drm/amdgpu: fix queue reset issue by mmio

Initialize the queue type before resetting the queue using mmio.

Signed-off-by: Jesse Zhang <jesse.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add kdoc entry for 'program_isharp_1dlut' in 'dpp401_dscl_program_isharp'

Added a descriptor for the 'program_isharp_1dlut' parameter, which is a
flag used to determine whether to program the isharp 1D LUT.

Fixes the below with gcc W=1:
drivers/gpu/drm/amd/amdgpu/../display/dc/dpp/dcn401/dcn401_dpp_dscl.c:963: warning: Function parameter or struct member 'program_isharp_1dlut' 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>
Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Replace 'amdgpu_job_submit_direct' with 'drm_sched_entity' in cleaner shader

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: d361ad5d2fc0 ("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>

drm/amdgpu/: Add missing kdoc entry in amdgpu_vm_handle_fault function

This commit adds a description for the 'ts' parameter in the
amdgpu_vm_handle_fault function's comment block.

Fixes the below with gcc W=1:
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c:2781: warning: Function parameter or struct member 'ts' not described in 'amdgpu_vm_handle_fault'

Cc: Xiaogang.Chen <Xiaogang.Chen@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202408251419.vgZHg3GV-lkp@intel.com/
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Xiaogang Chen <Xiaogang.Chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dccg root clock optimization related hang

[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>

drm/amd/display: Refactor dccg35_get_other_enabled_symclk_fe

[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>

drm/amdgpu: Normalize reg offsets on JPEG v4.0.3

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>

drm/amd/display: Avoid race between dcn35_set_drr() and dc_state_destruct()

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: 06ad7e164256 ("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>

drm/amd/display: Avoid race between dcn10_set_drr() and dc_state_destruct()

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: 06ad7e164256 ("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>

drm/amdgpu: use clamp() in amdgpu_vm_adjust_size()

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>

drm/amd: use clamp() in amdgpu_pll_get_fb_ref_div()

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>

drm/amdgpu: enable gfxoff quirk on HP 705G4

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>

drm/amdgpu: add raven1 gfxoff quirk

Fix screen corruption with openkylin.

Link: https://bbs.openkylin.top/t/topic/171497
Signed-off-by: Peng Liu <liupeng01@kylinos.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix spelling mistake "recompte" -> "recompute"

There is a spelling mistake in a DRM_DEBUG_DRIVER message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Add cache line size info

Populate cache line size info in topology based on information from IP
discovery table.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Sreekant Somasekharan <Sreekant.Somasekharan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add missing kdoc entry for 'bs_coeffs_updated' in dpp401_dscl_program_isharp

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>

drm/amdgpu: fix invalid fence handling in amdgpu_vm_tlb_flush

CPU based update doesn't produce a fence, handle such cases properly.

Fixes: d8a3f0a0348d ("drm/amdgpu: implement TLB flush fence")
Signed-off-by: Lang Yu <lang.yu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: re-work VM syncing

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>

Revert "drm/amdgpu: align pp_power_profile_mode with kernel docs"

This reverts commit bbb05f8a9cd87f5046d05a0c596fddfb714ee457.

This breaks some manual setting of the profile mode in
certain cases.

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3600
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: use rlc safe mode for soft recovery

Protect the MMIO access with safe mode.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: use rlc safe mode for soft recovery

Protect the MMIO access with safe mode.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx12: use rlc safe mode for soft recovery

Protect the MMIO access with safe mode.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx12: use proper rlc safe mode helpers

Rather than open coding it for the queue reset.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: use proper rlc safe mode helpers

Rather than open coding it for the queue reset.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: use proper rlc safe mode helpers

Rather than open coding it for the queue reset.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx12: per queue reset only on bare metal

It's not supported under SR-IOV at the moment.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: per queue reset only on bare metal

It's not supported under SR-IOV at the moment.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: per queue reset only on bare metal

It's not supported under SR-IOV at the moment.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes11: implement mmio queue reset for gfx11

Implement queue reset for graphic and compute queue.

