drm/amdgpu: Remove unneeded variable 'ret' in amdgpu_device.c

Fixes coccicheck warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:1036:5-8: Unneeded variable: "ret". Return "0" on line 1079

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Ma Feng <mafeng.ma@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx: Add mmSDMA2-7_EDC_COUNTER to support Arcturus

Add mmSDMA2-7_EDC_COUNTER to support Arcturus

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx: Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7 to support Arcturus

Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7 to support Arcturus

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx: Replace ARRAY_SIZE with size variable

Replace ARRAY_SIZE with size variables to support
different ASICs.

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7 to support Arcturus

Arcturus has 8 SEs. Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7
for EDC GPR _workarounds,

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Added ASIC specific check in gmc v9.0 ECC interrupt programming sequence

Devices newer then VEGA10/12 shall have these programming sequences performed by PSP BL

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: enlarge agp_start address into 48bit

max range of the agp aperture is 48 bits, so
enlarge agp_start address into 48bit with all bits set

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

drm/amdgpu: disable VCN2.5 ib test for Arcturus sriov

currently using TMR loading VCN fw MMSCH would fail
to init after FLR, just disable ib test for temporarily
daily testing, continuing debug with mm team.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Acked-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix ctx init failure for asics without gfx ring

This workaround does not affect other asics because amdgpu only need expose
one gfx sched to user for now.

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: attempt xgmi perfmon re-arm on failed arm

The DF routines to arm xGMI performance will attempt to re-arm both on
performance monitoring start and read on initial failure to arm.

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

drm/amdgpu: add perfmons accessible during df c-states

During DF C-State, Perfmon counters outside of range 1D700-1D7FF will
encounter SLVERR affecting xGMI performance monitoring.  PerfmonCtr[7:4]
is being added to avoid SLVERR during read since it falls within this
range.  PerfmonCtl[7:4] is being added in order to arm PerfmonCtr[7:4].
Since PerfmonCtl[7:4] exists outside of range 1D700-1D7FF, DF routines
will be enabled to opportunistically re-arm PerfmonCtl[7:4] on retry
after SLVERR.

Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Acked-by: Alex Deucher <Alexander.Deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: simplify padding calculations (v2)

Simplify padding calculations.

v2: Comment update and spacing.

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

drm/amdkfd: expose num_cp_queues data field to topology node (v2)

Thunk driver would like to know the num_cp_queues data, however this data relied
on different asic specific. So it's better to get it from kfd driver.

v2: don't update name size.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: expose num_sdma_queues_per_engine data field to topology node (v2)

Thunk driver would like to know the num_sdma_queues_per_engine data, however
this data relied on different asic specific. So it's better to get it from kfd
driver.

v2: don't update the name size.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: skip disable dynamic state management

Under sriov, the disable operation is no allowed.

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

drm/amdgpu: enable VCN0 and VCN1 sriov instances support for Arcturus

v1: compared to bare-metal: sriov support psp loading VCN firmware; only one
encoding ring would be used in each instance.
v2: keep unchange for bare-metal VCN2.5 hw_init, just add a flag with sriov
and also remove multiple lines.
v3: squash in warning fix

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: skip VCN2.5 power gating and clock gating for sriov Arcturus

v1: skip gating in serveral called functions by power gating and clock gating
v2: from suggestion, skip setting gate in both set function, which is where
it being called.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: update VCN1(dual instances) fw types ID and VCN ip block type

Previously there is no VCN1 type ID in psp gfx interface. Also add VCN ip
block type unless the reinit after FLR for sriov would fail.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add VCN2.5 sriov start for Arctrus

Use MMSCH V1 to finish Memory Controller
programming as well as start MMSCH to do
VCN2.5 initialization.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add VCN2.5 MMSCH start for Arcturus

Use MMSCH to do the initialization since MMSCH
manages VCN2.5 instances and its world switch.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move umc offset to one new header file for Arcturus

Code refactor and no functional change.

