drm/amdgpu/umsch: enable doorbell for umsch

Program vcn_doorbell_range with vcn_ring0_1.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Veerabadhran Gopalakrishnan <Veerabadhran.Gopalakrishnan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: make dc_set_power_state() return type `void` again

As dc_set_power_state() no longer allocates memory, it's not necessary
to have return types and check return code as it can't fail anymore.

Change it back to `void`.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Destroy DC context while keeping DML and DML2

If there is memory pressure at suspend time then dynamically
allocating a large structure as part of DC suspend code will
fail.

Instead re-use the same structures and clear all members except
those that should be maintained.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2362
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Catch errors from drm_atomic_helper_suspend()

drm_atomic_helper_suspend() can return PTR_ERR(), in which case the
error gets stored into `dm->cached_state`.  This can cause failures
during resume.  Catch the error during suspend and fail the suspend
instead.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2362
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Split up UVD suspend into prepare and suspend steps

amdgpu_uvd_suspend() allocates memory and copies objects into that
allocated memory.  This fails under memory pressure.  Instead move
majority of this code into a prepare step when swap can still be
allocated.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Add concept of running prepare_suspend() sequence for IP blocks

If any IP blocks allocate memory during their hw_fini() sequence
this can cause the suspend to fail under memory pressure.  Introduce
a new phase that IP blocks can use to allocate memory before suspend
starts so that it can potentially be evicted into swap instead.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Evict resources during PM ops prepare() callback

Linux PM core has a prepare() callback run before suspend.

If the system is under high memory pressure, the resources may need
to be evicted into swap instead.  If the storage backing for swap
is offlined during the suspend() step then such a call may fail.

So move this step into prepare() to move evict majority of
resources and update all non-pmops callers to call the same callback.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2362
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable GFX IP v11.5.0 CG and PG support

Add CG support for GFX/MC/HDP/ATHUB/IH/BIF.
Add PG support for GFX.

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add support to power up/down UMSCH by SMU

Power up/down UMSCH by SMU.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Acked-by: Veerabadhran Gopalakrishnan <Veerabadhran.Gopalakrishnan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add power up/down UMSCH ppt callback

Add ppt callback to power up/down UMSCH.

v2: squash in updates (Alex)

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add support to powerup VPE by SMU

Powerup VPE by SMU.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/discovery: add SMU 14 support

add smu 14 into the IP discovery list.

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/swsmu: add smu14 ip support

Add initial swSMU support for smu 14 series ASIC.

v2: squash in build fixes and updates (Li Ma)
    fix warnings (Alex)
v3: squash in updates (Alex)
v4: squash in updates (Alex)
v5: squash in avg/current power updates (Alex)

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/swsmu: add smu v14_0_0 pmfw if file

Add initial smu v14_0_0 pmfw if file

v2: squash in updates (Alex)

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/swsmu: add smu v14_0_0 ppsmc file

Add initial smu v14_0_0 ppsmc file

v2: squash in updates (Alex)
v3: squash in updates (Alex)

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: add smu v14_0_0 driver if file

Add initial smu v14_0_0 driver if file

v2: squash in updates (Alex)
v3: update interface (Alex)

Signed-off-by: Li Ma <li.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/umsch: power on/off UMSCH by DLDO

VCN 4.0.5 uses DLDO.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Veerabadhran Gopalakrishnan <Veerabadhran.Gopalakrishnan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/umsch: fix psp frontdoor loading

These changes are missed in rebase.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Veerabadhran Gopalakrishnan <Veerabadhran.Gopalakrishnan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Increase IP discovery region size

IP discovery region has increased to > 8K on some SOCs.Maximum reserve
size is upto 12K, but not used. For now increase to 10K.

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

drm/amd/pm: Use gpu_metrics_v1_4 for SMUv13.0.6

Use gpu_metrics_v1_4 for SMUv13.0.6 to fill
gpu metric info

v3: Removed filling gpu metric instantaneous
pcie bw

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add gpu_metrics_v1_4

Add new gpu_metrics_v1_4 to acquire XGMI data transfer,
pcie bandwidth & Clock lock status

v2:
Add pcie error counter to gpu metric table v1_4

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Update metric table for smu v13_0_6

Update pmfw metric table to include xgmi transfer
data and pci instantaneous bandwidth for smu v13_0_6

v2:
Updated metric table version

v3: Removed inst pcie bw with alignment to metrics table
version 8

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Return -EINVAL when MMSCH init status incorrect

Return -EINVAL when MMSCH init fail which can be handle by function
amdgpu_device_reset_sriov correctly.

