drm/amd/display: Refactor DCN3X into component folder

[why]
Move DCN3X files to unique component folder.

[how]
Create respective component folder in dc, move the DCN3X files into
corresponding new folders and made appropriate changes for compilation
in Makefiles.

Reviewed-by: Martin Leung <martin.leung@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Mounika Adhuri <moadhuri@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: On clock init, maintain DISPCLK freq

[Why]
On init if a display is connected, we need to maintain the DISPCLK
frequency Even though DPG_EN=1, the display still requires the correct
timing or it could cause audio corruption (if DISPCLK freq is reduced).

[How]
Read the current DISPCLK freq and request the same value to ensure the
timing is valid and unchanged.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Chris Park <chris.park@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix minor coding errors where dml21 phase 5 uses wrong variables

There is a coding error which causes incorrect variables to be assigned
in DML21 phase 5.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove redundant condition in VBA 314 func

[WHY]
Coverity analysis this conditional code as DEADCODE.
The conditional statement is never true since
'MacroTileSizeBytes' is either 256 or 65536. Thus, the
code inside is the conditional statement is never reached.

[HOW]
Removed the conditional statement.

Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Reviewed-by: Alex Hung <alex.hung@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove redundant condition with DEADCODE

[WHY]
Coverity analysis flagged this condition as DEADCODE since the
variable 'req128_c' is always false, thus the condition is never
true.

[HOW]
Remove the condition.

Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: workaround for oled eDP not lighting up on DCN401"

This reverts commit e902dd7f3e3b93a401e1d3c0322cce933e75e809.

A proper fix for this issue has been implemented in DMUB FW. So, no need
to keep the workaround.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add dcn401 DIG fifo enable/disable

[Why]
Found while hotplugging MST daisy chain displays. Changing dispclk
during this sequence caused SMU hang due to DIG fifo not being disabled
correctly (caused by missing functions).

[How]
Adding disable/enable DIG fifo functions for dcn401

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Relja Vojvodic <relja.vojvodic@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable DCN401 idle optimizations by default

[WHY&HOW]
Re-enable idle optimizations by default.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: DCN401 full power down in HW init if any link enabled

[Why]
During HW init, certain operations the driver performs are invalid on
enabled hardware in an unknown state (for example, setting all clock
values to minimum when the GPU is actively driving a display). There is
already code present to call HWSS->power_down during init when any link
is enabled in HW, but that function pointer is unpopulated for most asics.
We want to enable this codepath for DCN401, as it resolves the issue with
being unable to drive certain display configs on adapter re-enable, and we
can restore boot optimizations.

[How]
 - add power_down HWSS function for DCN401
 - remove debug bit to disable boot optimizations for DCN401

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine gfx8 firmware loading

refine gfx8 firmware loading

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

drm/amdkfd: add ASIC version check for the reset selection of RAS poison

GFX v9.4.3 uses mode1 reset, other ASICs choose mode2.

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Acked-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: use mode1 reset for RAS poison consumption

Per firmware's requirement, replace mode2 with mode1.

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

drm/amd/pm: update check condition for SMU mode1 reset

The fed status does indicate RAS fatal error.

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

drm/amdgpu: set RAS fed status for more cases

Indicate fatal error for each RAS block and NBIO.

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

drm/amdgpu: create amdgpu_ras_in_recovery to simplify code

Reduce redundant code and user doesn't need to pay attention to RAS
details.

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

drm/amdgpu: trigger mode1 reset for RAS RMA status

Check RMA status in bad page retirement flow.

v2: fix coding bugs in v1.

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

drm/amd/display: Disable PHYSYMCLK RCO

[Why]
PHYSYMCLK RCO has been found to lead to crashes in some
corner cases

[How]
Disable PHYSYMCLK RCO debug bit

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: [FW Promotion] Release 0.0.221.0

 - Create a general command and fix Replay desync error with general cmd

Acked-by: Zaeem Mohamed <zaeem.mohamed@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: 3.2.288

* FW Release 0.0.221.0
* Fixed missing targets in FAMS2
* Populate hardware_release hook for dcn401
* Disable DMCUB timeout for DCN35
* Move PRIMARY plane zpos higher
* Introduce overlay cursor mode
* Change dram_clock_latency for dcn35 and dcn351
* DCN401 cursor code update

Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Update idle hardmins if uclk or fclk requirement changed

[WHY&HOW]
Update the idle hardmin with SMU if either clock changed.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Avoid overflow from uint32_t to uint8_t

[WHAT & HOW]
dmub_rb_cmd's ramping_boundary has size of uint8_t and it is assigned
0xFFFF. Fix it by changing it to uint8_t with value of 0xFF.

