drm/amdgpu/sdma5: use common function to init sdma fw

Use common function to init sdma v5 firmware ucode.

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

drm/amdgpu/sdma4: use common function to init sdma fw

Use common function to init sdma v4 firmware ucode.

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

drm/amdgpu: add function to init SDMA microcode

Add an common function to init SDMA related microcode.

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

drm/amdgpu/gfx11: use common function to init cp fw

Use common function to init gfx v11 CP firmware ucode.

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

drm/amdgpu/gfx10: use common function to init CP fw

Use common function to init gfx v10 CP firmware ucode.

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

drm/amdgpu/gfx9: use common function to init cp fw

Use common function to init gfx v9 CP firmware ucode.

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

drm/amdgpu: add function to init CP microcode

Add an common function to init CP related microcode.

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

drm/amd/pm: use adverse selection for dpm features unsupported by driver

It's vbios and pmfw instead of driver who decide whether some dpm features
is supported or not. Driver just de-selects those features which are not
permitted on user's request. Thus, we use adverse selects model.

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: enable gfxoff feature for SMU 13.0.0

The feature is ready with latest 78.58.0 PMFW.

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/amdgpu: avoid gfx register accessing during gfxoff

Make sure gfxoff is disabled before gfx register accessing.

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>

drivers/amd/pm: check the return value of amdgpu_bo_kmap

amdgpu_bo_kmap() returns error when fails to map buffer object. Add the
error check and propagate the error.

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

drm/amdgpu: Use simplified API for p2p dist calc

Use the simpified API that calculates distance between two devices.

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

drm/amdgpu: Disable verbose for p2p dist calc

Disable verbose while getting p2p distance. With verbose, it shows
warning if ACS redirect is set between the devices. Adds noise
to dmesg logs when a few GPU devices are on the same platform.

Example log:

amdgpu 0000:34:00.0: ACS redirect is set between the client and provider (0000:31:00.0)
amdgpu 0000:34:00.0: to disable ACS redirect for this path, add the kernel parameter:
pci=disable_acs_redir=0000:30:00.0;0000:2e:00.0;0000:33:00.0;0000:2e:10.0

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

drm/amdgpu: Fixed ras warning when uninstalling amdgpu

  For the asic using smu v13_0_2, there is the following
warning when uninstalling amdgpu:
  amdgpu: ras disable gfx failed poison:1 ret:-22.

[Why]:
  For the asic using smu v13_0_2, the psp .suspend and
  mode1reset is called before amdgpu_ras_pre_fini during
  amdgpu uninstall, it has disabled all ras features and
  reset the psp. Since the psp is reset, calling
  amdgpu_ras_disable_all_features in amdgpu_ras_pre_fini
  to disable ras features will fail.

[How]:
  If all ras features are disabled, amdgpu_ras_disable_all_features
  will not be called to disable all ras features again.

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

drm/amdgpu/gfx11: switch to amdgpu_gfx_rlc_init_microcode

switch to common helper to initialize rlc firmware
for gfx11

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

drm/amdgpu/gfx10: switch to amdgpu_gfx_rlc_init_microcode

switch to common helper to initialize rlc firmware
for gfx10

v2: squash in size validation fix (Alex)

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

drm/amdgpu/gfx9: switch to amdgpu_gfx_rlc_init_microcode

switch to common helper to initialize rlc firmware
for gfx9

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

drm/amdgpu: add helper to init rlc firmware

To initialzie rlc firmware according to rlc
firmware header version

v2: squash in backwards compat fix

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

drm/amdgpu: add helper to init rlc fw in header v2_4

To initialize rlc firmware in header v2_4

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

drm/amdgpu: add helper to init rlc fw in header v2_3

To initialize rlc firmware in header v2_3

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

drm/amdgpu: add helper to init rlc fw in header v2_2

To initialize rlc firmware in header v2_2

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

drm/amdgpu: add helper to init rlc fw in header v2_1

To initialize rlc firmware in header v2_1

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

drm/amdgpu: add helper to init rlc fw in header v2_0

To initialize rlc firmware in header v2_0

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

drm/amdgpu: Fix amdgpu_vm_pt_free warning

Free page table BO from vm resv unlocked context generate below
warnings.