Signed-off-by: Lin.Cao <lincao12@amd.com>
Reviewed-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: wait for completion of the EnableGfxImu command

Wait for completion of sending the EnableGfxImu message
when using the PSP FW loading.

Signed-off-by: Tim Huang <Tim.Huang@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vpe: fix insert_nop ops

Avoid infinite loop when count is 0.
This is missed in rebase.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Address member 'gart_placement' not described in 'amdgpu_gmc_gart_location'

Fixes the below:
drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c:274: warning: Function parameter or member 'gart_placement' not described in 'amdgpu_gmc_gart_location'

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vpe: align with mcbp changes

MCBP is decided by adev->gfx.mcbp now.
This is missed in rebase.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vpe: remove IB end boundary requirement

Remove IB end boundary requirement,
VPE has no such limitions, use existing
amdgpu_ring_generic_pad_ib() instead.
This is missed in rebase.

Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Improve MES responsiveness during oversubscription

When MES is oversubscribed it may not frequently check for new
command submissions from driver if the scheduling load is high.
Response latency as high as 5 seconds has been observed.

Enable a flag which adds a check for new commands between
scheduling quantums.

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Cc: Alexandru Tudor <alexandru.tudor@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix SI failure due to doorbells allocation

SI hardware does not have doorbells at all, however currently the code
will try to do the allocation and thus fail, makes SI AMDGPU not usable.

Fix this failure by skipping doorbells allocation when doorbells count
is zero.

Fixes: 54c30d2a8def ("drm/amdgpu: create kernel doorbell pages")
Reviewed-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/discovery: enable DCN 3.5.0 support

Enable DCN 3.5.0 support.

Signed-off-by: Aaron Liu <aaron.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: get doorbell's absolute offset based on the db_size

Here, Adding db_size in byte to find the doorbell's
absolute offset for both 32-bit and 64-bit doorbell sizes.
So that doorbell offset will be aligned based on the doorbell
size.

v2:
- Addressed the review comment from Felix.
v3:
- Adding doorbell_size as parameter to get db absolute offset.
v4:
  Squash the two patches into one.

Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Arvind Yadav <Arvind.Yadav@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add missing NULL check

bo->tbo.resource can easily be NULL here.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2902
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
CC: stable@vger.kernel.org

drm/amd/display: 3.2.255

This version brings along following fixes:
- Refactor DPG test pattern logic for ODM cases
- Refactor HWSS into component folder
- Revert "drm/amd/display: Add a check for idle power optimization"
- Revert "drm/amd/display: remove duplicated edp relink to fastboot
- Update cursor limits based on SW cursor fallback limits
- Update stream mask
- Update pmfw_driver_if new structure
- Modify SMU message logs
- Don't set dpms_off for seamless boot

Known issue:
DWB (Writeback functionality) is broken.
Fix will be available in DC 3.2.256

Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable virtual links

[Why]
This could work around the issue with DP tunneling producing
a NULL pointer dereference.

[How]
Disable the virtual links.

Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Stylon Wang <stylon.wang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable SubVP if test pattern is enabled

[Why]
Enabling DPG causes HUBP to stay in blank constantly. If DPG is enabled
while an MCLK switch is taking place with SubVP, it will cause the MCLK
to never complete. This is because SubVP MCLK switch relies a HUBP
VLine interrupt, which will never occur when HUBP is constantly in
blank.

[How]
Disable SubVP when test pattern is enabled.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Nevenko Stupar <nevenko.stupar@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: George Shen <george.shen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Refactor DPG test pattern logic for ODM cases

[Why]
Current DPG test pattern logic does not account for ODM configuration
changes after test pattern has already been programmed. For example, if
ODM2:1 is enabled after test pattern is already being output, the second
pipe is not programmed to output test pattern, causing half the screen
to be black.

[How]
Move DPG test pattern parameter calculations into separate function.
Whenever ODM pipe configuration changes, re-calculate DPG test pattern
parameters and program DPG if test pattern is currently enabled.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: George Shen <george.shen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't set dpms_off for seamless boot

[Why]
eDPs fail to light up with seamless boot enabled

[How]
When seamless boot is enabled don't configure dpms_off
in disable_vbios_mode_if_required.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Refactor HWSS into component folder

[why]
Rename hw_sequencer to hwseq.
Move all hwseq files to unique
folder hwss.