v2: use amdgpu_gfx_rlc funcs to enter/exit safe mode.
v3: use gfx_v11_0_request_gfx_index_mutex()
v4: fix mutex handling

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>

drm/amdgpu/mes: implement amdgpu_mes_reset_hw_queue_mmio

The reset_queue api could be used from kfd or kgd.

v2: add use_mmio parameter for mes_reset_legacy_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>

drm/amdgpu/mes: modify mes api for mmio queue reset

Add me/pipe/queue parameters for queue reset input.

v2: fix build (Alex)

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>

drm/amdgpu/gfx12: fallback to driver reset compute queue directly

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>

drm/amdgpu/gfx12: add ring reset callbacks

Add ring reset callbacks for gfx and compute.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: rework reset sequence

To match other GFX IPs.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: wait for reset done before remap

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>

drm/amdgpu/gfx10: remap queue after reset successfully

Kiq command unmap_queues only does the dequeueing action.
We have to map the queue back with clean mqd.

v2: fix up error handling (Alex)

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>

drm/amdgpu/gfx10: add ring reset callbacks

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>

drm/amdgpu/gfx11: wait for reset done before remap

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>

drm/amdgpu/gfx11: rename gfx_v11_0_gfx_init_queue()

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>

drm/amdgpu/gfx11: fallback to driver reset compute queue directly (v2)

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>

drm/amd/display: 3.2.299

This version brings along the following:

- DCN35 fixes
- DML2 fixes
- IPS fixes
- ODM fixes
- Miscellaneous cleanups
- MST fixes
- SPL fixes

Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix flickering caused by dccg

Always allow un-gating. Follow legacy workaround for repeated
dppclk dto updates

Reviewed-by: Muhammad Ahmed <ahmed.ahmed@amd.com>
Signed-off-by: Hansen Dsouza <Hansen.Dsouza@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Block timing sync for different signals in PMO

PMO assumes that like timings can be synchronized, but DC only allows
this if the signal types match.

Reviewed-by: Austin Zheng <austin.zheng@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix graphics hang in multi-display mst case

[what]
Graphics hang observed with 3 displays connected to DP2.0 mst dock.

[why]
There's a mismatch in dml and dc between the assignments of hpo link
encoders.

[how]
Add a new array in dml that tracks the current mapping of HPO stream
encoders to HPO link encoders in dc.

Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Gabe Teeger <Gabe.Teeger@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add sharpness control interface

- Add interface for controlling shapness level input into DCN.
- Update SPL to support custom sharpness values.
- Add support for different sharpness values depending on YUV/RGB
  content.

Reviewed-by: Samson Tam <samson.tam@amd.com>
Signed-off-by: Relja Vojvodic <Relja.Vojvodic@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Wait for all pending cleared before full update"

This reverts commit f0b7dcf25834afd17df316367dfe5d4c890c713c.

It is causing graphics hangs.

Reviewed-by: Martin Leung <martin.leung@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable sharpness if HDR Multiplier is too large

[Why]
Certain profiles have higher HDR multiplier than SDR boost max which
is not currently supported

[How]
Disable sharpness for these profiles

Fixes: 1b0ce903fe74 ("drm/amd/display: add improvements for text display and HDR DWM and MPO")
Reviewed-by: Martin Leung <martin.leung@amd.com>
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add dpia debug option to control power management

[Why]
To provide option to dpia control power management

[How]
By adding disable_usb4_pm_support bit field in dpia_debug option to
control dpia power management

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: add ring reset callbacks

Add ring reset callbacks for gfx and compute.

Acked-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: re-enable Dynamic ODM policy

[Why]
Previous disable ODM policy due to underflow issue with sharpener.
Issue is resolved after updating sharpening policy to apply to
both windowed and fullscreen video

[How]
Remove sharpness check disabling Dynamic ODM policy

Reviewed-by: Martin Leung <martin.leung@amd.com>
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Lock DC and exit IPS when changing backlight

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>

drm/amd/display: only trigger BIOS related assert for older ASICs

[Why]
Some asserts are always hit on startup/Pnp when they should only be used
to indicate when something has gone wrong.