Add a pt_free_work in vm to free page table BO from vm->pt_freed list.
pass vm resv unlock status from page table update caller, and add vm_bo
entry to vm->pt_freed list and schedule the pt_free_work if calling with
vm resv unlocked.

WARNING: CPU: 12 PID: 3238 at
drivers/gpu/drm/ttm/ttm_bo.c:106 ttm_bo_set_bulk_move+0xa1/0xc0
Call Trace:
 amdgpu_vm_pt_free+0x42/0xd0 [amdgpu]
 amdgpu_vm_pt_free_dfs+0xb3/0xf0 [amdgpu]
 amdgpu_vm_ptes_update+0x52d/0x850 [amdgpu]
 amdgpu_vm_update_range+0x2a6/0x640 [amdgpu]
 svm_range_unmap_from_gpus+0x110/0x300 [amdgpu]
 svm_range_cpu_invalidate_pagetables+0x535/0x600 [amdgpu]
 __mmu_notifier_invalidate_range_start+0x1cd/0x230
 unmap_vmas+0x9d/0x140
 unmap_region+0xa8/0x110

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

drm/amdgpu: Use vm status_lock to protect pt free

Use vm_status_lock to protect all vm_status state transitions to allow
them to happen without a reservation lock in unlocked page table
updates.

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

drm/amdgpu: Use vm status_lock to protect vm evicted list

Use vm_status_lock to protect all vm_status state transitions to allow
them to happen without a reservation lock in unlocked page table
updates.

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

drm/amdgpu: Use vm status_lock to protect vm moved list

Use vm_status_lock to protect all vm_status state transitions to allow
them to happen without a reservation lock in unlocked page table
updates.

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

drm/amdgpu: Use vm status_lock to protect vm idle list

Use vm_status_lock to protect all vm_status state transitions to allow
them to happen without a reservation lock in unlocked page table
updates.

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

drm/amdgpu: Use vm status_lock to protect relocated list

Use vm_status_lock to protect all vm_status state transitions to allow
them to happen without a reservation lock in unlocked page table
updates.

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

drm/amdgpu: Rename vm invalidate lock to status_lock

The vm status_lock will be used to protect all vm status lists.

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

drm/amd/pm: Remove unneeded result variable

Return the value atomctrl_initialize_mc_reg_table_v2_2() directly instead
of storing it in another redundant variable.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Remove the unneeded result variable

Return the value append_vbios_pptable() directly instead of storing it in
another redundant variable.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix initial connector audio value

This got lost somewhere along the way, This fixes
audio not working until set_property was called.

Signed-off-by: hongao <hongao@uniontech.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Reduce number of arguments of dml314's CalculateFlipSchedule()

Most of the arguments are identical between the two call sites and they
can be accessed through the 'struct vba_vars_st' pointer. This reduces
the total amount of stack space that
dml314_ModeSupportAndSystemConfigurationFull() uses by 112 bytes with
LLVM 16 (1976 -> 1864), helping clear up the following clang warning:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn314/display_mode_vba_314.c:4020:6: error: stack frame size (2216) exceeds limit (2048) in 'dml314_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
  void dml314_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
       ^
  1 error generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/1710
Reported-by: "kernelci.org bot" <bot@kernelci.org>
Tested-by: Maíra Canal <mairacanal@riseup.net>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Reduce number of arguments of dml314's CalculateWatermarksAndDRAMSpeedChangeSupport()

Most of the arguments are identical between the two call sites and they
can be accessed through the 'struct vba_vars_st' pointer. This reduces
the total amount of stack space that
dml314_ModeSupportAndSystemConfigurationFull() uses by 240 bytes with
LLVM 16 (2216 -> 1976), helping clear up the following clang warning:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn314/display_mode_vba_314.c:4020:6: error: stack frame size (2216) exceeds limit (2048) in 'dml314_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
  void dml314_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
       ^
  1 error generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/1710
Reported-by: "kernelci.org bot" <bot@kernelci.org>
Tested-by: Maíra Canal <mairacanal@riseup.net>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: don't register a dirty callback for non-atomic

Some asics still support non-atomic code paths.

Fixes: 66f99628eb2440 ("drm/amdgpu: use dirty framebuffer helper")
Reported-by: Arthur Marsh <arthur.marsh@internode.on.net>
Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: bump minor for gang submit

Since that has now landed bump the minor to let userspace know about it.