[how]
creating hwss repo in dc, and moved the dcnxx_hwseq.c
and .h files into corresponding new folders inside the hwss
and cleared the linkage errors by adding relative paths
in the Makefile.template.

Reviewed-by: Martin Leung <martin.leung@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Mounika Adhuri <moadhuri@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Revert "drm/amd/display: Add a check for idle power optimization"

Revert commit 434cf7af492f ("drm/amd/display: Add a check for idle power optimization")
Because it cause Freesync and S4 regression

Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Sung Joon Kim <sungkim@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add hub->ctx_distance in setup_vmid_config

add hub->ctx_distance when read CONTEXT1_CNTL, align w/
write back operation.

v2: fix coding style errors reported by checkpatch.pl (Christian)

Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lang Yu <lang.yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't use fsleep for PSR exit waits

[Why]
These functions can be called from high IRQ levels and the OS will hang
if it tries to use a usleep_highres or a msleep.

[How]
Replace the fsleep with a udelay.

Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update cursor limits based on SW cursor fallback limits

[Why&How]
For determining the cursor size limit, use the same checks that
are used for determining SW cursor fallback instead of only
using SubVP

Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update dml ssb from pmfw clock table

[why]
Need to use real clock table

[How]
Update the clock table

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Muhammad Ahmed <ahmed.ahmed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update stream mask

[Why]
Whenever stream changes because of new
pipe arrangements such as ODM. The new
stream mask is not reflected in DMCUB.

The mismatch in stream mask is blocking ips
entry in some scenarios.

[How]
Whenever stream arrangement changes,
update stream mask and notify DMCUB.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Duncan Ma <duncan.ma@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Revert "drm/amd/display: remove duplicated edp relink to fastboot"

Revert commit 984abb5384b0 ("drm/amd/display: remove duplicated edp relink to fastboot")

Because it cause 4k EDP not light up on boot

Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Modify Vmin default value

Fine tune the Vmin clock value

Reviewed-by: Robin Chen <robin.chen@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Max Tseng <max.tseng@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update pmfw_driver_if new structure

[why]
pmfw header file updated, need align with data structure.

[How]
Update the data structure.

Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: VSIF v3 set Max Refresh Rate

[WHY]
FreeSync spec requires PB8 and PB12 to be set to nominal
refresh rate regardless of fixed rate or variable

[HOW]
Removed the condition that checks and overwrites max refresh rate
and set PB8/PB12 to be set to max refresh rate always

Reviewed-by: Anthony Koo <anthony.koo@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Muhammad Ansari <muhammad.ansari@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Modify SMU message logs

[why]
It's important to make sure SMU messages
are logged by default to improve debugging for
power optimization use cases.

[how]
Change logs to warnings when SMU message
returns non-success id.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Sung Joon Kim <sungkim@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix potential null pointer derefernce

The amdgpu_ras_get_context may return NULL if device
not support ras feature, so add check before using.

Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: enable S/G display for for recent APUs by default

With S/G display becomes stable, enable S/G display for recent APUs
by default rather than white list.

v2: explicitly disable sg on pre-CZ chips (Alex)
v3: add parens for every clause (Alex)

Co-authored-by: Alex Deucher <alexander.deucher@amd.com>
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>

Documentation/amdgpu: Add FRU attribute details

Add documentation for the newly added manufacturer and fru_id attributes
in sysfs.

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

drm/amdgpu: Add more FRU field information

Add support to read Manufacturer Name and FRU File Id fields. Also add
sysfs device attributes for external usage.

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

drm/amdgpu: Refactor FRU product information

Keep FRU related information together in a separate structure.

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

drm/amdgpu: enable FRU device for SMU v13.0.6

v1:
enable GFX v9.4.3 FRU device to query board information.

v2:
use MP1 version to identify different asic

Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: update ib start and size alignment

Update IB starting address alignment and size alignment with correct values
for decode and encode IPs.

Decode IB starting address alignment: 256 bytes
Decode IB size alignment: 64 bytes
Encode IB starting address alignment: 256 bytes
Encode IB size alignment: 4 bytes

Also bump amdgpu driver version for this update.

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add check in validate_only in dml2

[what]
does_configuration_meet_sw_policies check was not done in the
validate_only portion of dml2, so some unsupported modes were passing bw
validation, only to fail the same check later in validate_and_build. now
we add the check to validate_only.