[How]
Ignore result of getting function from bios cmd table for newer asics.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Daniel Sa <Daniel.Sa@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix DCN35 set min dispclk logic

[Why]

Setting min dispclk to 50Mhz outside clock lowering function causes
unnecessary calls to SMU to lower dispclk and causes dentist hangs when
there is no stream on the pipes.

[How]

Move the set minimum dispclk logic inside the lowering dispclk if
statement.

Fixes: 234441320552 ("DCN35 set min dispclk to 50Mhz")
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>

drm/amdgpu/gfx9.4.3: Implement compute pipe reset

Implement the compute pipe reset, and the driver will
fallback to pipe reset when queue reset fails.
The pipe reset only deactivates the queue which is
scheduled in the pipe, and meanwhile the MEC pipe
will be reset to the firmware _start pointer. So,
it seems pipe reset will cost more cycles than the
queue reset; therefore, the driver tries to recover
by doing queue reset first.

Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: always allocate cleared VRAM for GEM allocations

This adds allocation latency, but aligns better with user
expectations.  The latency should improve with the drm buddy
clearing patches that Arun has been working on.

In addition this fixes the high CPU spikes seen when doing
wipe on release.

v2: always set AMDGPU_GEM_CREATE_VRAM_CLEARED (Christian)

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3528
Fixes: a68c7eaa7a8f ("drm/amdgpu: Enable clear page functionality")
Acked-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com> (v1)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Cc: Christian König <christian.koenig@amd.com>

drm/amdgpu/mes: add mes mapping legacy queue switch

For mes11 old firmware has issue to map legacy queue,
add a flag to switch mes to map legacy queue.

Fixes: f9d8c5c7855d ("drm/amdgpu/gfx: enable mes to map legacy queue support")
Reported-by: Andrew Worsley <amworsley@gmail.com>
Link: https://lists.freedesktop.org/archives/amd-gfx/2024-August/112773.html
Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Don't drain ih1 for APU

ih1 is not initialized for APUs. Don't drain it or NULL pointer
error will be triggered.

Fixes: 6ef29715ac06 ("drm/amdkfd: Change kfd/svm page fault drain handling")
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx12: return early in preempt_ib()

When MES is enabled KIQ is not available.  Return an error
when someone uses the debugfs preempt test interface in
that case.

Acked-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx11: return early in preempt_ib()

When MES is enabled KIQ is not available.  Return an error
when someone uses the debugfs preempt test interface in
that case.

Acked-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Determine IPS mode by ASIC and PMFW versions

[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>

drm/amdgpu: Move the dumping log out of for loop

log message "Dumping IP State Completed" needs to
be logged only once when state dumping is complete.

Hence moving it out of the for loop.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Acked-by: Trigger Huang <Trigger.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: move drain_workqueue before shutdown is set

[background] when unloading amdgpu driver right after running a
workload, drain_workqueue is causing "Fence fallback timer
expired on ring sdma0.0". Under sriov, this issue will cause sriov
full access timeout and a reset happening.

move drain_workqueue before shutdown is set to allow ih process and
before enter full access under sriov to avoid full access time cost.

Signed-off-by: Victor Zhao <Victor.Zhao@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Do core dump immediately when job tmo

Do the coredump immediately after a job timeout to get a closer
representation of GPU's error status.

V2: This will skip printing vram_lost as the GPU reset is not
happened yet (Alex)

V3: Unconditionally call the core dump as we care about all the reset
functions(soft-recovery and queue reset and full adapter reset, Alex)

V4: Do the dump after adev->job_hang = true (Sunil)

Signed-off-by: Trigger Huang <Trigger.Huang@amd.com>
Acked-by: Sunil Khatri <sunil.khatri@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: skip printing vram_lost if needed

The vm lost status can only be obtained after a GPU reset occurs, but
sometimes a dev core dump can be happened before GPU reset. So a new
argument is added to tell the dev core dump implementation whether to
skip printing the vram_lost status in the dump.
And this patch is also trying to decouple the core dump function from
the GPU reset function, by replacing the argument amdgpu_reset_context
with amdgpu_job to specify the context for core dump.

