drm/amd/display: Adjust ABM 2.2 contrast parameters

[Why]
Improved contrast in ABM 2.2 is desired

[How]
Increase the contrast factor for ABM levels 2, 3 and 4

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Default to linear output gamma

[Why] Our output TF calculation doesn't work if no user-specified gamma
correction. Normally, user provides this, but driver sohuld just assume
default (linear) gamma otherwise.

[How] Remove output TF dependency on user gamma being provided.

Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Sivapiriyan Kumarasamy <Sivapiriyan.Kumarasamy@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix use of uninitialized union

[Why]
An uninitialized variable would randomly initialize to a large
value. This caused enough delay to fail DP Compliance Test 400.2.1.

[How]
Initialize the variable.

Signed-off-by: John Barberiz <John.Barberiz@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Create switching mechanism for ABM 2.2

[Why]
Need method of detecting which version of the DMCU FW is loaded and
load the appropriate iRAM.

[How]
Create definition for ABM 2.2 iRAM, and load it if the DMCU FW version
number matches the ABM 2.2 version; otherwise load ABM 2.1 iRAM.

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update DMCU versioning mechanism

[Why]
Current date based versioning doesn't tell us about feature version
and build version, and is not useful for debug.

[How]
Add versioning based on feature and build

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Know what a pageflip is

[Why]
We were assuming that any commit with allow_modeset == false
was a pageflip.  This was against drm intention and only
worked by sheer luck

[How]
A pageflip is the change from one framebuffer to another

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Perform plane updates only when needed

[Why]
Our old logic: if pageflip, update freesync and plane address.
Otherwise, update everything.
This over-updated on non-pageflip cases, and it failed to
update if pageflip and non-pageflip changes occurred on
the same commit

[How]
Update flip_addrs on pageflips.
Update scaling_info when it changes.
Update color fields on color changes.
Updates plane_info always because we don't have a good way of
knowing when it needs to be updated.

Unfortunately, this means that every stream commit involves two
calls into DC.  In particular, on pageflips there is a second,
pointless update that changes nothing but costs several
microseconds (about a 50% increase in time taken). The update is
fast, but there are comparisons and some useless programming.

Leave TODOs indicating dissatisfaction.

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Call into DC once per multiplane flip

[Why]
amdgpu_dm_commit_planes was performing multi-plane
flips incorrectly:

It waited for vblank once per flipped plane

It prepared flip ISR and acquired the corresponding vblank ref
once per plane, although it closed ISR and put the ref once
per crtc

It called into dc once per flipped plane, duplicating some work

[How]
Wait for vblank, get vblank ref, prepare flip ISR, and call into
DC only once, and only if there is a pageflip

Make freesync continue to update planes even if vrr information
has already been changed

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Simplify underscan and ABM commit

[Why]
Underscan and ABM are connector properties but require updates
to DC stream state.  Previously, on updates to these properties
the affected stream and all its planes were committed.
This is unnecessary; only a few fields on the stream need
to be changed.

[How]
If scaling or ABM have changed, change the stream and
create a stream update with those changes.  Call
DC with only those fields.

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Let updates with no scaling changes be fast

[Why]
DC was assuming that any surface_update->scaling_info
meant the update was at least medium.  However, if nothing
has changed there is no scaling to program, so there is
no problem with the update being fast

[How]
If every update flag is not set, the update is fast

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Introduce DC VM interface

Virtual memory allows display to support flipping to surfaces which
are not allocated contiguously in memory with physical addresses,
instead a 1-4 level page table is used.

This is beneficial because it allows the scattering of large surfaces
to improve memory efficiency and security.

Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Restructure DCN10 hubbub

[Why]
Change DCN10 hubbub to use hubbub as a base and allow all future DCN
hubbubs to do the same instead of using DCN10_hubbub. This increases
readability and doesn't require future hubbubs to inherit anything
other than the base hubbub struct.

[How]
Create separate DCN10_hubbub struct which uses the hubbub struct as
a base.

Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix eDP fast bootup for pre-raven asic

[Why]
For fastboot, Bios will light up eDP before SW driver is loaded. SW
driver will check if eDP is lit by bios through reading the
BIOS_SCRATCH_3 register. If lit, SW driver will not power down eDP
power and phy to save time.