Also add line in dcn35_resource to ensure that value set for
enable_windowed_mpo_odm gets passed to dml.

[why]
Immediate black screen during video playback at 4k144hz. The debugger
showed that we were failing validation in dml on every updateplanes().

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Gabe Teeger <gabe.teeger@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Port replay vblank logic to DML2

Update DML2 with replay vblank logic found in DML1.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Modify Pipe Selection for Policy for ODM

[Why]
There are certain cases during a transition to ODM that might cause
corruption on the display. This occurs when we choose certain pipes in a
particular state.

[How]
We now will store the pipe indexes of the any pipes that might be
problematic to switch to during an ODM transition, and only use them as
a last resort.

Reviewed-by: Dmytro Laktyushkin <dmytro.laktyushkin@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Saaem Rizvi <syedsaaem.rizvi@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: correct dml2 input and dlg_refclk

dc->dml2_options.use_native_pstate_optimization flag will make driver
use dcn32 legacy_svp_drr related tuning. Set this to false fixed the
stutter underflow issue also based on HW suggest disable ODM by default
and let DML choose it.

Reviewed-by: Zhan Liu <zhan.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix Chroma Surface height/width initialization

[why]
Surface height/width for Chroma has another variable that it should be
intialized to, chroma_size. Fixing this will help pass DML2.0 validation
for YCbCr420 tests, DCHB006.109,129, DCHB014.011,012.

[how]
Assign SurfaceHeight/WidthC to chroma_size.height/width

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Sung Joon Kim <sungkim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move stereo timing check to helper

Rework dml2_map_dc_pipes to keep the logic clean.

Reviewed-by: Chaitanya Dhere <chaitanya.dhere@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Split pipe for stereo timings

[Why & How]
DML2 did not carry over DML1 logic that splits pipe for stero timings. Pipe
splitting is needed in this case to pass stereo tests.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use fixed DET Buffer Size

[why]
Regression from DML1.0 where we use differen DET buffer sizes for each
pipe. From the spec, we need to use DET buffer size of 384 kb for each
pipe

[how]
Ensure to use 384 kb DET buffer sizes for each available pipe.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Sung Joon Kim <sungkim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Handle multiple streams sourcing same surface

[why]
There are cases where more than 1 stream can be mapped to the same
surface. DML2.0 does not seem to handle these cases.

[how]
Make sure to account for the stream id when deriving the plane id. By
doing this, each plane id will be unique based on the stream id.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Sung Joon Kim <sungkim@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add z8_marks in dml

Add z8 watermarks to struct for later ASIC use.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add DCN35 DML2 support

Enable DML2 for DCN35.

Changes since V1:
- Remove hard coded values

Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Roman Li <roman.li@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Introduce DML2

DC is transitioning from DML to DML2, and this commit introduces all the
required changes for some of the already available ASICs and adds the
required code infra to support new ASICs under DML2. DML2 is also a
generated code that provides better mode verification and programming
models for software/hardware, and it enables a better way to create
validation tools. This version is more like a middle step to the
complete transition to the DML2 version.

Changes since V1:
- Alex: Fix typos

Changes since V2:
- Update DC includes

Changes since V3:
- Fix 32 bit compilation issues on x86

Changes since V4:
- Avoid compilation of DML2 on some not supported 32-bit architecture
- Update commit message

Co-developed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Co-developed-by: Roman Li <roman.li@amd.com>
Signed-off-by: Roman Li <roman.li@amd.com>
Signed-off-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move dml code under CONFIG_DRM_AMD_DC_FP guard

For some reason, the dml code is not guarded under CONFIG_DRM_AMD_DC_FP
in the Makefile. This commit moves the dml code under the DC_FP guard.

Reviewed-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move bw_fixed from DML folder

bw_fixed does not need any FPU operation, and it is used on DCE and DCN.
For this reason, this commit moves bw_fixed to the basic folder outside
DML.

Reviewed-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move custom_float from DML folder

The custom_float file does not have any FPU operation, so it should be
inside DML. This commit moves the file to the basic folder.

Reviewed-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move dce_calcs from DML folder

dce_calcs does not have FPU operations, and it is required for DCE and
DCN. Remove this file from the DML folder and add it to the basic folder
visible for DCE and DCN.

