drm/amd/display: Fix MST topology debugfs

[why]
The drm dump_topology function was previously called on all
DP connectors. This resulted in empty topology dumps for those
connectors which weren't root MST nodes.

[how]
Make sure we only dump topology from the root MST node.

Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Reviewed-by: Aurabindo Jayamohanan Pillai <Aurabindo.Pillai@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: LTTPR config logic

[WHY]
Some systems can enable LTTPR through bits in BIOS, while other systems
can be configured at boot to enable LTTPR. Some configs enable
Non-Transparent mode, while others enable Transparent mode.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enumerate LTTPR modes

[WHY]
There are three possible modes for LTTPR:
- Non-LTTPR mode, where AUX timeout is 400 us and no per-hop link
training is done
- LTTPR Transparent mode, where AUX timeout is 3200 us and no per-hop
link training is done
- LTTPR Non-Transparent mode, where AUX timeout is 3200 us and per-hop
link training is done

[HOW]
Use an enum instead of a bool to track LTTPR state; modify comparisons
accordingly.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Interface for LTTPR interop

[WHY]
The logic to toggle LTTPR transparent/non-transparent requires 2 flags
provided by BIOS

[HOW]
Repurpose the interface to get dce caps so both LTTPR querying functions
can use them.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Rename fs_params to hdr_tm_params

[Why&How]
Renaming structure to better indicate its meaning.

Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Acked-by: Anthony Koo <Anthony.Koo@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix typo for variable name

[why]
Word "remainder" was misspelled as "reminder" in
reduceSizeAndFraction method variable.

[how]
Fix the spelling.

Signed-off-by: Vladimir Stempen <vladimir.stempen@amd.com>
Reviewed-by: Alexander Deucher <alexander.deucher@amd.com>
Reviewed-by: Bindu R <Bindu.R@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add mod hdcp interface for supporting encryption state query

Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: define mod_hdcp_display_disable_option struct

Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: enable DP DSC Compliance automation

[Why]
Color depth data is not parsed during test requests.

[How]
Update display color depth according to color depth
request from the test equipment.

Signed-off-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Guard ASSR with internal display flag

[Why]
ASSR enabling only considers capability declared in DPCD.
We also need to check whether the connector is internal.

[How]
ASSR enabling need to check both DPCD capability and internal display
flag.

Signed-off-by: Stylon Wang <stylon.wang@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix static checker warnings on tracebuff_fb

[Why]
Static analysis on linux-next has found a potential null pointer
dereference;

[How]
Refactor the function, add ASSERT and remove the unnecessary check.

Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add refresh rate trace

When we have to debug VRR issues, we usually want to know the current
refresh rate; for this reason, it is handy to have a way to check in
real-time the refresh rate value. This commit introduces a kernel trace
that can provide such information.

Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: BIOS LTTPR Caps Interface

[WHY]
Some platforms will have LTTPR capabilities forced on by VBIOS flags;
the functions added here will access those flags.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Dan Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: skip PP_MP1_STATE_UNLOAD on aldebaran

This message is not needed on Aldebaran.

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

drm/amd/amdgpu: set MP1 state to UNLOAD before reload its FW for vega20/ALDEBARAN

When resume from gpu reset, need set MP1 state to UNLOAD before reload SMU
FW otherwise will cause following errors:
[  121.642772] [drm] reserve 0x400000 from 0x87fec00000 for PSP TMR [  123.801051] [drm] failed to load ucode id (24) [  123.801055] [drm] psp command (0x6) failed and response status is (0x0) [  123.801214] [drm:psp_load_smu_fw [amdgpu]] *ERROR* PSP load smu failed!
[  123.801398] [drm:psp_resume [amdgpu]] *ERROR* PSP resume failed [  123.801536] [drm:amdgpu_device_fw_loading [amdgpu]] *ERROR* resume of IP block <psp> failed -22 [  123.801632] amdgpu 0000:04:00.0: amdgpu: GPU reset(9) failed [  123.801691] amdgpu 0000:07:00.0: amdgpu: GPU reset(9) failed [  123.802899] amdgpu 0000:04:00.0: amdgpu: GPU reset end with ret = -22

v2: add error info and including ALDEBARAN also

Signed-off-by: Chengming Gui <Jack.Gui@amd.com>
Reviewed-and-tested-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Reset error code for 'no handler' case

If reset handler is not implemented, reset error before proceeding.