This fixes 2 INTEGER_OVERFLOW issues reported by Coverity.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Do not override dml2.1 reinit

[WHY&HOW]
Reinit should return after completing version 2.1 reinit instead of calling
version 2 reinit after.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix missed targets in FAMS2+HWFQ

[WHY&HOW]
Add additional delay factor when considering a safe time to flip for HWFQ
to be passed in by the driver.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add null check for phantom pipes in prepare mcache

[WHY&HOW]
Sometimes this function is called with a partially deconstructed phantom
stream toplolgy, and should ignore phantoms with no plane state.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Block SubVP if hardware rotation is used in DML2.1

[WHY&HOW]
SubVP is not supported when hardware rotation is in use.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Explicitly cast v_total to signed in a subtraction

[WHY & HOW]
v_total is an uint32_t and subtracting an unsigned to a signed will
result in an unsigned which is always >= 0. As a result, the ternary
conditions are always true and thus has no effect.

This is fixed by casting v_total to signed explicitly. This also
avoids v_total subtraction to overflow.

This fixes 1 NO_EFFECT, 2 DEADCODE and 2 INTEGER_OVERFLOW issues reported
by Coverity.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove NO_EFFECT self-assignment

[WHAT & HOW]
The self-assignments have no effects and thus are removed.

Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove reduntant comparisons by Coverity

[WHY]
Coverity analysis of the Upstream display driver code
(amd-staging-drm-next) flagged these three functions as
containing 'CONSTANT_EXPRESSION_RESULT' errors, i. e. the
conditionals are reduntant since their result is predetermined.

fixpt31_32.c:
The two flagged 'ASSERT' lines redundant since they are always true:
- LONG_MAX equals to the largest 'signed long long' number
- res.value is type 'signed long long',

So, the condition inside the 'ASSERTS's is always

link_dp_training.c:
The flagged conditional statement is always false:
- 'max_requested.FFE_PRESET.settings.level' is 4 bits, so its max
  possible value is 15
- 'DP_FFE_PRESET_MAX_LEVEL' equals to 15.

So, the conditional statement is always false and the wrapped code
never executes.

[HOW]
Removed lines flagged by Coverity analysis.

Reviewed-by: Sun peng Li <sunpeng.li@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: populate hardware_release hook for dcn401

[Description]
hardare_release() is called when driver is removed. Add the missing hook for DCN401

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: make ODM segment width of YCbCr422 two pixel aligned

[why]
In YCbCr422 format hardware shares 1 set of chromas CbCr with 2 sets of
lumas Y. Therefore each ODM segment needs to be two pixel aligned. The
commit adds this missing hardwware requirement into ODM segment width
decision logic.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Disable IPS when it is not allowed.

[Why&How]
Add flag to disable IPS when it is not allowed.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: JinZe.Xu <jinze.xu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amd/display: Handle HPD_IRQ for internal link"

[How&Why]
This reverts commit 239b31bd5c3fef3698440bf6436b2068c6bb08a3.

Due to the it effects Replay resync.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Sung-huai Wang <danny.wang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove useless comparison of unsigned int vs. 0

[WHY & HOW]
The comparisons of unsigned int with 0 can have no meanings, i.e.
unsigned int >= 0 (always true) or unsigned int < 0 (always false), and
therefore they are removed.

This fixes 12 NO_EFFECT issues reported by Coverity.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix uninitialized variables in dcn401

This fixes an UNINIT issue reported by Coverity.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix incorrect size calculation for loop in dcn401

[WHY]
fe_clk_en and be_clk_sel have size of 4 but sizeof(fe_clk_en) has
byte size 16 which is lager than the array size.

[HOW]
Use ARRAY_SIZE for calculating size.