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/amdgpu: properly initialize return value during CS

The return value is no longer initialized before the loop because of
moving code around.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: c2b08e7a6d27 ("drm/amdgpu: move entity selection and job init earlier during CS")
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: drop the pptable related workarounds for SMU 13.0.0

The pptable in the vbios is fully ready. The related workarounds
in driver are not needed any more.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-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/amd/pm: add support for 3794 pptable for SMU13.0.0

Enable 3794 pptable support for SMU13.0.0.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-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: add gang submit frontend v6

Allows submitting jobs as gang which needs to run on multiple engines at the
same time.

All members of the gang get the same implicit, explicit and VM dependencies. So
no gang member will start running until everything else is ready.

The last job is considered the gang leader (usually a submission to the GFX
ring) and used for signaling output dependencies.

Each job is remembered individually as user of a buffer object, so there is no
joining of work at the end.

v2: rebase and fix review comments from Andrey and Yogesh
v3: use READ instead of BOOKKEEP for now because of VM unmaps, set gang
    leader only when necessary
v4: fix order of pushing jobs and adding fences found by Trigger.
v5: fix job index calculation and adding IBs to jobs
v6: fix typo found by Alex

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/amdgpu: add gang submit backend v2

Allows submitting jobs as gang which needs to run on multiple
engines at the same time.

Basic idea is that we have a global gang submit fence representing when the
gang leader is finally pushed to run on the hardware last.

Jobs submitted as gang are never re-submitted in case of a GPU reset since this
won't work and will just deadlock the hardware immediately again.

v2: fix logic inversion, improve documentation, fix rcu

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/amdgpu: cleanup instance limit on VCN4 v4

Similar to what we did for VCN3 use the job instead of the parser
entity. Cleanup the coding style quite a bit as well.

v2: merge improved application check into this patch
v3: finally fix the check
v4: limit to the correct engine

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/amdgpu: getting fan speed pwm for vega10 properly

Instead of using RPM speed, we will use a function from vega20 based on PWM registers.

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

drm/amdgpu: revert "fix limiting AV1 to the first instance on VCN3" v3

This reverts commit 250195ff744f260c169f5427422b6f39c58cb883.

The job should now be initialized when we reach the parser functions.

v2: merge improved application check into this patch
v3: back to the original test, but use the right ring

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/amdgpu: move entity selection and job init earlier during CS

Initialize the entity for the CS and scheduler job much earlier.

v2: fix job initialisation order and use correct scheduler instance

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/amdgpu: cleanup error handling in amdgpu_cs_parser_bos

Return early on success and so remove all those "if (r)" in the error
path.

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/amdgpu: cleanup CS pass2 v6

Cleanup the coding style and function names to represent the data
they process for pass2 as well.

Go over the chunks only twice now instead of multiple times.

v2: fix job initialisation order and use correct scheduler instance
v3: try to move all functional changes into a separate patch.
v4: separate reordering, pass1 and pass2 change
v5: fix va_start calculation
v6: fix user fence check

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/pm: Remove redundant check condition

V3:
[Why]:
  Since ASICs using smu v13_0_2 have called smu
.suspend before calling amdgpu_pm_sysfs_fini,
pm.dpm_enabled was already set to 0 when smu
.suspend was called. This makes it impossible
to delete all pm sys nodes.

[How]:
  Since all functions in amdgpu_pm_sysfs_fini
have their own condition checks, so the first
conditional check in amdgpu_pm_sysfs_fini can
be removed.

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

drm/amdgpu: Fixed psp fence and memory issues when removing amdgpu device

V3:
Fixed psp fence and memory issues for the asic
using smu v13_0_2 when removing amdgpu device.

[Why]:
1. psp_suspend->psp_free_shared_bufs->
       psp_ta_free_shared_buf->
           amdgpu_bo_free_kernel->
             ...->amdgpu_bo_release_notify->
                    amdgpu_fill_buffer
   psp will free vram memory used by psp when psp_suspend
   is called. But for the asic using smu v13_0_2, because
   psp_suspend is called before adev->shutdown is set to
   true when removing the first hive device, amdgpu fill_buffer
   will be called, which will cause fence issues when evicting
   all vram resources in amdgpu vram mgr_fini.
2. Since psp_hw_fini is not called after calling psp_suspend
   and psp_suspend only calls psp_ring_stop, the psp ring memory
   will not be released when amdgpu device is removed.