Fixes issue with the following trace -
[  106.508592] amdgpu 0000:b1:00.0: amdgpu: ASIC reset failed with error, -38 for drm dev, 0000:b1:00.0
[  106.508972] amdgpu 0000:b1:00.0: amdgpu: GPU reset succeeded, trying to resume
[  106.509116] [drm] PCIE GART of 512M enabled.
[  106.509120] [drm] PTB located at 0x0000008000000000
[  106.509136] [drm] VRAM is lost due to GPU reset!
[  106.509332] [drm] PSP is resuming...

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-and-tested-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix black screen with scaled modes on some eDP panels

[why]
This was a regression introduced by commit:

        drm/amd/display: Skip modeset for front porch change

Due to the change how timing parameters were set, scaled modes would
cause a black screen on some eDP panels. Would probably apply to
other displays (i.e. even non-eDP) that only have scaled modes,
but such case is not that usual for external displays.

[how]
Pick up crtc frame dimensions when programming the timing unless
it's FreeSync video mode.

Fixes: 6f59f229f8ed7a ("drm/amd/display: Skip modeset for front porch change")
Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: ih reroute for newer asics than vega20

Starting Arcturus, it supports ih reroute through mmio directly
in bare metal environment. This is also valid for newer asics
such as Aldebaran.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: dqm fence memory corruption

Amdgpu driver uses 4-byte data type as DQM fence memory,
and transmits GPU address of fence memory to microcode
through query status PM4 message. However, query status
PM4 message definition and microcode processing are all
processed according to 8 bytes. Fence memory only allocates
4 bytes of memory, but microcode does write 8 bytes of memory,
so there is a memory corruption.

Changes since v1:
  * Change dqm->fence_addr as a u64 pointer to fix this issue,
also fix up query_status and amdkfd_fence_wait_timeout function
uses 64 bit fence value to make them consistent.

Signed-off-by: Qu Huang <jinsdb@126.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix offset calculation in amdgpu_vm_bo_clear_mappings()

Offset calculation wasn't correct as start addresses are in pfn
not in bytes.

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

drm/amd/amdgpu: Add CP_IB1_BASE_* to gc_10_3_0 headers

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Fix DPM level count on aldebaran

Firmware returns zero-based max level, increment by one to get
total levels. This fixes the issue of not showing all levels and current
frequency when frequency is at max DPM level.

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

drm/amd/pm: unify the interface for gfx state setting

No need to have special handling for swSMU supported ASICs.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: unify the interface for power gating

No need to have special handling for swSMU supported ASICs.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: fix missing static declarations

Add "static" declarations for those APIs used internally.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: unify the interface for loading SMU microcode

No need to have special handling for swSMU supported ASICs.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: no need to force MCLK to highest when no display connected

Correct the check for vblank short.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix build warnings

Fix header guard and make internal functions static. Fixes the below warnings:

drivers/gpu/drm/amd/amdgpu/../amdgpu/amdgpu_reset.h:24:9: warning: '__AMDUGPU_RESET_H__' is used as a header guard here, followed by #define of a different macro [-Wheader-guard]
drivers/gpu/drm/amd/amdgpu/aldebaran.c:110:6: warning: no previous prototype for function 'aldebaran_async_reset' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu13/aldebaran_ppt.c:1435:5: warning: no previous prototype for function 'aldebaran_mode2_reset' [-Wmissing-prototypes]

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Enable recovery on aldebaran

Add aldebaran to devices which support recovery

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 mode2 reset support for aldebaran

v1: Aldebaran uses reset control to support mode2 reset. The sequences to
reset and restore hardware context are specific to a particular
configuration.

v2: Clear bus mastering before reset.
Fix coding style issues, drop unwanted variables and info log.