This fixes 2 OVERRUN issues reported by Coverity.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Minor cleanup for DCN401 cursor related code

Move pipe_ctx variables to start of the function and add a helpful comment

Co-authored-by: Sridevi Arvindekar <sarvinde@amd.com>
Reviewed-by: Ilya Bakoulin <ilya.bakoulin@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Sridevi Arvindekar <sarvinde@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Check otg_master pointer before use it

Coverity highlighted that the parameter otg_master is referenced before
the if condition that validates it, which means that the code might have
some attempt to access a null pointer. This commit addresses this issue
by moving the pointer verification to the beginning of the function and
initializing all the values right after it.

Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Disable DMCUB timeout for DCN35

[Why]
DMCUB can intermittently take longer than expected to process commands.

Old ASIC policy was to continue while logging a diagnostic error - which
works fine for ASIC without IPS, but with IPS this could lead to a race
condition where we attempt to access DCN state while it's inaccessible,
leading to a system hang when the NIU port is not disabled or register
accesses that timeout and the display configuration in an undefined
state.

[How]
We need to investigate why these accesses take longer than expected, but
for now we should disable the timeout on DCN35 to avoid this race
condition. Since the waits happen only at lower interrupt levels the
risk of taking too long at higher IRQ and causing a system watchdog
timeout are minimal.

Reviewed-by: Ovidiu Bunea <ovidiu.bunea@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: use preferred link settings for dp signal only

[why]
We set preferred link settings for virtual signal. However we don't support
virtual signal for UHBR link rate. If preferred is set to UHBR link rate, we
will allow virtual signal with UHBR link rate which causes system crashes.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: remove ODM combine before bandwidth validation

[why]
DML1 validation code doesn't have the ability to remove ODM combine.
It will directly translate currently used ODM combine config into ODM
override. If ODM combine is used in the initial state it will only
validate the timing if ODM is used. This is not correct for dynamic ODM u
se case, as ODM is used as an optimization not a hard requirement.

[how]
Remove ODM combine in the initial state before bandwidth validation so
DML will not translate it into ODM override in the initial bandwidth
validation.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: remove dpp pipes on failure to update pipe params

[why]
There are cases where update pipe params could fail but dpp pipes are already
added to the state. In this case, we should remove dpp pipes so dc state is
restored back. If it is not restored, dc state is corrupted after calling this
function, so if we call the same interface with the corrupted state again, we
may end up programming pipe topology based on a corrupted dc state.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Remove register from DCN35 DMCUB diagnostic collection

[Why]
These registers should not be read from driver and triggering the
security violation when DMCUB work times out and diagnostics are
collected blocks Z8 entry.

[How]
Remove the register read from DCN35.

Reviewed-by: Duncan Ma <duncan.ma@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Move PRIMARY plane zpos higher

[Why]

Compositors have different ways of assigning surfaces to DRM planes for
render offloading. It may decide between various strategies: overlay,
underlay, or a mix of both (see here for more info:
https://gitlab.freedesktop.org/emersion/libliftoff/-/issues/76)

One way for compositors to implement the underlay strategy is to assign
a higher zpos to the DRM_PRIMARY plane than the DRM_OVERLAY planes,
effectively turning the DRM_OVERLAY plane into an underlay plane.

Today, amdgpu attaches an immutable zpos of 0 to the DRM_PRIMARY plane.
This however, is an arbitrary restriction. DCN pipes are general
purpose, and can be arranged in any z-order. To support compositors
using this allocation scheme, we can set a non-zero immutable zpos for
the PRIMARY, allowing the placement of OVERLAYS (mutable zpos range
0-254) beneath the PRIMARY.

[How]

Assign a zpos = #no of OVERLAY planes to the PRIMARY plane. Then, clean
up any assumptions in the driver of PRIMARY plane having the lowest
zpos.

v2: Fix typo s/decending/descending/

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Introduce overlay cursor mode

[Why]

DCN is the display hardware for amdgpu. DRM planes are backed by DCN
hardware pipes, which carry pixel data from one end (memory), to the
other (output encoder).

Each DCN pipe has the ability to blend in a cursor early on in the
pipeline. In other words, there are no dedicated cursor planes in DCN,
which makes cursor behavior somewhat unintuitive for compositors.