[How]:
1. Set shutdown to true before calling amdgpu_device_gpu_recover,
   then amdgpu_fill_buffer will not be called when psp_suspend is
   called.
2. Free psp ring memory in psp_sw_fini.

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

drm/amdgpu: Adjust removal control flow for smu v13_0_2

Adjust removal control flow for smu v13_0_2:
   During amdgpu uninstallation, when removing the first
device, the kernel needs to first send a mode1reset message
to all gpu devices. Otherwise, smu initialization will fail
the next time amdgpu is installed.

V2:
1. Update commit comments.
2. Remove the global variable amdgpu_device_remove_cnt
   and add a variable to the structure amdgpu_hive_info.
3. Use hive to detect the first removed device instead of
   a global variable.

V3:
 1. Update commit comments.
 2. Split a patch into multiple patches.
 3. The current patch does:
    a. Add a work mode of AMDGPU_RESET_FOR_DEVICE_REMOVE into
       the existing gpu recover path, which make all devices
       in hive list only have HW reset but no resume (except
       the base IP).
    b. Call AMDGPU_RESET_FOR_DEVICE_REMOVE and
       AMDGPU_NEED_FULL_RESET mode of amdgpu_device_gpu_recover
       in amdgpu_pci_remove when removing the first device in
       hive list.
    c. When removing the first device, the IP blocks keyword
       function call sequence is as follows:
.suspend->mode1reset->.resume(basic ip)->.hw_fini->.early_fini->.sw_fini.
   ^                           |
   |-<----------<---------<----|
The first three sequences are because of a call to
        amdgpu_device_gpu_recover. The three sequences will be
        executed in a loop until all devices in the hive list
        are iterated.
        The sequences starting from .hw_fini only apply to the
        first device. Since .suspend has been called before,
        except the resumed phase1 basic ip blocks, all other ip
        blocks .hw_fini of current device will do nothing.
     d. When removing other devices, the calling sequences is the
        same as legacy:
   .hw_fini -> .early_fini -> .sw_fini.
Since .suspend has been called when removing the first device,
        except the resumed phase1 basic ip blocks, all of other ip
        blocks .hw_fini of current device will do nothing.

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

drm/amd/display: 3.2.204

This version brings along following fixes:
- Fix urgent latency override for DCN32/DCN321
- Correct hostvm flag in DCN31
- Added new Asic Id for DCN301
- Adjust to 2 phys in DCN301
- Update dummy P-state search to use DCN32 DML
- Increase dcn315 pstate change latency
- Disable OTG WA for the plane_state NULL case on DCN314
- Update dccg based on HW delta
- Correct num_dsc for DCN314
- Fix DP DSC compliance failure with DCN314
- Refactor edp power sequencer codes
- update gamut remap if plane has changed
- Skip audio setup when audio stream is enabled
- Display distortion after hotplug LG Ultrafine 5K tiled display
- Fix DP MST timeslot issue when fallback happened
- Uncomment SubVP pipe split assignment in driver
- Enable committing subvp config
- Do second pass through DML for DET calculation
- Don't allocate DET for phantom pipes
- Modify DML to adjust Vstartup Position
- Ignore k1/k2 values for virtual signal to avoid assert
- Fix failures of disabling primary planes
- Do not compare integers of different widths
- Fix empty slot problem of EDID management
- Revise Sink device string ID width in dmub psr
- Log vertical interrupt 1 for debug

Acked-by: Wayne Lin <wayne.lin@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: correct num_dsc based on HW cap

[why]
num_dsc is 3 for dcn314 based on HW capablity.

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: solve regression in update phy state refactor

[Why]
There is a coding error when moving dp disable link phy to
hw sequencer, where the receiver power control is missed during
this refactor.