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

drm/amdgpu: Make set PG/CG state functions public

Expose PG/CG set states functions for other clients

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

drm/amdgpu: Add PSP public function to load a list of FWs

v1: Adds a function to load a list of FWs as passed by the caller. This is
needed as only a select need to loaded for some use cases.

v2: Omit unrelated change, remove info log, fix return value when count is 0

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

drm/amdgpu: Add reset control handling to reset workflow

This prefers reset control based handling if it's implemented
for a particular ASIC. If not, it takes the legacy path. It uses
the legacy method of preparing environment (job, scheduler tasks)
and restoring environment.

v2: remove unused variable (Alex)

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

drm/amdgpu: Add reset control to amdgpu_device

v1: Add generic amdgpu_reset_control to handle different types of resets. It
may be added at device, hive or ip level. Each reset control has a list
of handlers associated with it to handle different types of reset. Reset
control is responsible for choosing the right handler given a particular
reset context.

Handler objects may implement a set of functions on how to handle a
particular type of reset.

prepare_env = Prepare environment/software context (not used currently).
prepare_hwcontext = Prepare hardware context for the reset.
perform_reset = Perform the type of reset.
restore_hwcontext = Restore the hw context after reset.
restore_env = Restore the environment after reset (not used currently).

Reset context carries the context of reset, as of now this is based on
the parameters used for current set of resets.

v2: Fix coding style

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

drm/amd/pm: Add support for reset completion on aldebaran

v1: On aldebaran, after hardware context restore, another handshake
needs to happen with PMFW so that reset recovery is complete from
PMFW side. Treat this as RESET_COMPLETE event for aldebaran.

v2: Cleanup coding style, info logs

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

drm/amd/pm: Add function to wait for smu events

v1: Add function to wait for specific event/states from PMFW

v2: Add mutex lock, simplify sequence

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

drm/amd/pm: Modify mode2 msg sequence on aldebaran

v1: During mode2 reset, PCI space is lost after message is sent.
Restore PCI space before waiting for response from firmware.

v2: Move mode2 sequence to aldebaran and update PMFW version.
Handle generic sequence in smu13 without PMFW version check.

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

drm/amd/amdgpu implement tdr advanced mode

[Why]
Previous tdr design treats the first job in job_timeout as the bad job.
But sometimes a later bad compute job can block a good gfx job and
cause an unexpected gfx job timeout because gfx and compute ring share
internal GC HW mutually.

[How]
This patch implements an advanced tdr mode.It involves an additinal
synchronous pre-resubmit step(Step0 Resubmit) before normal resubmit
step in order to find the real bad job.

1. At Step0 Resubmit stage, it synchronously submits and pends for the
first job being signaled. If it gets timeout, we identify it as guilty
and do hw reset. After that, we would do the normal resubmit step to
resubmit left jobs.

2. For whole gpu reset(vram lost), do resubmit as the old way.

v2: squash in build fix (Alex)

Signed-off-by: Jack Zhang <Jack.Zhang1@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: make BO type check less restrictive

BO with ttm_bo_type_sg type can also have tiling_flag and metadata.
So so BO type check for only ttm_bo_type_kernel.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reported-by: Tom StDenis <Tom.StDenis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use amdgpu_bo_user bo for metadata and tiling flag

Tiling flag and metadata are only needed for BOs created by
amdgpu_gem_object_create(), so we can remove those from the
base class.

v2: * squash tiling_flags and metadata relared patches into one
    * use BUG_ON for non ttm_bo_type_device type when accessing
    tiling_flags and metadata._
v3: *include to_amdgpu_bo_user

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

drm/amdgpu: use amdgpu_bo_create_user() for when possible

Use amdgpu_bo_create_user() for all the BO allocations for
ttm_bo_type_device type.

v2: include amdgpu_amdkfd_alloc_gws() as well it calls amdgpu_bo_create()
    for  ttm_bo_type_device

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

drm/amdgpu: introduce struct amdgpu_bo_user

Implement a new struct amdgpu_bo_user as subclass of
struct amdgpu_bo and a function to created amdgpu_bo_user
bo with a flag to identify the owner.

v2: amdgpu_bo_to_amdgpu_bo_user -> to_amdgpu_bo_user()

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

drm/amdgpu: allow variable BO struct creation

Allow allocating BO structures with different structure size
than struct amdgpu_bo.

v2: Check bo_ptr_size in all amdgpu_bo_create() caller.