Fixes: 4cf781c24c3b ("drm/amdgpu: Added RAS UMC error query support for Arcturus")
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: include delay.h

For udelay.  This is needed for some platforms.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazluaskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu: add metrics table lock for vega20 (v2)

To protect access to the metrics table.

v2: unlock on error

Bug: https://gitlab.freedesktop.org/drm/amd/issues/900
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu: add metrics table lock for renoir (v2)

To protect access to the metrics table.

v2: unlock on error

Bug: https://gitlab.freedesktop.org/drm/amd/issues/900
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu: add metrics table lock for navi (v2)

To protect access to the metrics table.

v2: unlock on error

Bug: https://gitlab.freedesktop.org/drm/amd/issues/900
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu: add metrics table lock for arcturus (v2)

To protect access to the metrics table.

v2: unlock on error

Bug: https://gitlab.freedesktop.org/drm/amd/issues/900
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/smu: add metrics table lock

This table is used for lots of things, add it's own lock.

Bug: https://gitlab.freedesktop.org/drm/amd/issues/900
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

gpu: drm: dead code elimination

this set adds support for removal of gpu drm dead code.

patch3 is similar with patch 1:
`num` is a data of u8 type and ATOM_MAX_HW_I2C_READ == 255,

so there is a impossible condition '(num > 255) => (0-255 > 255)'.

Signed-off-by: Pan Zhang <zhangpan26@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: wait for all rings to drain before runtime suspending

Add a safety check to runtime suspend to make sure all outstanding
fences have signaled before we suspend.  Doesn't fix any known issue.

We already do this via the fence driver suspend function, but we
just force completion rather than bailing.  This bails on runtime
suspend so we can try again later once the fences are signaled to
avoid missing any outstanding work.

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

drm/amdgpu/smu: fix spelling

s/dispaly/display/g

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

drm/amdgpu: Switch from system_highpri_wq to system_unbound_wq

This is to avoid queueing jobs to same CPU during XGMI hive reset
because there is a strict timeline for when the reset commands
must reach all the GPUs in the hive.

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

drm/amdgpu: Redo XGMI reset synchronization.

Use task barrier in XGMI hive to synchronize ASIC resets
across devices in XGMI hive.

v2: Return right away with a warning if no xgmi hive, update doc.

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

drm/amdgpu: Add task barrier to XGMI hive.

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

drm: Add Reusable task barrier.

It is used to synchronize N threads at a rendevouz point before execution
of critical code that has to be started by all the threads at approximatly
the same time.

v2: Remove mention of reset use case, improve doc.

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

drm/amdgpu: reverts commit ce316fa55ef0f1751276b846a54fb3b835bd5e64.

In preparation for doing XGMI reset synchronization using task barrier.

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

drm/amdgpu/vcn: remove unnecessary included headers

Esp. VCN1.0 headers should not be here

v2: add back the <linux/module.h> to keep consistent.

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix KIQ ring test fail in TDR of SRIOV

issues:
MEC is ruined by the amdkfd_pre_reset after VF FLR done

fix:
amdkfd_pre_reset() would ruin MEC after hypervisor finished the VF FLR,
the correct sequence is do amdkfd_pre_reset before VF FLR but there is
a limitation to block this sequence:
if we do pre_reset() before VF FLR, it would go KIQ way to do register
access and stuck there, because KIQ probably won't work by that time
(e.g. you already made GFX hang)

so the best way right now is to simply remove it.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix double gpu_recovery for NV of SRIOV

issues:
gpu_recover() is re-entered by the mailbox interrupt
handler mxgpu_nv.c

fix:
we need to bypass the gpu_recover() invoke in mailbox
interrupt as long as the timeout is not infinite (thus the TDR
will be triggered automatically after time out, no need to invoke
gpu_recover() through mailbox interrupt.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: skip soc clk setting under pp one vf

Under sriov pp one vf mode, there is no need to set
soc clk under pp one vf because smu firmware will depend
on the mclk to set the appropriate soc clk for it.