[how]
1. Add back missing receiver power control in disable link phy.
2. minor modifications to ensure there is no undesired sequence
changes in dp link enable/disable.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: update dccg based on HW delta

[why]
update hw dccg based on HW delta, and reuse common src code

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Do second pass through DML for DET calculation

[Why & How]
If we find that DML requires pipe split, run through
DML again because the DET allocation per pipe must
be re-assigned.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: log vertical interrupt 1 for debug

[Why & How]
Extend existing OTG state collection function to include the vertical
interrupt 1 state.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Revise Sink device string ID

[Why]
The Sink device string ID1/ID2 use 5 bytes instead of 6 bytes,
so the driver should compare the first 5 bytes only.

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Robin Chen <po-tchen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Modify DML to adjust Vstartup Position

[WHY]
The Vstartup position should be as late as possible to
maximize power saving with the current. Calculation of
Vstartup in DML does not take into account as SDP signal.

[HOW]
Made necessary changes to calculate the correct Vstartup
position in DML to account for AS SDP
* Overriding the VBlankNom value in certain cases
  otherwise it will use the default value
* Bypassing the condition for adjust_sync_Vstartup
* Overriding vblank_nom_input with VBlankNom that is
  set in the SW Layer

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: muansari <muansari@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable OTG WA for the plane_state NULL case on DCN314

[Why]
This shouldn't trigger during tiled display hotplug/unplug but it does
because one of the tiles can end up with a NULL plane state.

This also doesn't guard against the hang that it was originally trying
to resolve, and can instead cause DIO corruption due to OTG sync
being lost.

[How]
This was reverted at one point out of DCN31 so revert it here too.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Add shift and mask for ICH_RESET_AT_END_OF_LINE

[Why]
DP DSC compliance failing for dcn314 due to ICH_RESET_AT_END_OF_LINE
shift and mask being missing

[How]
Add in shift and mask for ICH_RESET_AT_END_OF_LINE

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Enable committing subvp config

[Why and How]
Enable committing subvp config through DMCUB for DCN32

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Assume connectors are on single slot

[Why]
v1_5 display object table has no way for connectors to
indicate which slot they are a part of, resulting in additional
empty slots to appear in EDID management UI.

[How]
Assume that all connectors belong to the same slot.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: do not compare integers of different widths

[Why & How]
Increase width of some variables to avoid comparing integers of
different widths

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: increase dcn315 pstate change latency

[Why & How]
Update after new measurment came in

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Ignore k1/k2 values for virtual signal

[Why and How]
We are hitting k1/k2 assert when we are using a virtual signal in the
test; as a result, we are failing some automated tests with a false
positive. This commit addresses this issue by ignoring the assert
condition if we use SIGNAL_TYPE_VIRTUAL.

Reviewed-by: Aurabindo Pillai <Aurabindo.Pillai@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't allocate DET for phantom pipes

[Why & How]
We should not allocate any DET for the phantom pipes.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Fix DP MST timeslot issue when fallback happened

[Why]
When USB4 DP link training failed and fell back to lower link rate,
the time slot calculation uses the verified_link_cap.
And the verified_link_cap was not updated to the new one.
It caused the wrong VC payload time-slot was allocated.

[How]
Updated verified_link_cap with the new one from cur_link_settings
after the LT completes successfully.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Cruise Hung <Cruise.Hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Display distortion after hotplug 5K tiled display

[Why]
During hot plug of specific 5K tiled display, sometimes both the tiles
are not synchronized resulting in distortion. The reason is that otgs of
both the tiles goes out of sync when otg workaround (dcnxxx_disable_otg_wa)
is applied for bandwidth optimization. The otg workaround reenables otg
but otg synchronization context is not reset and hence dc_trigger_sync()
does not resynchronize otg again.

[How]
Implement reset_sync_context_for_pipe() to reset the otg synchronization
context for the disabled pipe and its slave pipes when otg workaround is
applied.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Meenakshikumar Somasundaram <meenakshikumar.somasundaram@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update dummy P-state search to use DCN32 DML

[Why]
Current DCN3.2 logic for finding the dummy P-state index uses the
DCN3.0 DML validation function instead of DCN3.2 DML.

This can result in either unexpected DML VBA values, or unexpected
dummy P-state index to be used.