Definition of BIOS_SCRATCH_3 are missing for pre-raven asic. This
causes eDP fast boot to not work property. For some eDP panels, even
if dp tx sends NoVideoStream_flag =1 and dpcd 0x600=2, eDP rx may not
handle properly. This may cause a short flash on screen.

[How] Add definition of BIOS_SCRATCH_3 for all asic

Signed-off-by: hersen wu <hersenxs.wu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Yongqiang Sun <yongqiang.sun@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: Rework HW fence processing.

Expedite job deletion from ring mirror list to the HW fence signal
callback instead from finish_work, together with waiting for all
such fences to signal in drm_sched_stop we garantee that
already signaled job will not be processed twice.
Remove the sched finish fence callback and just submit finish_work
directly from the HW fence callback.

v2: Fix comments.
v3: Attach  hw fence cb to sched_job
v5: Rebase

Suggested-by: Christian Koenig <Christian.Koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: Refactor ring mirror list handling.

Decauple sched threads stop and start and ring mirror
list handling from the policy of what to do about the
guilty jobs.
When stoppping the sched thread and detaching sched fences
from non signaled HW fenes wait for all signaled HW fences
to complete before rerunning the jobs.

v2: Fix resubmission of guilty job into HW after refactoring.

v4:
Full restart for all the jobs, not only from guilty ring.
Extract karma increase into standalone function.

v5:
Rework waiting for signaled jobs without relying on the job
struct itself as those might already be freed for non 'guilty'
job's schedulers.
Expose karma increase to drivers.

v6:
Use list_for_each_entry_safe_continue and drm_sched_process_job
in case fence already signaled.
Call drm_sched_increase_karma only once for amdgpu and add documentation.

v7:
Wait only for the latest job's fence.

Suggested-by: Christian Koenig <Christian.Koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: enable MGPU fan boost feature on Vega10

For those SKUs which support this feature only.

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

drm/amd/powerplay: support Vega10 fan table V3

MGPU fan boost related parameter is added.

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

drm/amdgpu: remove old CI DPM implementation

The power smu7 powerplay code is much more robust and has
been the default for a while now.  Remove the old code.

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

drm/radeon: check if device is root before getting pci speed caps

Check if the device is root rather before attempting to see what
speeds the pcie port supports.  Fixes a crash with pci passthrough
in a VM.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=109366
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add missing power attribute to APU check

Add missing power_average to visible check for power
attributes for APUs.  Was missed before.

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

drm/amd/display: change the max clock level to 16

As the gfxclk for SMU11 can have at most 16 discrete levels.

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

drm/amd/powerplay: run btc before enabling all SMU features

BTC is needed before enabling all SMU features.

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

drm/amd/powerplay: fit the SOC clock also to the new performance level

The SOC clock needs also to fit the new performance level.

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

drm/amd/powerplay: enforce display related settings only on needed

No display related settings are needed on dpm level change.

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

drm/amd/powerplay: avoid unnecessary dpm level setting

No dpm level setting is needed when the request level
is actually same as current.

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

drm/amdgpu: add flags to emit_ib interface v2

Replace the last bool type parameter with a general flags parameter,
to make the last parameter be able to contain more information.

v2: drop setting need_ctx_switch = false

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

drm/amd/powerplay: OD setting fix on Vega10

gfxclk for OD setting is limited to 1980M for non-acg
ASICs of Vega10

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

drm/amdgpu: sriov should skip asic_reset in device_init

sriov would meet guest driver load failure,
if calling amdgpu_asic_reset in amdgpu_device_init.
sriov should skip asic_reset in device_init.

Signed-off-by: Wentao Lou <Wentao.Lou@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add APTX quirk for Lenovo laptop

Needs ATPX rather than _PR3 for dGPU power control.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=202263
Reviewed-by: Jim Qu <Jim.Qu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amdgpu: fix wrong APU judgement

Fix the APU judgement to make it really work as expected.

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

drm/amdgpu: check PSP support before adding the ip block

So that we do not need to check this in every internal function.

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

drm/amdgpu: Replace kzalloc with kcalloc

Replace kzalloc() function with its 2-factor argument form, kcalloc().

This patch replaces cases of:

kzalloc(a * b, gfp)

with:
kcalloc(a, b, gfp)

Also, improve the coding style and the use of sizeof during
allocation by changing sizeof(struct dc_surface_update) and
sizeof(struct dc_plane_state) to sizeof(*updates) and
sizeof(*surfaces), correspondingly.