For example, if the cursor is in RGB format, but the top-most DRM plane
is in YUV format, DCN will not be able to blend them. Because of this,
amdgpu_dm rejects all configurations where a cursor needs to be enabled
on top of a YUV formatted plane.

From a compositor's perspective, when computing an allocation for
hardware plane offloading, this cursor-on-yuv configuration result in an
atomic test failure. Since the failure reason is not obvious at all,
compositors will likely fall back to full rendering, which is not ideal.

Instead, amdgpu_dm can try to accommodate the cursor-on-yuv
configuration by opportunistically reserving a separate DCN pipe just
for the cursor. We can refer to this as "overlay cursor mode". It is
contrasted with "native cursor mode", where the native DCN per-pipe
cursor is used.

[How]

On each crtc, compute whether the cursor plane should be enabled in
overlay mode. If it is, mark the CRTC as requesting overlay cursor mode.

Overlay cursor should be enabled whenever there exists a underlying
plane that has YUV format, or is scaled differently than the cursor. It
should also be enabled if there is no underlying plane, or if underlying
planes do not cover the entire CRTC.

During DC validation, attempt to enable a separate DCN pipe for the
cursor if it's in overlay mode. If that fails, or if no overlay mode is
requested, then fallback to native mode.

v2:
* Update commit message for when overlay cursor should be enabled
* Also consider scale and no-underlying-plane case (cursor on crtc bg)
* Consider all underlying planes when determinig overlay/native, not
  just the plane immediately beneath the cursor, as it may not cover the
  entire CRTC.
* Fix typo s/decending/descending/
* Force native cursor on pre-DCN hardware

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: define abm debug interface

[Why & How]
Define debug interface to dmub for reading back abm data.

Reviewed-by: Anthony Koo <anthony.koo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Refactor function dm_dp_mst_is_port_support_mode()

[Why]
dm_dp_mst_is_port_support_mode() is a bit not following the original design rule and cause
light up issue with multiple 4k monitors after mst dsc hub.

[How]
Refactor function dm_dp_mst_is_port_support_mode() a bit to solve the light up issue.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't refer to dc_sink in is_dsc_need_re_compute

[Why]
When unplug one of monitors connected after mst hub, encounter null pointer dereference.

It's due to dc_sink get released immediately in early_unregister() or detect_ctx(). When
commit new state which directly referring to info stored in dc_sink will cause null pointer
dereference.

[how]
Remove redundant checking condition. Relevant condition should already be covered by checking
if dsc_aux is null or not. Also reset dsc_aux to NULL when the connector is disconnected.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: change dram_clock_latency to 34us for dcn35

[Why & How]
Current DRAM setting would cause underflow on customer platform.
Modify dram_clock_change_latency_us from 11.72 to 34.0 us as per recommendation from HW team

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Paul Hsieh <paul.hsieh@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change dram_clock_latency to 34us for dcn351

[Why]
Intermittent underflow observed when using 4k144 display on
dcn351

[How]
Update dram_clock_change_latency_us from 11.72us to 34us

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Correct the defined value for AMDGPU_DMUB_NOTIFICATION_MAX

[Why & How]
It actually exposes '6' types in enum dmub_notification_type. Not 5. Using smaller
number to create array dmub_callback & dmub_thread_offload has potential to access
item out of array bound. Fix it.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Defer the setting of link hpd status for usb4/tbt

[Why & How]
Link hpd status is set during link detection process via dpia_query_hpd_status(),
doesn't need to explicitly set it during outbox irq. Remove it.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Adjust debug msg for usb4/tbt

[Why & How]
Debug msg for usb4/tbt now is a bit confusing. Adjust it for better reading.

Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Avoid programming DTO if Refclk is 0

[Why]
Reference clock, either DPREFCLK or DTBCLK can be a value of 0
which then will encounter division by 0.

[How]
Avoid further calculation and programming if refclk is not
populated.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Disable boot optimization for DCN401

[Why]
DCN401 currently has an issue re-enabling when pipe splitting is enabled,
while the root cause is being investigated we can make sure everything is
being reset as a workaround, by disabling boot optimization.