[How]
Update the dummy P-state logic to use DCN3.2 DML validation function.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Expose few dchubbub functions

[why]
Expose few dchubbun functions in dcn31 and dcn32 to leverage.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Removing 2 phys

[WHY]
New dcn301 has 2 less phys

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Pavle Kotarac <Pavle.Kotarac@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Added new DCN301 Asic Id

[WHY]
Adding new asic id for dcn301

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Pavle Kotarac <Pavle.Kotarac@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: correct hostvm flag

[Why]
Hostvm should be enabled/disabled accordding to
the status of riommu_active, but hostvm always
be disabled on DCN31 which causes underflow

[How]
Set correct hostvm flag on DCN31

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Sherry Wang <Yao.Wang1@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix urgent latency override for DCN32/DCN321

[Why]
The urgent latency override is useful when debugging issues
relating to underflow.

Current overridden variable is not correct and has no effect
on DCN3.2 and DCN3.21 DML calculations.

[How]
For DCN3.2 and DCN3.21, override the correct urgent latency
variable when bounding box override is present.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Uncomment SubVP pipe split assignment in driver

[Why & How]
Uncomment SubVP pipe split assignment in driver since FW headers
are now promoted

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: skip audio setup when audio stream is enabled

[why]
We have minimal pipe split transition method to avoid pipe
allocation outage.However, this method will invoke audio setup
which cause audio output stuck once pipe reallocate.

[how]
skip audio setup for pipelines which audio stream has been enabled

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: zhikzhai <zhikai.zhai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: update gamut remap if plane has changed

[Why]
The desktop plane and full-screen game plane may have different
gamut remap coefficients, if switching between desktop and
full-screen game without updating the gamut remap will cause
incorrect color.

[How]
Update gamut remap if planes change.

Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Hugo Hu <hugo.hu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Refactor edp panel power sequencer(PPS) codes

[Why & How]
Move extra panel power sequencer settings into panel_cofig struct.

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Ian Chen <ian.chen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.203

This version brings along following fixes:
- Port DCN30 420 logic to DCN32
- Remove some unused definitions from DCN32/321
- Remove dp dig pixle rate div policy from dcn314
- Fix dcn315 reading of memory channel count and width

- Fix SubVP and ODM relevant issues
- Fix pipe split, MPO and ODM relevant issues
- Support proper mst payload removal when link is not in mst mode in dc
- Assume an LTTPR is always present on fixed_vs links
- Rework recent update PHY state commit
- Add debug option and logs

Acked-by: Wayne Lin <wayne.lin@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: [FW Promotion] Release 0.0.134.0

- Handle pipe split case for SubVP:
  Pass in pipe split index for main and phantom pipes

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix SubVP way calculation

[Why and How]
- Don't skip bottom and next odm pipe when calculating
  num ways for subvp
- Don't need to double cache lines for DCC (divide by 256)

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Disable SubVP on driver disable

[Why and How]
- For driver disable cases in current implementation, if P-State
  is unsupported or still supported by firmware, we force it
  supported by DCN.

- SubVP now needs to be included in this case along with FPO.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: SubVP pipe split case

[Why and How]
For SubVP pipe split case, pass in split index for
main and phantom pipes to ensure that the P-State
sequence will force P-State for all required pipes.

Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Add debug option for allocating extra way for cursor

[Why and How]
- Add a debug option for allocating extra way for cursor
- Remove usage of cache_cursor_addr since it's not gaurenteed
  to be populated
- Include cursor size in MALL calculation if it exceeds the
  DCN cursor buffer size (and don't need extra way for cursor)

Reviewed-by: Aurabindo Pillai <Aurabindo.Pillai@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: For ODM seamless transition require AUTO mode

[Why & How]
ODM seamless transitions require DIV_MODE_AUTO. However,
DIV_MODE_AUTO only works when all the horizontal timing params
are divisible by the ODM combine factor. Therefore, disable the
ODM 2:1 policy when the horizontal timing params are not divisible
by 2.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: support proper mst payload removal when link is not in mst mode in dc

[why]
When user unplugs mst hubs, the current code will forcefully zero
entire mst payload allocation table structure stored in link before we
deallocate actual payload when disabling stream.

During the first disable stream sequence, we will use current mst
payload allocation table to determine if link should be turned off.
Because we zero out it before we are disabling stream, the payload
allocation table stored in link doesn't represent the actual allocation status,
so we turn off link at the first disable stream without waiting until all
streams' payloads have been deallocated. This avoilates the designed
deallocation sequence and caused system hang in DP2 scenario.