This code was detected with the help of Coccinelle.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add support for self irq on Vega10 v2

This finally enables processing of ring 1 & 2.

v2: fix copy&paste error

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-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: add support for processing IH ring 1 & 2

Previously we only added the ring buffer memory, now add the handling as
well.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-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: enable IH ring 1 and ring 2 v4

The entries are ignored for now, but it at least stops crashing the
hardware when somebody tries to push something to the other IH rings.

v2: limit ring size, add TODO comment
v3: only program rings if they are actually allocated
v4: limit the ring init to Vega10

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-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: set the executable flag on unused Vega10 PTEs v2

Otherwise we run into a non-retry fault on access.

It seems to be a hardware bug that the executable bit has
higher priority than the valid bit.

v2: handle clears as well

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

drm/ttm: stop always moving BOs on the LRU on page fault

Move the BO on the LRU only when it is actually moved by a DMA
operation.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Tested-And-Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: support retrieving and adjusting dcefclock power levels V2

User can use "pp_dpm_dcefclk" to retrieve and adjust dcefclock power
levels.

V2: expose this interface for Vega10 and later ASICs only

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

drm/amd/powerplay: support retrieving and adjusting fclock power levels V2

User can use "pp_dpm_fclk" to retrieve and adjust fclock power
levels.

V2: expose this interface for Vega20 and later ASICs only

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

drm/amd/powerplay: support retrieving and adjusting SOC clock power levels V2

User can use "pp_dpm_socclk" to retrieve and adjust SOC clock power
levels.

V2: expose this interface for Vega10 and later ASICs only

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

drm/amd/powerplay: support enabled ppfeatures retrieving and setting V3

User can use "ppfeatures" sysfs interface to retrieve and set enabled
powerplay features.

V2: expose this feature for Vega10 and later dGPUs
V3: squash in removal of unused variable (Alex)

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

drm/amdgpu: set bulk_moveable to false when lru changed v2

if lru is changed, we cannot do bulk moving.
v2:
root bo isn't in bulk moving, skip its change.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: add lru notify to bo driver v2

allow driver do somethings when lru changed.
v2:
address Michel's comments.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/psp: ignore psp response status

In some cases, psp response status is not 0 even there is no
problem while the command is submitted. Some version of PSP FW
doesn't write 0 to that field.
So here we would like to only print a warning instead of an error
during psp initialization to avoid breaking hw_init and it doesn't
return -EINVAL.

Signed-off-by: Aaron Liu <aaron.liu@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Xiangliang Yu<Xiangliang.Yu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Paul Menzel <pmenzel+amd-gfx@molgen.mpg.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Setting doorbell range registers earlier

HW doorbell writing routing policy: writing to doorbell
not in SDMA/IH/MM/ACV doorbell range will be routed to CP.
So CP doorbell routing depends on doorbell range setting
of above blocks. Setting doorbell range of above blocks
earlier (soc15_common_hw_init) to make sure CP doorbell
writing be routed to CP block.

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

drm/amdgpu: Fix sdma doorbell range setting

Different ASIC has different SDMA queue number so
different SDMA doorbell range. Introduce an extra
parameter to sdma_doorbell_range function and set
sdma doorbell range correctly.

Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add per device sdma_doorbell_range field

Different ASIC has different sdma doorbell range. Add
a per device sdma_doorbell_range field and initialize
it.

Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/soc15: return proper error codes in baco reset

Rather than just -1.

Reviewed-by: JimQu <Jim.Qu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Include names of all PP clock types

This fixes printing clock names in cases like:
[    5.352311] [drm] DM_PPLIB: values for Invalid clock
[    5.352313] [drm] DM_PPLIB:   400000 in kHz
[    5.352313] [drm] DM_PPLIB:   933000 in kHz
[    5.352314] [drm] DM_PPLIB:   1067000 in kHz
[    5.352315] [drm] DM_PPLIB:   1200000 in kHz
[    5.352317] [drm] DM_PPLIB: values for Invalid clock
[    5.352318] [drm] DM_PPLIB:   300000 in kHz
[    5.352318] [drm] DM_PPLIB:   600000 in kHz
[    5.352319] [drm] DM_PPLIB:   626000 in kHz
[    5.352320] [drm] DM_PPLIB:   654000 in kHz
(source: HP EliteBook 745 G5 w. RAVEN 0x1002:0x15DD 0x103C:0x83D5 0xD1)

On my system above "Invalid" names got replaced by "F" and "DCF".