[How]
 - use enable_accelerated_mode instead of init_pipes to fully reset asic

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Prevent overflow on DTO calculation

[Why]
uint32_t is implicitly converted to uint64_t while multiplication
still happens on uint32_t side.  This creates digit overflow
for large pixel clock which is meant to be retained in uint64_t.

[How]
Calculate multiplication of units in uint64_t domain instead of
uint32_t in DTO parameter clock caculation.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: DCN401 cusor code update

Scaling and rotation changes for cursor.

Reviewed-by: Ariel Bernstein <eric.bernstein@amd.com>
Reviewed-by: Nevenko Stupar <nevenko.stupar@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Sridevi Arvindekar <sarvinde@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use the SPDX license identifier for dmub_replay files

Use the SPDX format for dmub_replay.c|.h files.

Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Remove duplicate 'update_idle_uclk' in dcn401 clk_mgr code

[WHY]
The coverity analysis flagged this if expression as it contains a
'CONSTANT_EXPRESSION_RESULT': 'update_idle_uclk' is 'ORd' with itself.

[HOW]
Removed the duplicate 'update_idle_uclk'.

Reviewed-by: Alex Hung <alex.hung@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Ignore UHBR13.5 cable ID cap for passive cable max link rate

[Why]
Passive DP40 cables were updated in the latest DP spec to support
UHBR13.5 link rate. Current max link rate logic checks against the
cable ID DPCD even for passive cables.

[How]
Ignore UHBR13.5 cable ID DPCD cap in get_max_link_rate logic.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Add config support entry to replay caps debugfs

[Why]
replay_capability debugfs tells whether sink and driver support
replay feature. However replay enablement also depends on
whether it is enabled/disabled via amdgpu module params.

[How]
Add 'Config support' entry to output current replay config.

Reviewed-by: ChiaHsuan Chung <chiahsuan.chung@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Roman Li <roman.li@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine gfx7 firmware loading

refine gfx7 firmware loading

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

drm/amd/display: added NULL check at start of dc_validate_stream

[Why]
prevent invalid memory access

[How]
check if dc and stream are NULL

Co-authored-by: winstang <winstang@amd.com>
Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: winstang <winstang@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Extend gfx12 trap handler fix to gfx10/11

In commit fda812ebe3d9 ("drm/amdkfd: gfx12 context save/restore trap handler fixes")
the following fix was introduced but incorrectly restricted to gfx12.
The same issue and a corresponding fix apply to gfx10 and gfx11.

Do not overwrite TRAPSTS.{SAVECTX,HOST_TRAP} when restoring this
register. Both of these fields can assert while the wavefront is
running the trap handler.

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Lancelot Six <lancelot.six@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.220.0

 - Change ordering of structs to put enums together
 - Add new define DMUB_TRACE_ENTRY_DEFINED to guard
   the trace code enum

Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Add sequential ONO sequencing for DCN35

[Why]
Adds support for performing the sequential ONO changes from DCN351
into DCN35 ASIC based on revision.

[How]
Check the revision and run the DCN351 sequences on applicable revisions.

Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Ignore phantom DTBCLK requirement

[Description]
Phantom DTBCLK can be calculated different from main because phantom
has no DSC and thus will have a different output BPP. Ignore phantom
DTBCLK requirement and only consider non-phantom DTBCLK requirements.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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 fallback defaults for invalid LTTPR DPCD caps

[WHY]
Some sinks are observed to return invalid LTTPR revision and/or invalid
LTTPR link rate capabilities.

[HOW]
Assume any LTTPR which reports invalid max link rate supports HBR3.
Don't validate LTTPR DPCD revision in dp_is_lttpr_present check.

Reviewed-by: George Shen <george.shen@amd.com>
Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Update FIXED_VS retimer training pattern delay

[WHY]
Needed for PHY patterns as well to perform electrical compliance.
Also need to increase wait time from 30ms to 50ms to resolve very
intermittent UHBR20 link training failures.

Reviewed-by: Wenjing Liu <wenjing.liu@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: reduce ODM slice count to initial new dc state only when needed

[why]
We need to decrease ODM slice when adding or removing planes because MPO
support takes precedence over dynamic ODM combine. However there is a case where
we remove ODM combine even for ODM combine required timing in the initial new
dc state. This is normally okay because ODM will be added back after we pass DML
bandwidth validation. However since we remove ODM combine in the initial new
state, the previous ODM pipe allocation is lost. This may cause the new plane to
take away the original secondary OPP head pipe that is still required in the new
state.