V2: Inform user if VRAM lost check is skipped so users don't assume
VRAM wasn't lost (Alex)

Signed-off-by: Trigger Huang <Trigger.Huang@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx9: put queue resets behind a debug option

Pending extended validation.

Reviewed-and-tested-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Acked-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add experimental resets debug flag

Add this flag to enable experimental resets for testing before they
are fully validated.

Reviewed-and-tested-by: Jiadong Zhu <Jiadong.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: Fix a mistake in revert commit

[why]
It is to fix in try_disable_dsc() due to misrevert of
commit 338567d17627 ("drm/amd/display: Fix MST BW calculation Regression")

[How]
Fix restoring minimum compression bw by 'max_kbps', instead of native bw 'stream_kbps'

Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: always force a state reprogram on init

Always reprogram the hardware state on init.  This ensures
the PMFW state is explicitly programmed and we are not relying
on the default PMFW state.

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3131
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: remove unnecessary TODO spl_os_types.h

Remove unnecessary TODO from spl_os_types.h

Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: SPDX copyright for spl_os_types.h

Use appropriate SPDX copyright for spl_os_types.h

Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add DSC Debug Log

Add DSC log in each critical routines to facilitate debugging.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.298

This version brings along the following fixes:
- Fix MS/MP mismatches in dml21 for dcn401
- Resolved Coverity issues
- Add back quality EASF and ISHARP and dc dependency changes
- Add sharpness support for windowed YUV420 video
- Add improvements for text display and HDR DWM and MPO
- Fix Synaptics Cascaded Panamera DSC Determination
- Allocate DCN35 clock table transfer buffers in GART
- Add Replay Low Refresh Rate parameters in dc type

Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add sharpness support for windowed YUV420 video

[Why]
Previous only applied sharpness for fullscreen YUV420 video.

[How]
Remove fullscrene restriction and apply sharpness for windowed
 YUV420 video as well.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add improvements for text display and HDR DWM and MPO

[Why]
Tune settings for improved text display.
Handle differences between DWM and MPO in HDR path.

[How]
Update sharpener LBA table.
Use HDR multiplier to calculate scalar matrix coefficients
 for HDR RGB MPO path.
Update unit tests.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add Replay Low Refresh Rate parameters in dc type.

Why:
To supported Low Refresh Rate panel for Replay Feature,
Adding some parameters to record Low Refresh Rate information.

Reviewed-by: Robin Chen <robin.chen@amd.com>
Signed-off-by: Dennis Chan <dennis.chan@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add back quality EASF and ISHARP and dc dependency changes

[Why]
Addressed previous issues with quality changes and new issues due to
 rolling back quality changes.

[How]
This reverts commit f9e6759888866748f31b6b6c2142a481d587f51f, fixes merge conflicts, and fixed some
 formatting errors.
Store current sharpness level for each pregen table to minimize
 calculating sharpness table every time.
Disable dynamic ODM when sharpness is enabled.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Notify DMCUB of D0/D3 state

[Why]
We want to avoid arming the HPD timer in firmware when preparing for
S0i3 entry when DC is considered in D3.

[How]
Notify DMCUB of the power state transitions so it can decide to arm
the HPD timer for idle in DCN35 only in D0.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Ovidiu Bunea <Ovidiu.Bunea@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix Synaptics Cascaded Panamera DSC Determination

Synaptics Cascaded Panamera topology needs to unconditionally
acquire root aux for dsc decoding.

Reviewed-by: Roman Li <roman.li@amd.com>
Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Retry Replay residency

[Why]
Because sometime DMUB GPINT will time out,
 it will cause we return 0 as residency number.

[How]
Retry to avoid this happened.

Reviewed-by: Robin Chen <robin.chen@amd.com>
Signed-off-by: ChunTao Tso <ChunTao.Tso@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Allocate DCN35 clock table transfer buffers in GART

[Why]
Request from PMFW to use GART for clock table transfer tables as
framebuffer is being deprecated on APU.

[How]
Switch over to GART via the allocation flag.

Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: do not set traslate_by_source for DCN401 cursor

translate_by_source need not be set for DCN401 onwards since cursor
cursor composition comes after scaler in the hardware pipeline.
Hence offset calculation has been reworked, and this setting is not
necessary to be enabled anymore.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Resolve Coverity Issues

[WHY]
Remove coverity issues that were originally ignored.

[HOW]
Ran coverity locally on driver, used output report to find existing
coverity issues, resolved them

Reviewed-by: Nicholas Choi <nicholas.choi@amd.com>
Signed-off-by: Daniel Sa <Daniel.Sa@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix MS/MP mismatches in dml21 for dcn401

[WHY]
Prefetch calculations did not guarantee that bandwidth required in
mode support was less than mode programming which can cause failures.

[HOW]
Fix bandwidth calculations to assume fixed times for OTO schedule,
and choose which schedule to use based on time to fetch pixel data.

Reviewed-by: Jun Lei <jun.lei@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Wait for all pending cleared before full update

[Description]
Before every full update we must wait for all pending updates to be
cleared - this is particularly important for minimal transitions
because if we don't wait for pending cleared, it will be as if
there was no minimal transition at all. In OTG we must read 3 different
status registers for pending cleared, one specifically for OTG updates,
one specifically for OPTC updates, and the last for surface related
updates

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: guard write a 0 post_divider value to HW

[why]
post_divider_value should not be 0.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Ahmed, Muhammad <Ahmed.Ahmed@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't skip clock updates in overclocking

[Description]
Skipping clock updates is not a hard requirement for overclocking
and only an optimization. Remove the skip as this can cause issues
for FAMS transitions during the overclock sequence. If FAMS
is enabled we must disable UCLK switch on any full update (which
requires update clocks to be called).

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Introduce additional IPS debug flags

[Why]

Idle power states (IPS) describe levels of power-gating within DCN. DM
and DC is responsible for ensuring that we are out of IPS before any DCN
programming happens. Any DCN programming while we're in IPS leads to
undefined behavior (mostly hangs).

Because IPS intersects with all display features, the ability to disable
IPS by default while ironing out the known issues is desired. However,
disabing it completely will cause important features such as s0ix entry
to fail.

Therefore, more granular IPS debug flags are desired.

[How]

Extend the dc debug mask bits to include the available list of IPS
debug flags.

All the flags should work as documented, with the exception of
IPS_DISABLE_DYNAMIC. It requires dm changes which will be done in
later changes.

v2: enable docs and fix docstring format

Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu13.0.7: print index for profiles

Print the index for the profiles.

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3543
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: fix ordering for setting workload_mask

No change in functionality for the current code, but we
need to set the index properly before changing it if we
ever use a non-0 index.

Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: align pp_power_profile_mode with kernel docs

The kernel doc says you need to select manual mode to
adjust this, but the code only allows you to adjust it when
manual mode is not selected.  Remove the manual mode check.

Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'amd-drm-next-6.12-2024-08-26' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.12-2024-08-26:

amdgpu:
- SDMA devcoredump support
- DCN 4.0.1 updates
- DC SUBVP fixes
- Refactor OPP in DC
- Refactor MMHUBBUB in DC
- DC DML 2.1 updates
- DC FAMS2 updates
- RAS updates
- GFX12 updates
- VCN 4.0.3 updates
- JPEG 4.0.3 updates
- Enable wave kill (soft recovery) for compute queues
- Clean up CP error interrupt handling
- Enable CP bad opcode interrupts
- VCN 4.x fixes
- VCN 5.x fixes
- GPU reset fixes
- Fix vbios embedded EDID size handling
- SMU 14.x updates
- Misc code cleanups and spelling fixes
- VCN devcoredump support
- ISP MFD i2c support
- DC vblank fixes
- GFX 12 fixes
- PSR fixes
- Convert vbios embedded EDID to drm_edid
- DCN 3.5 updates
- DMCUB updates
- Cursor fixes
- Overdrive support for SMU 14.x
- GFX CP padding optimizations
- DCC fixes
- DSC fixes
- Preliminary per queue reset infrastructure
- Initial per queue reset support for GFX 9
- Initial per queue reset support for GFX 7, 8
- DCN 3.2 fixes
- DP MST fixes
- SR-IOV fixes
- GFX 9.4.3/4 devcoredump support
- Add process isolation framework
- Enable process isolation support for GFX 9.4.3/4
- Take IOMMU remapping into account for P2P DMA checks