Signed-off-by: Yintian Tao <yttao@amd.com>
Reviewed-by : Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/scheduler: do not keep a copy of sched list

entity should not keep copy and maintain sched list for
itself.

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>

amd/amdgpu: add sched array to IPs with multiple run-queues

This sched array can be passed on to entity creation routine
instead of manually creating such sched array on every context creation.

v2: squash in missing break fix

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: replace vm_pte's run-queue list with drm gpu scheds list

drm_sched_entity_init() takes drm gpu scheduler list instead of
drm_sched_rq list. This makes conversion of drm_sched_rq list
to drm gpu scheduler list unnecessary

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/scheduler: rework entity creation

Entity currently keeps a copy of run_queue list and modify it in
drm_sched_entity_set_priority(). Entities shouldn't modify run_queue
list. Use drm_gpu_scheduler list instead of drm_sched_rq list
in drm_sched_entity struct. In this way we can select a runqueue based
on entity/ctx's priority for a  drm scheduler.

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/pm_runtime: update usage count in fence handling

Increment the usage count in emit fence, and decrement in
process fence to make sure the GPU is always considered in
use while there are fences outstanding.  We always wait for
the engines to drain in runtime suspend, but in practice
that only covers short lived jobs for gfx.  This should
cover us for longer lived fences.

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

drm/amd/powerplay: Add SMU WMTABLE Validity Check for Renoir

[Why]
SMU watermark table (WMTABLE) validity check is missing on Renoir.
This validity check is very useful for checking whether
WMTABLE is updated successfully.

[How]
Add SMU watermark validity check.

Signed-off-by: Zhan Liu <zhan.liu@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove unneeded semicolon in amdgpu_ras.c

Fixes coccicheck warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:318:2-3: Unneeded semicolon

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove unneeded semicolon in gfx_v10_0.c

Fixes coccicheck warning:

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c:1967:2-3: Unneeded semicolon

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove unneeded semicolon in amdgpu_pmu.c

Fixes coccicheck warning:

drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:110:3-4: Unneeded semicolon
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:133:2-3: Unneeded semicolon
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:163:2-3: Unneeded semicolon
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c:191:2-3: Unneeded semicolon

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove unneeded semicolon

Fixes coccicheck warning:

drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c:412:90-91: Unneeded semicolon

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/sdma5: make ring tests less chatty

We already did this for older generations.

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

drm/amdgpu/gfx10: make ring tests less chatty

We already did this for older generations.

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

drm/amd/display: Add debug option to override DSC target bpp increment

[why]
It's required for debug purposes.

[how]
Add a dsc_bpp_increment_div debug option that overrides DPCD
BITS_PER_PIXEL_INCREMENT value. The value dsc_bpp_increment_div should
be set to is the one after parsing, i.e. it could be 1, 2, 4, 8 or 16
(meaning 1pix, 1/2pix, ..., 1/16pix).

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Do not handle linkloss for eDP

[Why]
eDP is internal link and link loss is unexpected.
It is typically going to be PSR related errors, which is
handled separately.

[How]
Check for eDP and skip check for link loss

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix missing cursor on some rotated SLS displays

[Why]
Cursor disappears for some SLS displays that are rotated 180
and 270 degrees.  This occurs when there is no pipe split being
done ( ex. 3 or more displays ).  The cursor calculations assume
pipe splitting is done so when it calculates the new cursor
position in hwss.set_cursor_position(), it is out-of-bounds so
it disables the cursor in hubp.set_cursor_position().

[How]
In non pipe split cases, calculate cursor using viewport size
( width or height ) instead of viewport size * 2 ( the two
because pipe splitting divides the rectangle into two ).

Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Formula refactor for calculating DPP CLK DTO

[Why]
Previous formula for calculating DPP CLK DTO was
hard to understand.

[How]
Replace with easier to understand formula that produces
same results.

Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Lower DPP DTO only when safe

[Why]
A corner case currently exists where DPP DTO is lowered before
pipes are updated to a higher viewport. This causes underflow
as the DPPCLK is too low for the current viewport.

