If a queue is being destroyed but causes a HWS hang on removal, the KFD
may issue an unnecessary gpu reset if the destroyed queue can be fixed
by a queue reset.
This is because the queue has been removed from the KFD's queue list
prior to the preemption action on destroy so the reset call will fail to
match the HQD PQ reset information against the KFD's queue record to do
the actual reset.
To fix this, deactivate the queue prior to preemption since it's being
destroyed anyways and remove the queue from the KFD's queue list after
preemption.
Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Enables users to update SVM's default granularity, used in
buffer migration and handling of recoverable page faults.
Param value is set in terms of log(numPages(buffer)),
e.g. 9 for a 2 MIB buffer
Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit replaces the use of amdgpu_job_submit_direct which submits
the job to the ring directly, with drm_sched_entity in the cleaner
shader job submission process. The change allows the GPU scheduler to
manage the cleaner shader job.
- The job is then submitted to the GPU using the
drm_sched_entity_push_job function, which allows the GPU scheduler to
manage the job.
This change improves the reliability of the cleaner shader job
submission process by leveraging the capabilities of the GPU scheduler.
Fixes: d361ad5d2f ("drm/amdgpu: Add sysfs interface for running cleaner shader")
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
enable dpp rcg before we disable dppclk in hw_init cause system
hang/reboot
[How]
we remove dccg rcg related code from init into a separate function and
call it after we init pipe
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Qili Lu <qili.lu@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Function used to check the number of FEs connected to the current BE.
This was then used to determine if the symclk could be disabled, if
all FEs were disconnected. However, the function would skip over the
primary FE and return 0 when the primary FE was still connected. This
caused black screens on driver disable with an MST daisy chain hooked
up.
[How]
Refactor the function to correctly return the number of FEs connected
to the input BE. Also, rename it for clarity.
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Nicholas Susanto <Nicholas.Susanto@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
On VFs and SOCs with GC 9.4.4, VCN RRMT is disabled.
Only local register offsets should be used on JPEG v4.0.3 as they cannot
handle remote access to other AIDs. Since only local offsets are used,
the special write to MCM_ADDR register is no longer needed.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
dc_state_destruct() nulls the resource context of the DC state. The pipe
context passed to dcn35_set_drr() is a member of this resource context.
If dc_state_destruct() is called parallel to the IRQ processing (which
calls dcn35_set_drr() at some point), we can end up using already nulled
function callback fields of struct stream_resource.
The logic in dcn35_set_drr() already tries to avoid this, by checking tg
against NULL. But if the nulling happens exactly after the NULL check and
before the next access, then we get a race.
Avoid this by copying tg first to a local variable, and then use this
variable for all the operations. This should work, as long as nobody
frees the resource pool where the timing generators live.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3142
Fixes: 06ad7e1642 ("drm/amd/display: Destroy DC context while keeping DML and DML2")
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
dc_state_destruct() nulls the resource context of the DC state. The pipe
context passed to dcn10_set_drr() is a member of this resource context.
If dc_state_destruct() is called parallel to the IRQ processing (which
calls dcn10_set_drr() at some point), we can end up using already nulled
function callback fields of struct stream_resource.
The logic in dcn10_set_drr() already tries to avoid this, by checking tg
against NULL. But if the nulling happens exactly after the NULL check and
before the next access, then we get a race.
Avoid this by copying tg first to a local variable, and then use this
variable for all the operations. This should work, as long as nobody
frees the resource pool where the timing generators live.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3142
Fixes: 06ad7e1642 ("drm/amd/display: Destroy DC context while keeping DML and DML2")
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Tested-by: Raoul van Rüschen <raoul.van.rueschen@gmail.com>
Tested-by: Christopher Snowhill <chris@kode54.net>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Tested-by: Sefa Eyeoglu <contact@scrumplex.net>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When it needs to get a value within a certain interval, using clamp()
makes the code easier to understand than min(max()).
