
- Per-IP detect_and_reset callbacks for targeted recovery
- Falls back to full GPU reset when needed
2. Reset infrastructure:
- Adds userq_reset_work workqueue for async reset handling
- Implements pre/post reset handlers for queue state management
- Integrates with existing GPU reset framework
3. Error handling improvements:
- Enhanced state tracking with HUNG state
- Automatic reset triggering on critical failures
- VRAM loss handling during recovery
4. Integration points:
- Added to device init/reset paths
- Called during queue destroy, suspend, and isolation events
- Handles both individual queue and full GPU resets
The reset functionality works with both gfx/compute and sdma queues,
providing better resilience against queue failures while minimizing
disruption to unaffected queues.
v2: add detection and reset calls when preemption/unmaped fails.
add a per device userq counter for each user queue type.(Alex)
v3: make sure we hold the adev->userq_mutex when we call amdgpu_userq_detect_and_reset_queues. (Alex)
warn if the adev->userq_mutex is not held.
v4: make sure we have all of the uqm->userq_mutex held.
warn if the uqm->userq_mutex is not held.
v5: Use array for user queue type counters.(Alex)
all of the uqm->userq_mutex need to be held when calling detect and reset. (Alex)
v6: fix lock dep warning in amdgpu_userq_fence_dence_driver_process
v7: add the queue types in an array and use a loop in amdgpu_userq_detect_and_reset_queues (Lijo)
v8: remove atomic_set(&userq_mgr->userq_count[i], 0).
it should already be 0 since we kzalloc the structure (Alex)
v9: For consistency with kernel queues, We may want something like:
amdgpu_userq_is_reset_type_supported (Alex)
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
commit 978fa2f6d0 ("drm/amd/display: Use scaling for non-native
resolutions on eDP") started using the GPU scaler hardware to scale
when a non-native resolution was picked on eDP. This scaling was done
to fill the screen instead of maintain aspect ratio.
The idea was supposed to be that if a different scaling behavior is
preferred then the compositor would request it. The not following
aspect ratio behavior however isn't desirable, so adjust it to follow
aspect ratio and still try to fill screen.
Note: This will lead to black bars in some cases for non-native
resolutions. Compositors can request the previous behavior if desired.
Fixes: 978fa2f6d0 ("drm/amd/display: Use scaling for non-native resolutions on eDP")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4538
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Properly Check for return values from calls to debug functions in
runtime_disable().
v2: storing the last non zero returned value from the loop.
Signed-off-by: Sunday Clement <Sunday.Clement@amd.com>
Reviewed-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
If any hardware IPs involved with the second phase of suspend fail, unwind
all steps to restore back to original state.
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
If any hardware IPs involved with the first phase of suspend fail, unwind
all steps to restore back to original state.
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For S3 on vangogh, PMFW needs to be notified before the
driver powers down RLC. This already happens in smu_disable_dpms()
so drop the superfluous call in amdgpu_device_suspend().
Co-developed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The severity of CPER for BP threshold exceed event should be set as
FATAL to match the OOB implementation.
Signed-off-by: Xiang Liu <xiang.liu@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Correct valid_bits and ms_chk_bits of section info field for bad page
threshold exceed CPER to match OOB's behavior.
Signed-off-by: Xiang Liu <xiang.liu@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Commit e6a8a000cf ("drm/amd/display: Rename dml2 to dml2_0 folder")
renames the directory dml2 to dml2_0 in ./drivers/gpu/drm/amd/display/dc,
but misses to adjust the file entry in AMD DISPLAY CORE - DML.
Adjust the file entry after this directory renaming.
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The contents of si_dpm.h seem to have been copied from the
old radeon driver, including a lot of structs and fields which
were only relevant to GPU generations even older than SI.
A lot of these can be deleted without causing much churn to the
actual SI DPM code. Let's delete them to make the code easier
to understand.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