For a timing that requires ODM 2:1 but optimized with ODM 4:1, if we add an MPO
plane, we will not have enough pipe to preserve ODM 4:1. In this case we should
reduce ODM slice count then try to add the MPO plane again. By reducing, we are
gradually remove 1 ODM slice from right most side one at a time until we have
enough free pipes for the new plane. If we remove ODM combine entirely, we could
use the pipe at ODM slice index 1 as a DPP pipe for the new plane. But ODM slice
1 is still needed as the timing requires ODM 2:1. This transition is not
seamless and user will see corruption on the screen.

[how]
Remove single ODM slice one at time until we have enough pipes for a new plane.
Remove previous logic to always remove ODM combine entirely.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Fix multiple cursors when using 4 displays on a contiguous large surface

[Why & How]
Remove some cursor offset calculations for rotated cursor for fixing a bug where multiple cursors are seen.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Nevenko Stupar <nevenko.stupar@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Return after Init

why:
DML21 being overwritten after init.

how:
After initializing, early return.

Reviewed-by: Dillon Varone <dillon.varone@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Daniel Sa <daniel.sa@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change AVI Info Frame version to 4 for OVT

[Why]
Official Release CTA specification reverted the change and
no longer uses version 5 to indicate OVT timing.

[How]
Change the version used for AVI info Frame for OVT to 4.

Reviewed-by: Leo Ma <hanghong.ma@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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: Remove NULL assignment for stream_status

[Description]
No need to assign stream_status NULL because it is always
re-assigned before usage. This change is to fix coverity
errors.

Reviewed-by: Nicholas Choi <nicholas.choi@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@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 dcn32_smu13 duplicate header

[WHY]
Duplicate headers requiring unecessary maintenance of both headers

[HOW]
Removal of smu13_driver_if header and all referneces to it changed to dcn32_smu13_driver

Signed-off-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove redundant ternary operation

[WHY]
Coverity analysis flagged this ternary operation as
DEADCODE.

Since 'total_y_free_entry' is initialized as 0x200 and
'total_c_free_entry' is initialized as 0x140, and they
never change values before the ternary operator, so
'total_y_free_entry' is always greater 'total_c_free_entry',
'so small_free_entry' is always set to 'total_c_free_entry'.

[HOW]
Replace the ternary operator with the variable declaration.

Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove unreachable check

[WHY]
Coverity analysis flagged this code as DEADCODE
since the condition and return in the outer loop
are never reached.

All operations with the 'dwb_pipe' variable happen
in the inner loop, that already contains the same
check with the 'MAX_DWB_PIPES'. The later
check condition cannot occur before the check
condition in the inner loop, thus the later
condition is unreacable.

[HOW]
Remove the unreachable condition.

Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove redundant if conditions

[WHY]
Coverity analysis flagged the two if conditions in
dcn30_dpp.c and dcn401_dpp.c as DEADCODE since they
are never true, since the variable
'program_prealpha_dealpha' is initialized at 0 and
never chagnes.

[HOW]
Removed the variable 'program_prealpha_dealpha' and
the if conditions.

Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Signed-off-by: Ivan Lipski <ivlipski@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add odm_slice_rect parameter in spl_in

[why]
OPP input rect aka odm slice rect is a hardware dependent parameter that
can't be determined by SPL software logic. Therefore we need to
explicitly pass odm slice rect in. So ODM slice rect calculation is
moved out of SPL.

[how]
add odm_slice_rect parameter in spl_in

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@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: use even ODM slice width for two pixels per container

[why]
When optc uses two pixel per container, each ODM slice width must be an
even number.

[how]
If ODM slice width is odd number increase it by 1.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@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: Fix Replay Desync Error Test

When PHY power off, the DP_SEC_CNTL cannot be configured and cause
disable Adaptive sync SDP failed. Regarding the issue, the driver will
disabled AS-SDP in replay state machine.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Dennis Chan <dennis.chan@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Replay + IPS + ABM in Full Screen VPB

[Why]
Because ABM will wait VStart to start getting histogram data, it will
cause we can't enter IPS while full screnn video playing.