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

drm/amdgpu: load balance VCN3 decode as well v8

Add VCN3 IB parsing to figure out to which instance we can send the
stream for decode.

v2: remove VCN instance limit as well, fix amdgpu_cs_find_mapping,
    check supported formats instead of unsupported.
v3: fix typo and error handling
v4: make sure the message BO is CPU accessible
v5: fix addr calculation once more
v6: only check message buffers
v7: fix constant and use defines
v8: fix create msg calculation

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

drm/amdgpu: share scheduler score on VCN3 instances

The VCN3 instances can do both decode as well as encode.

Share the scheduler load balancing score and remove fixing encode to
only the second instance.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add the sched_score to amdgpu_ring_init

Allow separate ring to share the same scheduler score.

No functional change.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: check fb of primary plane

Sometimes the primary plane might not be initialized (yet), which
causes dm_check_crtc_cursor to divide by zero.
Apparently a weird state before a S3-suspend causes the aforementioned
divide-by-zero error when resuming from S3.  This was explained in
bug 212293 on Bugzilla.

To avoid this divide-by-zero error we check if the primary plane's fb
isn't NULL.  If it's NULL the src_w and src_h attributes will be 0,
which would cause a divide-by-zero.

This fixes Bugzilla report 212293
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=212293

Fixes: 12f4849a1cfd69f3 ("drm/amd/display: check cursor scaling")
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Sefa Eyeoglu <contact@scrumplex.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Allow idle optimization based on vblank.

[Why]
idle optimization was being disabled after commit.

[How]
check vblank count for display off and enable idle optimization based on this count.
Also,check added to ensure vblank count does not decrement, when count reaches 0.

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

drm/amd: Fix a typo in two different sentences

s/defintion/definition/ .....two different places.

Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu/gfx_v7_0: Trivial typo fixes

s/acccess/access/
s/inferface/interface/
s/sequnce/sequence/  .....two different places.
s/retrive/retrieve/
s/sheduling/scheduling/
s/independant/independent/
s/wether/whether/ ......two different places.
s/emmit/emit/
s/synce/sync/

Reviewed-by: Nirmoy Das<nirmoy.das@amd.com>
Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon/r600_cs: Few typo fixes

s/miror/mirror/
s/needind/needing/
s/informations/information/

Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: securedisplay: simplify i2c hexdump output

A previous fix I did left a rather complicated loop in
amdgpu_securedisplay_debugfs_write() for what could be expressed in a
simple sprintf, as Rasmus pointed out.

This drops the leading 0x for each byte, but is otherwise
much nicer.

Suggested-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Ensure that the modifier requested is supported by plane.

On initializing the framebuffer, call drm_any_plane_has_format to do a
check if the modifier is supported. drm_any_plane_has_format calls
dm_plane_format_mod_supported which is extended to validate that the
modifier is on the list of the plane's supported modifiers.

The bug was caught using igt-gpu-tools test: kms_addfb_basic.addfb25-bad-modifier

Tested on ChromeOS Zork by turning on the display, running an overlay
test, and running a YT video.

=== Changes from v1 ===
Explicitly handle DRM_FORMAT_MOD_INVALID modifier.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Mark Yacoub <markyacoub@chromium.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Convert sysfs sprintf/snprintf family to sysfs_emit

Fix the following coccicheck warning:
drivers/gpu/drm/amd/pm/amdgpu_pm.c:1940:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:1978:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2022:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:294:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:154:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:496:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:512:9-17: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:1740:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:1667:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2074:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2047:9-17: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2768:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2738:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2442:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3246:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3253:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2458:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3047:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3133:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3209:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3216:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2410:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2496:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2470:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2426:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2965:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:2972:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3006:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu/drm/amd/pm/amdgpu_pm.c:3013:8-16: WARNING:
use scnprintf or sprintf

Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Convert sysfs sprintf/snprintf family to sysfs_emit

Fix the following coccicheck warning:
drivers/gpu//drm/amd/amdgpu/amdgpu_ras.c:434:9-17: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_xgmi.c:220:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_xgmi.c:249:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/df_v3_6.c:208:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_psp.c:2973:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:75:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:112:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:58:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:93:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:125:9-17: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_gtt_mgr.c:52:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_gtt_mgr.c:71:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:140:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:164:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:186:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:208:8-16: WARNING:
use scnprintf or sprintf
drivers/gpu//drm/amd/amdgpu/amdgpu_atombios.c:1916:8-16: WARNING:
use scnprintf or sprintf

Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon/radeon_pm: Convert sysfs sprintf/snprintf family to sysfs_emit

Fix the following coccicheck warning:
drivers/gpu//drm/radeon/radeon_pm.c:521:9-17: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:475:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:418:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:363:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:734:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:688:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:704:8-16: WARNING: use scnprintf or
sprintf
drivers/gpu//drm/radeon/radeon_pm.c:755:8-16: WARNING: use scnprintf or
sprintf

Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/pm: bail on sysfs/debugfs queries during platform suspend

The GPU is in the process of being shutdown.  Spurious queries during
suspend and resume can put the SMU into a bad state.  Runtime PM is
handled dynamically so we check if we are in non-runtime suspend.