The same problem was occurring on Huawei Matebook D with just 667000 kHz
instead of 400000 kHz.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: fix typo

Fix spelling mistake: "lenght" -> "length"

Signed-off-by: Matteo Croce <mcroce@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Skip create new stream if crtc state doesn't change

Need to check if crtc state is changed so that mode set is
required before trying to create new stream.

It deals with the MST hotplug use case when plug back to the
same connector where the failure to create new stream for the
inactive crtc on the old connector.

Signed-off-by: Jerry (Fangzhi) Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use BACO reset on vega20 if platform support

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

drm/amdgpu: use BACO reset if platform support (v2)

It will fall back to use mode1 reset if platform does not support BACO
feature.

v2: squash in warning fix (Alex)

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

drm/amdgpu/powerper: add vega20 BACO functions

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

drm/amdgpu: update THM IP register header to support BACO

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

drm/amdgpu: also include NBIO v7.4 register mask header

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

drm/amdgpu: update NBIO v7.4 to support BACO

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

drm/amdgpu: expose BACO interfaces to upper level from PP

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

drm/amdgpu: add BACO interfaces for vega10

V2: delay 20ms before BACO out.
V3: rename function to vega10_baco_xxx

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

drm/amdpgu: add common functions for BACO feature in PP (v2)

V2: squash in crash fix for non-register commands (Alex)

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

drm/amdgpu: add BACO interfaces in pm and hwmgr function table

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

drm/amdgpu: update nbio v6.1 register/master to support BACO

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

drm/amd/display: Fully remove i2caux folder

This is a follow up to:
e28e1490794d ("drm/amd/display: Remove i2caux folder")

Some files were still left, so delete all of them.

CC: David Francis <David.Francis@amd.com>
CC: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.14

Signed-off-by: Steven Chiu <steven.chiu@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Check for NULL when creating gamma struct

[Wjy&How] Some stress test is causing unexpected memory allocation
failure. This prevents null dereference but there will likely be problems
later, hard to gracefully handle memalloc fail for critical objects.

Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Reza Amini <Reza.Amini@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Check if registers are available before accessing

Check if VERT_FILTER_INIT_BOT and BLACK_OFFSET registers
exists in the DCN SCL IP block before trying to access.

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add new infopacket definition

Modify freesync module to build VTEM infopackets when in HdmiVRR mode

Signed-off-by: Bayan Zabihiyan <Bayan.Zabihiyan@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Rename configure_encoder to enc1_configure_encoder

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add workaround for 4k video underflow

[Why]
On DCN1, there is an issue where on high BW config on single channel
systems, underflow will be observed if DCC is disabled. This issue
can be observed on several use cases. For this particular case,
it is observed when playing 4k video on 4k desktop with video downscaled
to a certain size.

[How]
Block MPO for this particular case, this will prevent extra BW consumed
from downscaling, working around the underflow.

Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: dp interlace MSA timing programming for Interlace mode.

[Why]
DP compliance box shows wrong MSA data.

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix warning on raven hotplug

[Why]
Hotplug on raven results in REG_WAIT_TIMEOUT warning
due to failing attempt to lock disabled otg for the hubp
interdependent pipes programming.

[How]
Don't setup pipe interdependencies for disabled otg.
Also removed the unnecessary duplicate logic checks.

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: dal-pplib interface refactor dal part

[WHY] clarify dal input parameters to pplib interface, remove
un-used parameters. dal knows exactly which parameters needed
and their effects at pplib and smu sides.

current dal sequence for dcn1_update_clock to pplib:

1.smu10_display_clock_voltage_request for dcefclk
2.smu10_display_clock_voltage_request for fclk
3.phm_store_dal_configuration_data {
   set_min_deep_sleep_dcfclk
   set_active_display_count
   store_cc6_data --- this data never be referenced

new sequence will be:

1. set_display_count  --- need add new pplib interface
2. set_min_deep_sleep_dcfclk -- new pplib interface
3. set_hard_min_dcfclk_by_freq
4. set_hard_min_fclk_by_freq

after this code refactor, smu10_display_clock_voltage_request,
phm_store_dal_configuration_data will not be needed for rv.