[How]
Modify the panel refresh rate to the maximun multiple of current refresh
rate

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: ChunTao Tso <ChunTao.Tso@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Expand supported Replay residency mode

[Why]
Dmub provides several Replay residency calculation methods, but current
interface only supports either ALPM or PHY mode

[How]
Modify the interface for supporting different types of Replay residency
calculation.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Leon Huang <Leon.Huang1@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Replay low hz

[Why]
The original coasting vtotal is 2 bytes, and it need to be amended to 4
bytes because low hz case.

[How]
Amend coasting vtotal from 2 bytes to 4 bytes.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: ChunTao Tso <ChunTao.Tso@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use TPS3 for ALPM on Specific Panel

For some specific panel, it need to use TPS3 rather than use TP2 in ALPM
when DSC is enabled.

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

drm/amd/display: Wake DMCUB before sending a command for replay feature

[Why]
We can hang in place trying to send commands when the DMCUB isn't
powered on.

[How]
For functions that execute within a DC context or DC lock we can wrap
the direct calls to dm_execute_dmub_cmd/list with code that exits idle
power optimizations and reallows once we're done with the command
submission on success.

For DM direct submissions the DM will need to manage the enter/exit
sequencing manually.

We cannot invoke a DMCUB command directly within the DM execution helper
or we can deadlock.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@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: Don't use fsleep for PSR exit waits on dmub replay

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

[How]
Replace the flseep with a udelay for dmub_replay_enable.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@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/amdgpu: revert "take runtime pm reference when we attach a buffer" v2

This reverts commit b8c415e3bf98 ("drm/amdgpu: take runtime pm reference
when we attach a buffer") and commit 425285d39afd ("drm/amdgpu: add amdgpu
runpm usage trace for separate funcs").

Taking a runtime pm reference for DMA-buf is actually completely
unnecessary and even dangerous.

The problem is that calling pm_runtime_get_sync() from the DMA-buf
callbacks is illegal because we have the reservation locked here
which is also taken during resume. So this would deadlock.

When the buffer is in GTT it is still accessible even when the GPU
is powered down and when it is in VRAM the buffer gets migrated to
GTT before powering down.

The only use case which would make it mandatory to keep the runtime
pm reference would be if we pin the buffer into VRAM, and that's not
something we currently do.

v2: improve the commit message

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

drm/amd/display: Use dc wake for get the replay state

Use dc_wake_and_execute_gpint with DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY to
get the replay state.

Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Zaeem Mohamed <zaeem.mohamed@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine imu firmware loading

refine imu firmware loading

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

drm/amdgpu: refine pmfw/smu firmware loading

refine pmfw/smu firmware loading

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

drm/amdgpu: refine gmc firmware loading

refine gmc firmware loading

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

drm/amdgpu: refine vpe firmware loading

refine vpe firmware loading

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

drm/amdgpu: refine vcn firmware loading

refine vcn firmware loading

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

drm/amdgpu: move aca/mca init functions into ras_init() stage

adjust the function position to better match aca/mca fini code in ras_fini().

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

drm/amdgpu: fix overflowed constant warning in mmhub_set_clockgating()

To fix potential overflowed constant warning, modify the variables to u32
for getting the return value of RREG32_SOC15().

Signed-off-by: Bob Zhou <bob.zhou@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Indicate CU havest info to CP

To achieve full occupancy CP hardware needs to know if CUs in SE are
symmetrically or asymmetrically harvested

v2: Reset is_symmetric_cus for each loop

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

drm/amdgpu: Skip coredump during resets for debug

Skip scheduling coredump when gpu reset is intentionally triggered
through debugfs.

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

drm/amdgpu: refine sdma firmware loading

refine sdma firmware loading

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

drm/amdgpu: refine psp firmware loading

refine psp firmware loading

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

drm/amdgpu: add lock in kfd_process_dequeue_from_device

We need to take the reset domain lock before talking to MES. While in
this case we can take the lock inside the mes helper. We can't do so for
most other mes helpers since they are used during reset. So for
consistency sake we add the lock here.

Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>