amdkfd:
- CRIU fixes
- Improved input validation for user queues
- HMM fix
- Enable process isolation support for GFX 9.4.3/4
- Initial per queue reset support for GFX 9
- Allow users to target recommended SDMA engines

radeon:
- remove .load and drm_dev_alloc
- Fix vbios embedded EDID size handling
- Convert vbios embedded EDID to drm_edid
- Use GEM references instead of TTM
- r100 cp init cleanup
- Fix potential overflows in evergreen CS offset tracking

UAPI:
- KFD support for targetting queues on recommended SDMA engines
  Proposed userspace:
  https://github.com/ROCm/ROCR-Runtime/commit/2f588a24065f41c208c3701945e20be746d8faf7
  https://github.com/ROCm/ROCR-Runtime/commit/eb30a5bbc7719c6ffcf2d2dd2878bc53a47b3f30

drm/buddy:
- Add start address support for trim function

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240826201528.55307-1-alexander.deucher@amd.com

Merge v6.11-rc5 into drm-next

amdgpu pr conconflicts due to patches cherry-picked to -fixes, I might
as well catch up with a backmerge and handle them all. Plus both misc
and intel maintainers asked for a backmerge anyway.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Merge tag 'drm-intel-gt-next-2024-08-23' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-next

UAPI Changes:

- Limit the number of relocations to INT_MAX (Tvrtko)

  Only impact should be synthetic tests.

Driver Changes:

- Fix for #11396: GPU Hang and rcs0 reset on Cherrytrail platform
- Fix Virtual Memory mapping boundaries calculation (Andi)
- Fix for #11255: Long hangs in buddy allocator with DG2/A380 without
  Resizable BAR since 6.9 (David)
- Mark the GT as dead when mmio is unreliable (Chris, Andi)
- Workaround additions / fixes for MTL, ARL and DG2 (John H, Nitin)
- Enable partial memory mapping of GPU virtual memory (Andi, Chris)

- Prevent NULL deref on intel_memory_regions_hw_probe (Jonathan, Dan)
- Avoid UAF on intel_engines_release (Krzysztof)

- Don't update PWR_CLK_STATE starting Gen12 (Umesh)
- Code and dmesg cleanups (Andi, Jesus, Luca)

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZshcfSqgfnl8Mh4P@jlahtine-mobl.ger.corp.intel.com

Merge tag 'drm-misc-next-2024-08-22' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-next

drm-misc-next for v6.12:

Cross-subsystem Changes:

string:
- add mem_is_zero()

Core Changes:

edid:
- use mem_is_zero()

Driver Changes:

ast:
- reorganize output code by type (VGA, DP, etc)
- convert to struct drm_edid
- fix BMC handling for all outputs

bridge:
- anx7625: simplify OF array handling
- dw-hdmi: simplify clock handling
- lontium-lt8912b: fix mode validation
- nwl-dsi: fix mode vsync/hsync polarity

panel:
- ili9341: fix comments
- jd9365da: fix "exit sleep" commands
- jdi-fhd-r63452: simplify error handling with DSI multi-style
  helpers
- mantix-mlaf057we51: simplify error handling with DSI multi-style
  helpers
- simple: support Innolux G070ACE-LH3 plus DT bindings; support
  On Tat Industrial Company KD50G21-40NT-A1 plus DT bindings
- st7701: decouple DSI and DRM code; add SPI support; support Anbernic
  RG28XX plus DT bindings

vc4:
- fix PM during detect
- replace DRM_ERROR() with drm_error()
- v3d: simplify clock retrieval

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240822150710.GA243952@localhost.localdomain

Linux 6.11-rc5

Merge tag 'bcachefs-2024-08-24' of git://evilpiepirate.org/bcachefs

Pull bcachefs fixes from Kent Overstreet:

 - assorted syzbot fixes

 - some upgrade fixes for old (pre 1.0) filesystems

 - fix for moving data off a device that was switched to durability=0
   after data had been written to it.