[How]
Only lower DPP DTO when it is safe to lower, or if
the newer clocks are higher than the current ones.

Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: support virtual DCN

[why]
DAL3 should support SRIOV

[how]
Add support for the virtual dal flag.  This flag should skip
most/all of DC construction since the HW isn't accessible, but
still construct WindowsDM (almost) normally but with only SW display
targets

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix manual trigger source for DCN2

Fix manual trigger source correctly be TRIGA for DCN2
rather than MANUAL_FLOW.

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update extended timeout support for DCN20 and DCN21

[Why]
DCN21 and DCN2 extended timeout support cap is not set correctly.

[How]
Set extended timeout support for ASIC families to their right values.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix update_bw_bounding_box Calcs

[Why]
Previously update_bw_bounding_box for RN was commented out
due to incorrect values causing BSOD on Hybrid Graphics.
However, commenting out this function also may cause issues
such as underflow in certain cases such as 2x4K displays.

[How]
Fix dram_speed_mts calculations.
Update from proper index of clock_limits[]

Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use absolute time stamp to follow the eDP T7 spec requirement

[Why]:
According to eDP spec, max T7 delay should be 50 ms. Current code uses 300
retry counters may not be accurate enough for different panels.

[How]:
Use absolute time stamp to achive accurate delay.

Signed-off-by: Dale Zhao <dale.zhao@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.64

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix 270 degree rotation for mixed-SLS mode

[Why]
When we rotate 270 in mixed SLS mode, the recouts occupy the
right side of the display.  So all the recout_skip_v values
are relative to the left side of the display.  This causes
adjust_vp_and_init_for_seamless_clip() to incorrectly increase
the data->viewport.height for that recout.  The rotation looks
like the bottom half is duplicated twice.

[How]
recout.x values are being adjusted based on
stream->timing.h_border_left.  Instead of using h_border_left,
use dst.x to represent the border.  Shift dst.x by the amount of
stream->timing.h_border_left and set
stream->timing.h_border_left to 0.  Do all the calculations
and then revert stream->timing.h_border_left and
stream->dst.x back to their original values.
When calculating pipe_ctx->plane_res.scl_data.h_active,
make sure to use the original stream->timing.h_border_left
value.

Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Get cache window sizes from DMCUB firmware

[Why]
Firmware state and tracebuffer shouldn't be considered stable API
between firmware versions.

Driver shouldn't be querying anything from firmware state or tracebuffer
outside of debugging.

Commands are the stable API for this once we have the outbox.

[How]
Add metadata struct to the end of the data firmware that describes
fw_state_size and some reserved area for future use.

Drop the tracebuffer and firmware state headers since they can differ
per version.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove reliance on pipe indexing

[Why]
In certain instances, there was a reliance on pipe indexing being accurate. However, this
assumption fails with harvesting of pipes 1 or 2, which can occur in production B6 parts.
HW hang would occur as a result.

[How]
Use hubp index for mpcc, and do mpc_init for all theoretical pipes (including disabled ones).

Signed-off-by: Noah Abradjian <noah.abradjian@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Specified VR patch skip to reset segment to 0

[Why]
After read the 3rd Edid blocks, we will reset segment to 0,
which causes this VR fail to read Edid successfully.

[How]
Skip to reset segment to 0 for this VR device.

Signed-off-by: Derek Lai <Derek.Lai@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: check link status before disable stream

[Why]
1. Set second screen only then unplug external monitor
2. Enter to S4 then plug in external monitor
3. Resume from S4, eDP will not turn off when OS set
   second screen only
Sometimes OS will not set eDP power up cause eDP dpms_off
keep true then driver skipp disable stream

[How]
When drvier try to disable stream, add link status condition

Signed-off-by: Paul Hsieh <paul.hsieh@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable lttpr for Navi

Signed-off-by: Hugo Hu <hugo.hu@amd.com>
Reviewed-by: Abdoulaye Berthe <Abdoulaye.Berthe@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Perform DMUB hw_init on resume

[Why]
The DMUB is put into reset on suspend and is not running on resume,
disabling PSR/ABM features.