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

drm/amdgpu/pm: mark pcie link/speed arrays as const

They are read only.

Noticed-by: Dave Airlie <airlied@linux.ie>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove irq_src->data handling

That is unused for quite some time now.

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>

drm/amd/display: Removing unused code from dmub_cmd.h

Removing code that is not used at the moment.

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

drm/amdgpu: Fix check for RAS support

Use positive logic to check for RAS
support. Rename the function to actually indicate
what it is testing for. Essentially, make the
function a predicate with the correct name.

Cc: Stanley Yang <Stanley.Yang@amd.com>
Cc: Alexander Deucher <Alexander.Deucher@amd.com>
Signed-off-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use appropriate DRM_DEBUG_... level

Convert IRQ-based prints from DRM_DEBUG_DRIVER to
the appropriate DRM log type, since IRQ-based
prints drown out the rest of the driver's
DRM_DEBUG_DRIVER messages.

v2: Update as per feedback to fine-tune for each
type of DRM log level.

Cc: Harry Wentland <Harry.Wentland@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Set amdgpu.noretry=1 for Arcturus

Setting amdgpu.noretry=1 as default for Arcturus.

Signed-off-by: Philip Cox <Philip.Cox@amd.com>
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: added support for dynamic GECC

updated host to send boot config to psp to enable GECC for sienna cichlid

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

drm/amdgpu: update host to psp interface

added interface support for setting boot config

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

drm/amd/pm: Update aldebaran pmfw interface

Update aldebaran PMFW interfaces to version 0x6

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

drm/amdgpu: move vram recover into sriov full access

[what]
currently driver recover vram after full access, which may hit
a corner case that meanwhile another whole gpu reset may be
triggered by another VF, which will cause vram recover fail
then fail the whole device reset.

[how]
move the recover vram into full access. So another bad VF will
not disturb the recover sequence for this vf.

Signed-off-by: Horace Chen <horace.chen@amd.com>
Reviewed by: Monk.Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: drop redundant and unneeded BACO APIs V2

Use other APIs which are with the same functionality but much
more clean.

V2: drop mediate unneeded interface

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: label these APIs used internally as static

Also drop unnecessary header file and declarations.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: make DAL communicate with SMU through unified interfaces

No need to have special handlings for swSMU supported ASICs.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: fix memory leak for dimgrey cavefish

We need to clean up the dcn3 clk_mgr.

Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: fix gcc -Wrestrict warning

gcc warns about an sprintf() that uses the same buffer as source
and destination, which is undefined behavior in C99:

drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c: In function 'amdgpu_securedisplay_debugfs_write':
drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c:141:6: error: 'sprintf' argument 3 overlaps destination object 'i2c_output' [-Werror=restrict]
  141 |      sprintf(i2c_output, "%s 0x%X", i2c_output,
      |      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  142 |       securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf[i]);
      |       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c:97:7: note: destination object referenced by 'restrict'-qualified argument 1 was declared here
   97 |  char i2c_output[256];
      |       ^~~~~~~~~~

Rewrite it to remember the current offset into the buffer instead.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: avoid incorrect %hu format string

clang points out that the %hu format string does not match the type
of the variables here:

drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:263:7: warning: format specifies type 'unsigned short' but the argument has type 'unsigned int' [-Wformat]
                                  version_major, version_minor);
                                  ^~~~~~~~~~~~~
include/drm/drm_print.h:498:19: note: expanded from macro 'DRM_ERROR'
        __drm_err(fmt, ##__VA_ARGS__)
                  ~~~    ^~~~~~~~~~~

Change it to a regular %u, the same way a previous patch did for
another instance of the same warning.

Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Rix <trix@redhat.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers: gpu: Remove duplicate include of amdgpu_hdp.h

amdgpu_hdp.h has been included at line 91, so remove
the duplicate include.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Wan Jiabing <wanjiabing@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix cat debugfs hang_hws file causes system crash bug

Here is the system crash log:
[ 1272.884438] BUG: unable to handle kernel NULL pointer dereference at
(null)
[ 1272.884444] IP: [<          (null)>]           (null)
[ 1272.884447] PGD 825b09067 PUD 8267c8067 PMD 0
[ 1272.884452] Oops: 0010 [#1] SMP
[ 1272.884509] CPU: 13 PID: 3485 Comm: cat Kdump: loaded Tainted: G
[ 1272.884515] task: ffff9a38dbd4d140 ti: ffff9a37cd3b8000 task.ti:
ffff9a37cd3b8000
[ 1272.884517] RIP: 0010:[<0000000000000000>]  [<          (null)>]
(null)
[ 1272.884520] RSP: 0018:ffff9a37cd3bbe68  EFLAGS: 00010203
[ 1272.884522] RAX: 0000000000000000 RBX: 0000000000000000 RCX:
0000000000014d5f
[ 1272.884524] RDX: fffffffffffffff4 RSI: 0000000000000001 RDI:
ffff9a38aca4d200
[ 1272.884526] RBP: ffff9a37cd3bbed0 R08: ffff9a38dcd5f1a0 R09:
ffff9a31ffc07300
[ 1272.884527] R10: ffff9a31ffc07300 R11: ffffffffaddd5e9d R12:
ffff9a38b4e0fb00
[ 1272.884529] R13: 0000000000000001 R14: ffff9a37cd3bbf18 R15:
ffff9a38aca4d200
[ 1272.884532] FS:  00007feccaa67740(0000) GS:ffff9a38dcd40000(0000)
knlGS:0000000000000000
[ 1272.884534] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1272.884536] CR2: 0000000000000000 CR3: 00000008267c0000 CR4:
00000000003407e0
[ 1272.884537] Call Trace:
[ 1272.884544]  [<ffffffffade68940>] ? seq_read+0x130/0x440
[ 1272.884548]  [<ffffffffade40f8f>] vfs_read+0x9f/0x170
[ 1272.884552]  [<ffffffffade41e4f>] SyS_read+0x7f/0xf0
[ 1272.884557]  [<ffffffffae374ddb>] system_call_fastpath+0x22/0x27
[ 1272.884558] Code:  Bad RIP value.
[ 1272.884562] RIP  [<          (null)>]           (null)
[ 1272.884564]  RSP <ffff9a37cd3bbe68>
[ 1272.884566] CR2: 0000000000000000

Signed-off-by: Qu Huang <jinsdb@126.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/atomic: Couple of typo fixes

s/seralization/serialization/
s/parallism/parallelism/

Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: don't evict if not initialized

TTM_PL_VRAM may not initialized at all when calling
radeon_bo_evict_vram(). We need to check before doing eviction.

[    2.160837] BUG: kernel NULL pointer dereference, address: 0000000000000020
[    2.161212] #PF: supervisor read access in kernel mode
[    2.161490] #PF: error_code(0x0000) - not-present page
[    2.161767] PGD 0 P4D 0
[    2.163088] RIP: 0010:ttm_resource_manager_evict_all+0x70/0x1c0 [ttm]
[    2.168506] Call Trace:
[    2.168641]  radeon_bo_evict_vram+0x1c/0x20 [radeon]
[    2.168936]  radeon_device_fini+0x28/0xf9 [radeon]
[    2.169224]  radeon_driver_unload_kms+0x44/0xa0 [radeon]
[    2.169534]  radeon_driver_load_kms+0x174/0x210 [radeon]
[    2.169843]  drm_dev_register+0xd9/0x1c0 [drm]
[    2.170104]  radeon_pci_probe+0x117/0x1a0 [radeon]

Reviewed-by: Christian König <christian.koenig@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: replace per_device_list by array

Remove per_device_list from kfd_process and replace it with a
kfd_process_device pointers array of MAX_GPU_INSTANCES size. This helps
to manage the kfd_process_devices binded to a specific kfd_process.
Also, functions used by kfd_chardev to iterate over the list were
removed, since they are not valid anymore. Instead, it was replaced by a
local loop iterating the array.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.128