Signed-off-by: hersen wu <hersenxs.wu@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove i2caux folder

[Why]
It is huge, unmaintainable, needlessly layered, and obsolete

[How]
Remove it.  All of it.  Also remove the i2caux struct in
dc_context and the code that created and destructed it

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change from aux_engine to dce_aux

[Why]
The aux_engine struct is needlessly complex and
is defined multiple times.  It contains function pointers
that each have only one version and are called only from
inside dce_aux.

[How]
Replace aux_engine with a new struct called dce_aux.
Remove all function pointers and call functions directly.
Remove unused functions

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Switch ddc to new aux interface

[Why]
The old aux interface goes through i2caux and the aux_engine
and engine function pointers.  The multiple layers of indirection
make it hard to tell waht is happening.  The aux algorithm
does not need to be this complicated: attempt to submit the
request.  If you get an ack (reply = 0), stop.  Otherwise,
retry, up to 7 times.

[How]
Add a new helper function in dce_aux that performs aux retries

Move the plumbing of the aux calling code into dce_aux

Add functions in ddc that redirect directly to dce_aux

Make all aux calls use these functions

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Shift dc link aux to aux_payload

[Why]
aux_payload should be the struct used inside dc to start
aux transactions.  This will allow the old aux interface
to be seamlessly replaced.

[How]
Add three fields to aux_payload: reply, mot, defer_delay
This will mean that aux_payload has all data required
to submit a request.  Shift dc_link to use this struct

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.13

Signed-off-by: Steven Chiu <steven.chiu@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix disabled cursor on top screen edge

[Why]
The cursor vanishes when touching the top of edge of the screen for
Raven on Linux.

This occurs because the cursor height is not taken into account when
deciding to disable the cursor.

[How]
Factor in the cursor height into the cursor calculations - and mimic
the existing x position calculations.

Fixes: 94a4ffd1d40b ("drm/amd/display: fix PIP bugs on Dal3")
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set link rate set if eDP ver >= 1.4.

[Why]
If eDP ver >= 1.4,
the Source device must use LINK_RATE_SET.

[How]
Get LINK_RATE_SET by reading DPCD 10h-1fh,
then write DPCD 00115h before link training.

Signed-off-by: Derek Lai <Derek.Lai@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix for NULL ramp pointer crashing driver

[Why]
In certain scenarios the ramp parameter come in as NULL, which crashes
because this function doesn't guard properly in the early return.

[How]
- parameter mapUserRamp should be the guard (false means no ramp)
- remove checking ramp in early return

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Sivapiriyan Kumarasamy <Sivapiriyan.Kumarasamy@amd.com>
Acked-by: Eryk Brol <Eryk.Brol@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Redefine DMCU_SCRATCH to identify DMCU state

[why]
To resume system before entering S0i3 completely will cause PSP not
reload DMCU FW since there is not HW power state change.
In this case, driver cannot get correct DMCU version from IRAM
since driver override it and DMCU didn't reload to update it.
It makes driver return false in dcn10_dmcu_init().

[how]
1.To redefine DMCU_SCRATCH to identify different DMCU state.
2.To reserve IRAM 0xF0~0xFF write by DMCU only.
3.To remove dcn10_get_dmcu_state

Signed-off-by: Martin Tsai <martin.tsai@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix PME notification not working in RV desktop

[Why]
PPLIB not receive the PME when unplug.

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: refactor reg_update

[why]
separate register access from logic to allow us abstract register sequences

[how]
consolidate mask and value first then apply to register.

Signed-off-by: Tony Cheng <tony.cheng@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Signal hw_done() after waiting for flip_done()"

This reverts commit 717276b9256f5d97b43e53adca1670cee2c45db2.

The reverted patch was a workaround that addressed an issue fixed by:
4364bcb2cd21 ("drm: Get ref on CRTC commit object when waiting for
flip_done")

Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix CTS 4.2.2.8

[Why]
1-Test equipment does not reset test automation flag after completing
current test causing the next test to fail.
2.When test request is empty, we shouldn't ack the test.

[How]
1-Driver should always clear test equipment automation test request
when request is completed.
2-Driver should clear test equipement test automation if driver does
not complete the request.