[how]
Remove payload during deallocation and never zero payload allocation structure
without actually deallocating payload.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: rework recent update PHY state commit

[why]
Original change 594b237b9a07 ("drm/amd/display: Add
interface to track PHY state") was implemented by assuming stream's
dpms off is equivalent to PHY power off.
This assumption doesn't hold in following situations:
1. MST multiple stream scenario, where multiple streams are sharing the
same PHY output. Toggle dpms off for one of the stream doesn't power
off the PHY due to the presence of other streams.
2. enable stream failure scenario, where enable stream fails due to
failure of link training. This will cause DPMS off is set to false, while
the actual PHY power state is off in certain cases.
Due to the problematic assumption, the logic will skip disabling
other streams for MST multiple stream scenario, therefore PHY is
not actually powered off.

[how]
1. Rework this refactor by moving PHY state update down to hardware
level, where we update PHY state in place when hardware sequencer
is actually changing the power state of the PHY hardware.
2. Reimplement symclk on TX off workaround in place when we are actually
calling transmitter control to power off PHY in dcn32. Note the workaround is
added due to the lack of proper software interface to set TX while keeping
symclk on. We plan to address this interface problem so we can set TX off
only without affecting symclk in future dcn versions.

Fixes: 594b237b9a07 ("drm/amd/display: Add interface to track PHY state")
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add MES and MES-KIQ version in debugfs

This patch addes MES and MES-KIQ version in debugfs.

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

drm/amd/display: Assume an LTTPR is always present on fixed_vs links

[WHY]
LTTPRs can in very rare instsances fail to increment DPCD LTTPR count.
This results in aux-i LTTPR requests to be sent to the wrong DPCD
address, which causes link training failure.

[HOW]
Override internal repeater count if fixed_vs flag is set for a given link

Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Michael Strauss <michael.strauss@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dcn315 memory channel count and width read

[Why & How]
Correctly set ddr5 channel width to 8 bytes

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix double cursor on non-video RGB MPO

[Why]

DC makes use of layer_index (zpos) when picking the HW plane to enable
HW cursor on. However, some compositors will not attach zpos information
to each DRM plane. Consequently, in amdgpu, we default layer_index to 0
and do not update it.

This causes said DC logic to enable HW cursor on all planes of the same
layer_index, which manifests as a double cursor issue if one of the
planes is scaled (and hence scaling the cursor as well).

[How]

Use DRM core helpers to calculate a normalized_zpos value for each
drm_plane_state under each crtc, within the atomic state.

This helper will first consider existing zpos values, and if
identical/unset, fallback to plane ID ordering.

The normalized_zpos is then passed to dc_plane_info during atomic check
for later use by the cursor logic.

Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Only consider pixle rate div policy for DCN32+

[Why and How]
- Only consider pixel rate div policy for DCN32+

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Various logs added

[Why & How]
Added logs for panel delays, spread_spectrum_percentage,
and gpuclk_ss_percentage to facilitate debugging.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Leo Chen <sancchen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add debug option for dramclk_change_latency in apu

[Why & How]
Support dramclk change latency change via debug option and add some
code isolation.

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Fix pipe split prediction

[Why & How]
- Pipe split prediction previously only took into
account MPC split. We must also consider when
ODM combine is required, and when we apply ODM
combine by policy.

- Also re-work DET allocation function as it wasn't
properly splitting the DET per stream, per plane.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Wayne Lin <wayne.lin@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: Remove some unused definitions from DCN32/321

[Why&How]
After reg offset initialization was switched to runtime rather than
compile time, some of the defintions are not needed anymore and can
be removed.

Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Port DCN30 420 logic to DCN32

[Why]
420 modes are limited by FMT buffer width of 4096
which requires multi-pipe support in form of ODM
combine.  If 420 modes have greater HActive than
4096, the DML logic should accomodate whether
it should be rejected, or ODM combine 2:1 or 4:1
is triggered accordingly.

[How]
FMT Buffer limit of 4096 in DCN32. Force ODM
combine depending on HActive and FMT Buffer limit.
Reject modes if TMDS 420 and above 4096.

Acked-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Chris Park <chris.park@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: clean up some inconsistent indentings

clean up some inconsistent indentings

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=2182
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: clean up some inconsistent indentings

clean up some inconsistent indentings

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=2181
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>