This version brings along following fixes:

- Populate socclk entries for dcn2.1
- hide VGH asic specific structs
- Add kernel doc to crc_rd_wrk field
- revert max lb lines change
- Log DMCUB trace buffer events
- Fix debugfs link_settings entry
- revert max lb use by default for n10
- Deallocate IRQ handlers on amdgpu_dm_irq_fini
- Fixed Clock Recovery Sequence
- Fix UBSAN: shift-out-of-bounds warning
- [FW Promotion] Release 0.0.57
- Change input parameter for set_drr
- Use pwrseq instance to determine eDP instance

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use pwrseq instance to determine eDP instance

[Why & How]
Link index doesn't always correspond to the appropriate eDP instance.
We can assume lower link index is a lower eDP instance and set panel
control instance accordingly.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Jake Wang <haonan.wang2@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change input parameter for set_drr

[Why]
Change set_drr to pass in the entire dc_crtc_timing_adjust
structure instead of passing in the parameters individually.
This is to more easily pass in required parameters in the
adjust structure when it gets updated.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.57

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix UBSAN: shift-out-of-bounds warning

[Why]
On NAVI14 CONFIG_UBSAN reported shift-out-of-bounds at
display_rq_dlg_calc_20v2.c:304:38

rq_param->misc.rq_c.blk256_height is 0 when chroma(*_c) is invalid.
dml_log2 returns -1023 for log2(0), although log2(0) is undefined.

Which ended up as:
rq_param->dlg.rq_c.swath_height = 1 << -1023

[How]
Fix applied on all dml versions.
1. Ensure dml_log2 is only called if the argument is greater than 0.
2. Subtract req128_l/req128_c from log2_swath_height_l/log2_swath_height_c
   only when it is greater than 0.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Anson Jacob <Anson.Jacob@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fixed Clock Recovery Sequence

[Why]
When performing clock recovery, if a pre-emphasis adjustment is
requested, but voltage swing remains constant, the the retry counter
will not be reset. This can lead to prematurely failing link training.

[How]
Reset the clock recovery retry counter if an adjustment is requested
for either voltage swing or pre-emphasis.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Calvin Hou <Calvin.Hou@amd.com>
Signed-off-by: David Galiffi <David.Galiffi@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Deallocate IRQ handlers on amdgpu_dm_irq_fini

[why]
The amdgpu_dm IRQ handlers are not freed during the IRQ teardown.

[how]
Add function to deallocate IRQ handlers on amdgpu_dm_irq_fini step.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: revert max lb use by default for n10

This is causing a pstate change underflow regression for
unknown reason

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix debugfs link_settings entry

1. Catch invalid link_rate and link_count settings
2. Call dc interface to overwrite preferred link settings, and wait
until next stream update to apply the new settings.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Log DMCUB trace buffer events

[Why]
We want to log DMCUB trace buffer events as Linux kernel traces.

[How]
Register an IRQ handler for DMCUB outbox0 interrupt in amdgpu_dm,
and log the messages in the DMCUB tracebuffer to a new DMCUB
TRACE_EVENT as soon as we receive the outbox0 IRQ from DMCUB FW.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: revert max lb lines change

Some hardware revisions do have a max number of lines limitation
not honouring which can cause pstate switch underflow.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add kernel doc to crc_rd_wrk field

[Why]
Receive warning message below:

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h:380: warning: Function
parameter or member 'crc_rd_wrk' not described in 'amdgpu_display_manager'

[How]
Add documentation for crc_rd_wrk.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Wayne Lin <Wayne.Lin@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: hide VGH asic specific structs

The pmfw structs are specific to the asic and should not be
present in base clk_mgr struct

v2: squash in SI fix (Alex)

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Populate socclk entries for dcn2.1

[Why]
Dcn2.1 socclk entries in bandwidth params are not initialized.
They are not used now, but will be needed for dml validation.

[How]
Populate socclk bw params from dpm clock table

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Roman Li <roman.li@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Solomon Chiu <solomon.chiu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Use correct size when access vram

To make size is 4 byte aligned. Use &~0x3ULL instead of &3ULL.