[How]
Move the allocation of the framebuffer to sw_init.

Do DMUB hardware init and framebuffer filling only from hw_init.

On resume the contents of the framebuffer will be invalid so those
should be cleared.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Map ODM memory correctly when doing ODM combine

[why]
Up to 4 ODM memory pieces are required per ODM combine and cannot
overlap, i.e. each ODM "session" has to use its own memory pieces.
The ODM-memory mapping is currently broken for generic case.

The maximum number of memory pieces is ASIC-dependent, but it's always
big enough to satisfy maximum number of ODM combines. Memory pieces
are mapped as a bit-map, i.e. one memory piece corresponds to one bit.
The OPTC doing ODM needs to select memory pieces by setting the
corresponding bits, making sure there's no overlap with other OPTC
instances that might be doing ODM.

The current mapping works only for OPTC instance indexes smaller than
3. For instance indexes 3 and up it practically maps no ODM memory,
causing black, gray or white screen in display configs that include
ODM on OPTC instance 3 or up.

[how]
Statically map two unique ODM memory pieces for each OPTC instance
and piece them together when programming ODM combine mode.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add warmup escape call support

Add warmup escape support, for diags, in a way that is possible to
choose a new or an existing sequence. For achieving this goal, this
commit adds separated MCIF buffer as VCN request.

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix regamma build optimization

[Why]
When the global variable pow_buffer_ptr is set to -1, by definition
optimizations should not be used to build the regamma. Since
translate_from_linear_space unconditionally increments this global, it
inadvertently enables the optimization.

[How]
Increment pow_buffer_ptr only if it is not -1.

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: scaling changes should also be a full update

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.63

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove integer scaling code from DC and fix cursor

[Why]
Scaling better handled by upper layers before pipe splitting.

[How]
Remove DC code for integer scaling and force cursor update if
viewport or scaling changes occur to prevent underflow from
invalid cursor position.

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable Seamless Boot Transition for Multiple Streams

[why]
dc previously had bugs that interfered with the ability to inherit a
timing from a device with multiple streams (without flash/blanking).
After this fix there is still a dependency on UEFI support.

[how]
fixed 3 bugs: loaded MPC state, changed bw_optimize flag to a counter
instead of a boolean, and reading dpp/disp clk from HW to ensure we
don't raise the clock's when we're not supposed to.

Signed-off-by: Martin Leung <martin.leung@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Default max bpc to 16 for eDP

[Why]
Some 10bit eDP panels don't lightup after we cap bpc to 8.

[How]
Set default max_bpc to 16 for edp connector type.

Signed-off-by: Roman Li <roman.li@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Collapse resource arrays when pipe is disabled

[Why]
Currently, pipe resources are assigned to an index that matches the pipe position.
However, if pipe 1 or 2 is disabled, there will be a gap in the arrays which causes a crash when iterating based on pipe_count.

[How]
Fix resource construct to assign resources to minimum available array index.

Signed-off-by: Noah Abradjian <noah.abradjian@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: update chroma viewport wa

[Why]
Need previously implemented chroma vp wa to work for rotation cases.

[How]
Implement rotation specific wa.

Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use pipe_count for num of opps

[Why]
There is one opp per pipe. For certain RN parts, the fourth pipe is disabled, so there is no opp for it.
res_cap->num_opp is hardcoded to 4, so if we use that to iterate over opps we will crash.

[How]
Use the pipe_count value instead, which is not hardcoded and so will have the correct number.

Signed-off-by: Noah Abradjian <noah.abradjian@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Reinstate LFC optimization

[why]
We want to streamline the calculations made when entering LFC.
Previously, the optimizations led to screen tearing and were backed out
to unblock development.

[how]
Integrate other calculations parameters, as well as screen tearing,
fixes with the original LFC calculation optimizations.