Signed-off-by: abdoulaye berthe <abdoulaye.berthe@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Pack DMCU iRAM alignment

[Why]
When the DMCU's iRAM definition was moved to the newly created
power_helpers, a #pragma pack was lost, causing the iRAM to be misaligned

[How]
Restore the #pragma pack

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix missing hwss function for dce

[Why]
The driver will crash on dce hardware due to a null function pointer.

[How]
- bring back "program_csc_matrix" functionality as "program_output_csc" for
dce110
- dce110 doesn't use the "opp_id" parameter, so use 0 when there's no hubp

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.12

Signed-off-by: Steven Chiu <steven.chiu@amd.com>
Reviewed-by: Yongqiang Sun <yongqiang.sun@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Only get the connector state for VRR when toggled

[Why]
This fixes a stuttering issue that occurs when moving a hardware cursor
when VRR is enabled.

Previously when VRR is enabled atomic check will grab the connector
state for every atomic update. This has to lock the connector in order
to do so. The locking is bad enough by itself for performance, but
it gets worse with what we do just below that - add all the planes
for the CRTC to the commit.

This prevents the cursor fast path from working - there's more than one
plane now. With state->allow_modeset = true on top of this, it also
adds and removes all the planes from the DC context triggering a full
(very slow) update in DC.

[How]
We need the connector state to get the VRR min/max capbilities, but we
only need them when there's a CRTC mode change or when VRR is toggled.

The condition has been updated accordingly.

Fixes: 3cc22f281318 ("drm/amdgpu: Set FreeSync state using drm VRR properties")
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: run acg btc for Vega12

acg btc was added to Vega12

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

drm/amdkfd: Don't assign dGPUs to APU topology devices

dGPUs need their own topology devices. Don't assign them to APU topology
devices with CPU cores.

Bug: https://github.com/RadeonOpenCompute/ROCK-Kernel-Driver/issues/66
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Tested-by: Elias Konstantinidis <ekondis@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Allow building KFD on ARM64 (v2)

ifdef x86_64 specific code.
Allow enabling CONFIG_HSA_AMD on ARM64.

v2: Fixed a compiler warning due to an unused variable

CC: Mark Nutter <Mark.Nutter@arm.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Tested-by: Mark Nutter <Mark.Nutter@arm.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: check if we need to reset at init time (v2)

To deal with situations like kexec or GPU VM passthrough
where the device may have been used previously without a
proper GPU reset between.

v2: rebase

bug: https://bugs.freedesktop.org/show_bug.cgi?id=108585
bug: https://bugs.freedesktop.org/show_bug.cgi?id=108754
Reviewed-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/soc15: add need_reset_on_init asic callback for SOC15 (v2)

SOC15 chips require a reset if the driver was previously loaded
because the PSP can only be loaded once between each reset.

v2: rebase, handle multiple asic funcs

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

drm/amdgpu/vi: add need_reset_on_init asic callback for VI (v2)

VI chips require a reset if the driver was previously loaded
because the SMU can only be loaded once between each reset.

v2: rebase

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

drm/amdgpu/cik: add need_reset_on_init asic callback for CIK (v2)

CIK chips require a reset if the driver was previously loaded
because the SMU can only be loaded once between each reset.

v2: rebase

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

drm/amdgpu/si: add need_reset_on_init asic callback for SI (v2)

SI chips don't require a reset on reload due to the nature of
the SMU on them.

v2: rebase

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

drm/amdgpu: add need_reset_on_init asic callback (v2)

Used to determine if we need to reset the asic on init due
to the driver having been previously loaded or not shutdown
cleanly.  E.g., kexec or VM passthrough.

v2: rebase

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

drm/amdgpu: expose sclk and mclk via hwmon

Expose sclk (gfx clock) and mclk (memory clock) via
hwmon compatible interface.  hwmon does not actually
formally specify a frequency type attribute, but these
are compatible with the format of the other attributes
exposed via hwmon.  Units are hertz.

freq1_input - GPU gfx/compute clock in hertz
freq2_input - GPU memory clock in hertz (dGPU only)

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

drm/amdgpu: Add sysfs file for PCIe usage v5

Add a sysfs file that reports the number of bytes transmitted and
received in the last second. This can be used to approximate the PCIe
bandwidth usage over the last second.

v2: Clarify use of mps as estimation of bandwidth
v3: Don't make the file on APUs
v4: Early exit for APUs in the read function, change output to
    display "packets-received packets-sent mps"
v5: fix missing header for si (Alex)

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