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

drm/amd/pm: fix gpu reset failure by MP1 state setting

Instead of blocking varied unsupported MP1 state in upper level,
defer and skip such MP1 state handling in specific ASIC.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: fix MP1 state setting failure in s3 test

Skip PP_MP1_STATE_NONE in MP1 state setting, otherwise, it will
break S3 sequence.

[   50.188269] [drm:amdgpu_device_ip_suspend_phase2 [amdgpu]] *ERROR* SMC failed to set mp1 state 0, -22
[   50.969901] amdgpu 0000:03:00.0: amdgpu: SMU is resuming...
[   50.970024] sd 0:0:0:0: [sda] Starting disk
[   50.979723] serial 00:02: activated
[   51.353644] ata4: SATA link down (SStatus 4 SControl 300)
[   51.353669] ata3: SATA link down (SStatus 4 SControl 300)
[   51.353747] ata6: SATA link down (SStatus 4 SControl 300)
[   51.357694] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[   51.357711] ata5: SATA link down (SStatus 4 SControl 300)
[   51.357729] ata2: SATA link down (SStatus 4 SControl 300)
[   51.358005] ata1.00: supports DRM functions and may not be fully accessible
[   51.360491] ata1.00: supports DRM functions and may not be fully accessible
[   51.362573] ata1.00: configured for UDMA/133
[   51.362610] ahci 0000:00:17.0: port does not support device sleep
[   51.362946] ata1.00: Enabling discard_zeroes_data
[   52.566438] amdgpu 0000:03:00.0: amdgpu: Msg issuing pre-check failed and SMU may be not in the right state!
[   54.126316] amdgpu 0000:03:00.0: amdgpu: Msg issuing pre-check failed and SMU may be not in the right state!
[   54.126317] amdgpu 0000:03:00.0: amdgpu: Failed to SetDriverDramAddr!
[   54.126318] amdgpu 0000:03:00.0: amdgpu: Failed to setup smc hw!
[   54.126319] [drm:amdgpu_device_ip_resume_phase2 [amdgpu]] *ERROR* resume of IP block <smu> failed -62
[   54.126398] amdgpu 0000:03:00.0: amdgpu: amdgpu_device_ip_resume failed (-62).
[   54.126399] PM: dpm_run_callback(): pci_pm_resume+0x0/0x90 returns -62
[   54.126403] PM: Device 0000:03:00.0 failed to resume async: error -62

Fixes: 1689fca0d62aa7 ("drm/amd/pm: fix Navi1x runtime resume failure V2")
Signed-off-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/powerplay/smu10: refactor AMDGPU_PP_SENSOR_GPU_LOAD

refactor AMDGPU_PP_SENSOR_GPU_LOAD to ensure code consistency with other
commands

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix amdgpu_res_first()

Fix size comparison in the resource cursor.

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

drm/amdgpu/display: restore AUX_DPHY_TX_CONTROL for DCN2.x

Commit 098214999c8f added fetching of the AUX_DPHY register
values from the vbios, but it also changed the default values
in the case when there are no values in the vbios.  This causes
problems with displays with high refresh rates.  To fix this,
switch back to the original default value for AUX_DPHY_TX_CONTROL.

Fixes: 098214999c8f ("drm/amd/display: Read VBIOS Golden Settings Tbl")
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1426
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Igor Kravchenko <Igor.Kravchenko@amd.com>
Cc: Aric Cyr <Aric.Cyr@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>

drm/amd/display: use GFP_ATOMIC in dcn20_resource_construct

Replace GFP_KERNEL with GFP_ATOMIC as dcn20_resource_construct()
can't sleep.

Partially fixes: https://bugzilla.kernel.org/show_bug.cgi?id=212311
as dcn20_resource_construct() also calls into SMU functions which does
mutex_lock().

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display/dc/calcs/dce_calcs: Remove some large variables from the stack

Fixes the following W=1 kernel build warning(s):

 drivers/gpu/drm/amd/amdgpu/../display/dc/calcs/dce_calcs.c: In function ‘bw_calcs_init’:
 drivers/gpu/drm/amd/amdgpu/../display/dc/calcs/dce_calcs.c:2726:1: warning: the frame size of 1336 bytes is larger than 1024 bytes [-Wframe-larger-than=]

v2: squash in sizeof fix

Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Leo Li <sunpeng.li@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>