Reviewed-by: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Enable SMU 13.0.0 optimizations when ROCm is active (v2)

When ROCm is active enable additional SMU 13.0.0 optimizations.
This reuses the unused powersave profile on PMFW.

v2: move to the swsmu code since we need both bits active in
    the workload mask.

Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move the memory allocation out of dcn20_validate_bandwidth_fp().

dcn20_validate_bandwidth_fp() is invoked while FPU access has been
enabled. FPU access requires disabling preemption even on PREEMPT_RT.
It is not possible to allocate memory with disabled preemption even with
GFP_ATOMIC on PREEMPT_RT.

Move the memory allocation before FPU access is enabled.
To preserve previous "clean" state of "pipes" add a memset() before the
second invocation of dcn20_validate_bandwidth_internal() where the
variable is used.

Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move the memory allocation out of dcn21_validate_bandwidth_fp().

dcn21_validate_bandwidth_fp() is invoked while FPU access has been
enabled. FPU access requires disabling preemption even on PREEMPT_RT.
It is not possible to allocate memory with disabled preemption even with
GFP_ATOMIC on PREEMPT_RT.

Move the memory allocation before FPU access is enabled.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=217928
Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add a warning if the FPU is used outside from task context.

Add a warning if the FPU is used from any context other than task
context. This is only precaution since the code is not able to be used
from softirq while the API allows it on x86 for instance.

Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Fix UBSAN array-index-out-of-bounds for Polaris and Tonga

For pptable structs that use flexible array sizes, use flexible arrays.

Link: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2036742
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: Fix UBSAN array-index-out-of-bounds for Radeon HD 5430

For pptable structs that use flexible array sizes, use flexible arrays.

Suggested-by: Felix Held <felix.held@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2894
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Fix UBSAN array-index-out-of-bounds for SMU7

For pptable structs that use flexible array sizes, use flexible arrays.

Suggested-by: Felix Held <felix.held@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2874
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Annotate struct amdgpu_bo_list with __counted_by

Prepare for the coming implementation by GCC and Clang of the __counted_by
attribute. Flexible array members annotated with __counted_by can have
their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for
array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
functions).

As found with Coccinelle[1], add __counted_by for struct amdgpu_bo_list.
Additionally, since the element count member must be set before accessing
the annotated flexible array member, move its initialization earlier.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Cc: David Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc: Luben Tuikov <luben.tuikov@amd.com>
Cc: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: linux-hardening@vger.kernel.org
Link: https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci
Reviewed-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Drop unnecessary return statements

There is no reason to call return at the end of function that
returns void.

Fixes the below:

WARNING: void function return statements are not generally useful

Thus remove such a statement in the affected functions.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Documentation/amdgpu: Add board info details

Add documentation for board info sysfs attribute.

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

drm/amdgpu: Add sysfs attribute to get board info

Add a sysfs attribute which shows the board form factor like OAM or
CEM.

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

drm/amdgpu: Get package types for smuio v13.0

Add support to query package types supported in smuio v13.0 ASICs.

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

drm/amdgpu: Add more smuio v13.0.3 package types

Expand support to get other board types like OAM or CEM.

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

drm/amdgpu: fix ip count query for xcp partitions

fix wrong ip count INFO on spatial partitions. update the query
to return the instance count corresponding to the partition id.

v2:
 initialize variables only when required to be (Christian)
 move variable declarations to the beginning of function (Christian)

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Remove set df cstate for SMUv13.0.6

Remove set df cstate as disallow df state is
not required for SMUv13.0.6

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Move package type enum to amdgpu_smuio

Move definition of package type to amdgpu_smuio header and add new
package types for CEM and OAM.

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

drm/amdgpu: Fix complex macros error

Fixes the below:

ERROR: Macros with complex values should be enclosed in parentheses

WARNING: macros should not use a trailing semicolon
+#define amdgpu_inc_vram_lost(adev) atomic_inc(&((adev)->vram_lost_counter));

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Use partial migrations in GPU page faults

This patch implements partial migration in gpu page fault according to migration
granularity(default 2MB) and not split svm range in cpu page fault handling.
A svm range may include pages from both system ram and vram of one gpu now.
These chagnes are expected to improve migration performance and reduce mmu
callback and TLB flush workloads.

Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Simplify the per-CPU usage.

The fpu_recursion_depth counter is used to ensure that dc_fpu_begin()
can be invoked multiple times while the FPU-disable function itself is
only invoked once. Also the counter part (dc_fpu_end()) is ballanced
properly.