Signed-off-by: Amanda Liu <amanda.liu@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: AVI info package change due to spec update

YQ should be limited range for all cases.

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

drm/amd/display: Add definition for number of backlight data points

[Why]
A hardcoded number is used today

[How]
Add definition for number of BL data points

Signed-off-by: Camille Cho <Camille.Cho@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add interface to adjust DSC max target bpp limit

[Why]
For some use cases we need to be able to adjust the maximum target bpp
allowed by DSC policy.

[How]
New interface dc_dsc_policy_set_max_target_bpp_limit

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Nikola Cornij <Nikola.Cornij@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Disable integerscaling for downscale and MPO

[Why]
Integer scaling is applied to MPO planes when downscaling,
MPO planes use variable taps and integer scaling sets taps=1

[How]
Disable integer scaling on MPO planes,
Disable integer scaling for downscaling planes

Signed-off-by: Aidan Yang <Aidan.Yang@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use physical addressing for DMCUB on both dcn20/21

[Why]
CW0 and CW1 need to use physical addressing mode for dcn20 and dcn21.

The current code for dcn20 is using virtual.

[How]
We already program the DMCUB like this on dcn21 so we should just use
the same sequence for both.

Copy the dcn21 sequences into the dmjub_dcn20.c file and rename them.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Get DMUB registers from ASIC specific structs

[Why]
These values can differ per ASIC and should follow the full DC style
register programming model.

[How]
Define a common list and fill in the common list separately for
dcn20 and dcn21.

Unlike DC we're not using designated initializers for better compiler
compatibility since this resides in the DMUB service.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add wait for flip not pending on pipe unlock

[Why]
Lack of proper timing caused intermittent underflow on unplug external DP.
A previous fix was invalid and caused S0i3 regression, so had to be reverted.

[How]
When unlocking pipe, wait for no pipes to have flip pending before unlocking.

Signed-off-by: Noah Abradjian <noah.abradjian@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable lttpr for RN

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: HDMI 2.x audio bandwidth check

Add HDMI 2.x audio bandwidth check

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: remove JPEG related code from idle handler and begin use

For VCN2.0 and above, VCN has been separated from JPEG

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn1.0: use its own idle handler and begin use funcs

Because VCN1.0 power management and DPG mode are managed together with
JPEG1.0 under both HW and FW, so separated them from general VCN code.
Also the multiple instances case got removed, since VCN1.0 HW just have
a single instance.

v2: override work func with vcn1.0's own

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

drm/amdgpu: enable gfxoff for raven1 refresh

When smu version is larger than 0x41e2b, it will load
raven_kicker_rlc.bin.To enable gfxoff for raven_kicker_rlc.bin,it
needs to avoid adev->pm.pp_feature &= ~PP_GFXOFF_MASK when it loads
raven_kicker_rlc.bin.

Signed-off-by: changzhu <Changfeng.Zhu@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/sriov: Tonga sriov also need load firmware with smu

Fix Tonga sriov load driver fail issue.

Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewd-by Yintian Tao <Yintian.tao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: drop useless BACO arg in amdgpu_ras_reset_gpu

BACO reset mode strategy is determined by latter func when
calling amdgpu_ras_reset_gpu. So not to confuse audience, drop
it.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add missing dcn link encoder regs

[Why]
The earlier change: "check phy dpalt lane count config"
uses link encoder registers not defined properly.
That caused regression with mst-enabled display not
lighting up.

[How]
Add missing reg definitions.

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add dpcs20 registers

add reg headers to dpcs includes

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move dpcs headers to dpcs includes

- create dpcs directory for dpcs asic_reg headers
- move dpcs21 reg headers from dcn to dpcs directory

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add CU info print log

The log will be useful for easily getting the CU info on various
emulation models or ASICs.

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

drm/amdkfd: Use Arcturus specific set_vm_context_page_table_base()

Since Arcturus has it own function pointer, we can move Arcturus
specific logic to there rather than leaving it entangled with
other GFX9 chips.

Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>