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Li Zetao <lizetao1@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When it needs to get a value within a certain interval, using clamp()
makes the code easier to understand than min(max()).
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Li Zetao <lizetao1@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Enabling gfxoff quirk results in perfectly usable
graphical user interface on HP 705G4 DM with R5 2400G.
Without the quirk, X server is completely unusable as
every few seconds there is gpu reset due to ring gfx timeout.
Signed-off-by: Peng Liu <liupeng01@kylinos.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit addresses a missing kdoc for the 'bs_coeffs_updated'
parameter in the 'dpp401_dscl_program_isharp' function. The
'bs_coeffs_updated' is a flag indicating whether the Blur and Scale
Coefficients have been updated.
The 'dpp401_dscl_program_isharp' function is responsible for programming
the isharp, which includes setting the isharp filter, noise gain, and
blur and scale coefficients. If the 'bs_coeffs_updated' flag is set to
true, the function updates the blur and scale coefficients.
Fixes the below with gcc W=1:
drivers/gpu/drm/amd/amdgpu/../display/dc/dpp/dcn401/dcn401_dpp_dscl.c:961: warning: Function parameter or struct member 'bs_coeffs_updated' not described in 'dpp401_dscl_program_isharp'
Cc: Tom Chung <chiahsuan.chung@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Alex Hung <alex.hung@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Suggested-by: Tom Chung <chiahsuan.chung@amd.com>
Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Rework how VM operations synchronize to submissions. Provide an
amdgpu_sync container to the backends instead of an reservation
object and fill in the amdgpu_sync object in the higher layers
of the code.
No intended functional change, just prepares for upcomming changes.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Acked-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Backlight updates require aux and/or register access. Therefore, driver
needs to disallow IPS beforehand.
So, acquire the dc lock before calling into dc to update backlight - we
should be doing this regardless of IPS. Then, while the lock is held,
disallow IPS before calling into dc, then allow IPS afterwards (if it
was previously allowed).
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Reviewed-by: Roman Li <roman.li@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 988fe28626)
Cc: stable@vger.kernel.org # 6.10+
[Why]
DCN IPS interoperates with other system idle power features, such as
Zstates.
On DCN35, there is a known issue where system Z8 + DCN IPS2 causes a
hard hang. We observe this on systems where the SBIOS allows Z8.
Though there is a SBIOS fix, there's no guarantee that users will get it
any time soon, or even install it. A workaround is needed to prevent
this from rearing its head in the wild.
[How]
For DCN35, check the pmfw version to determine whether the SBIOS has the
fix. If not, set IPS1+RCG as the deepest possible state in all cases
except for s0ix and display off (DPMS). Otherwise, enable all IPS
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 28d43d0895)
Cc: stable@vger.kernel.org
Since the MES FW resets kernel compute queue always failed, this
may caused by the KIQ failed to process unmap KCQ. So, before MES
FW work properly that will fallback to driver executes dequeue and
resets SPI directly. Besides, rework the ring reset function and make
the busy ring type reset in each function respectively.
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is a racing condition that cp firmware modifies
MQD in reset sequence after driver updates it for
remapping. We have to wait till CP_HQD_ACTIVE becoming
false then remap the queue.
v2: fix KIQ locking (Alex)
v3: fix KIQ locking harder (Jessie)
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Add ring reset callbacks for gfx and compute.
v2: fix gfx handling
v3: wait for KIQ to complete
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is a racing condition that cp firmware modifies
MQD in reset sequence after driver updates it for
remapping. We have to wait till CP_HQD_ACTIVE becoming
false then remap the queue.
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Rename to gfx_v11_0_kgq_init_queue() to better align with
the other naming in the file.
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Since the MES FW resets kernel compute queue always failed, this
may caused by the KIQ failed to process unmap KCQ. So, before MES
FW work properly that will fallback to driver executes dequeue and
resets SPI directly. Besides, rework the ring reset function and make
the busy ring type reset in each function respectively.
Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