Instead of using the get_cpu_ptr() dance around the inc it is simpler to
increment the per-CPU variable directly. Also the per-CPU variable has
to be incremented and decremented on the same CPU. This is ensured by
the inner-part which disables preemption. This is kind of not obvious,
works and the preempt-counter is touched a few times for no reason.

Disable preemption before incrementing fpu_recursion_depth for the first
time. Keep preemption disabled until dc_fpu_end() where the counter is
decremented making it obvious that the preemption has to stay disabled
while the counter is non-zero.
Use simple inc/dec functions.
Remove the nested preempt_disable/enable functions which are now not
needed.

Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove migrate_en/dis from dc_fpu_begin().

This is a revert of the commit mentioned below while it is not wrong, as
in the kernel will explode, having migrate_disable() here it is
complete waste of resources.

Additionally commit message is plain wrong the review tag does not make
it any better. The migrate_disable() interface has a fat comment
describing it and it includes the word "undesired" in the headline which
should tickle people to read it before using it.
Initially I assumed it is worded too harsh but now I beg to differ.

The reviewer of the original commit, even not understanding what
migrate_disable() does should ask the following:

- migrate_disable() is added only to the CONFIG_X86 block and it claims
  to protect fpu_recursion_depth. Why are the other the architectures
  excluded?

- migrate_disable() is added after fpu_recursion_depth was modified.
  Shouldn't it be added before the modification or referencing takes
  place?

Moving on.
Disabling preemption DOES prevent CPU migration. A task, that can not be
pushed away from the CPU by the scheduler (due to disabled preemption)
can not be pushed or migrated to another CPU.

Disabling migration DOES NOT ensure consistency of per-CPU variables. It
only ensures that the task acts always on the same per-CPU variable. The
task remains preemptible meaning multiple tasks can access the same
per-CPU variable. This in turn leads to inconsistency for the statement

                  *pcpu -= 1;

with two tasks on one CPU and a preemption point during the RMW
operation:

     Task A                           Task B
     read pcpu to reg  # 0
     inc reg           # 0 -> 1
                                      read pcpu to reg  # 0
                                      inc reg           # 0 -> 1
                                      write reg to pcpu # 1
     write reg to pcpu # 1

At the end pcpu reads 1 but should read 2 instead. Boom.

get_cpu_ptr() already contains a preempt_disable() statement. That means
that the per-CPU variable can only be referenced by a single task which
is currently running. The only inconsistency that can occur if the
variable is additionally accessed from an interrupt.

Remove migrate_disable/enable() from dc_fpu_begin/end().

Cc: Tianci Yin <tianci.yin@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Fixes: 0c316556d124 ("drm/amd/display: Disable migration to ensure consistency of per-CPU variable")
Acked-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: amdgpu_drm.h: fix comment typos

Correct typos of "occurred".

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Cc: amd-gfx@lists.freedesktop.org
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine fault cache updates

Don't update the fault cache if status is 0.  In the multiple
fault case, subsequent faults will return a 0 status which is
useless for userspace and replaces the useful fault status, so
only update if status is non-0.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new INFO ioctl query for the last GPU page fault

Add a interface to query the last GPU page fault for the process.
Useful for debugging context lost errors.

v2: split vmhub representation between kernel and userspace
v3: add locking when fetching fault info in INFO IOCTL

Mesa MR: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23238
libdrm MR: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23238

Cc: samuel.pitoiset@gmail.com
Reviewed-by: Christian König <christian.koenig@amd.com>
Acked-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add GC v9.4.3 thermal limits to hwmon

Publish max operating temperature of SOC and memory as temp*_emergency
nodes in hwmon. temp*_crit will show the throttle temperature limits.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Add throttle limit for SMU v13.0.6

CTF limit represents the max operating temperature and thermal limit
gives the limit at which throttling starts. Add support for both limits.
SOC and HBM may have different limit values.*_emergency_max gives  max
operating temperature and *_crit_max value represents throttle limit.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.254

DC v3.2.254

* Stability fixes through null pointer checks, blackscreen fixes on
  certain configurations
* Improvements for various functionality like Idle optimization, ILR, DSC, IPS
* Fixes for ILR functionality
* Trivial code cleanups

===================================================================

Acked-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Adjust code style for hw_sequencer.h

[Why&How]
* Rearrange some definitions for consistency
* Drop legacy code

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