 - nocow deadlock fix

 - fix for new rebalance_work accounting

* tag 'bcachefs-2024-08-24' of git://evilpiepirate.org/bcachefs: (28 commits)
  bcachefs: Fix rebalance_work accounting
  bcachefs: Fix failure to flush moves before sleeping in copygc
  bcachefs: don't use rht_bucket() in btree_key_cache_scan()
  bcachefs: add missing inode_walker_exit()
  bcachefs: clear path->should_be_locked in bch2_btree_key_cache_drop()
  bcachefs: Fix double assignment in check_dirent_to_subvol()
  bcachefs: Fix refcounting in discard path
  bcachefs: Fix compat issue with old alloc_v4 keys
  bcachefs: Fix warning in bch2_fs_journal_stop()
  fs/super.c: improve get_tree() error message
  bcachefs: Fix missing validation in bch2_sb_journal_v2_validate()
  bcachefs: Fix replay_now_at() assert
  bcachefs: Fix locking in bch2_ioc_setlabel()
  bcachefs: fix failure to relock in btree_node_fill()
  bcachefs: fix failure to relock in bch2_btree_node_mem_alloc()
  bcachefs: unlock_long() before resort in journal replay
  bcachefs: fix missing bch2_err_str()
  bcachefs: fix time_stats_to_text()
  bcachefs: Fix bch2_bucket_gens_init()
  bcachefs: Fix bch2_trigger_alloc assert
  ...

Merge tag '6.11-rc5-server-fixes' of git://git.samba.org/ksmbd

Pull smb server fixes from Steve French:

 - query directory flex array fix

 - fix potential null ptr reference in open

 - fix error message in some open cases

 - two minor cleanups

* tag '6.11-rc5-server-fixes' of git://git.samba.org/ksmbd:
  smb/server: update misguided comment of smb2_allocate_rsp_buf()
  smb/server: remove useless assignment of 'file_present' in smb2_open()
  smb/server: fix potential null-ptr-deref of lease_ctx_info in smb2_open()
  smb/server: fix return value of smb2_open()
  ksmbd: the buffer of smb2 query dir response has at least 1 byte

Merge tag 's390-6.11-4' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Vasily Gorbik:

 - Fix KASLR base offset to account for symbol offsets in the vmlinux
   ELF file, preventing tool breakages like the drgn debugger

 - Fix potential memory corruption of physmem_info during kernel
   physical address randomization

 - Fix potential memory corruption due to overlap between the relocated
   lowcore and identity mapping by correctly reserving lowcore memory

 - Fix performance regression and avoid randomizing identity mapping
   base by default

 - Fix unnecessary delay of AP bus binding complete uevent to prevent
   startup lag in KVM guests using AP

* tag 's390-6.11-4' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/boot: Fix KASLR base offset off by __START_KERNEL bytes
  s390/boot: Avoid possible physmem_info segment corruption
  s390/ap: Refine AP bus bindings complete processing
  s390/mm: Pin identity mapping base to zero
  s390/mm: Prevent lowcore vs identity mapping overlap

Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "The important core fix is another tweak to our discard discovery
  issues. The off by 512 in logical block count seems bad, but in fact
  the inline was only ever used in debug prints, which is why no-one
  noticed"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: sd: Do not attempt to configure discard unless LBPME is set
  scsi: MAINTAINERS: Add header files to SCSI SUBSYSTEM
  scsi: ufs: qcom: Add UFSHCD_QUIRK_BROKEN_LSDBS_CAP for SM8550 SoC
  scsi: ufs: core: Add a quirk for handling broken LSDBS field in controller capabilities register
  scsi: core: Fix the return value of scsi_logical_block_count()
  scsi: MAINTAINERS: Update HiSilicon SAS controller driver maintainer

bcachefs: Fix rebalance_work accounting

rebalance_work was keying off of the presence of rebelance_opts in the
extent - but that was incorrect, we keep those around after rebalance
for indirect extents since the inode's options are not directly
available

Fixes: 20ac515a9cc7 ("bcachefs: bch_acct_rebalance_work")
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>