drm/amd/display: HDMI remote sink need mode validation for Linux

[Why]
Currently mode validation is bypassed if remote sink exists. That
leads to mode set issue when a BW bottle neck exists in the link path,
e.g., a DP-to-HDMI converter that only supports HDMI 1.4.

Any invalid mode passed to Linux user space will cause the modeset
failure due to limitation of Linux user space implementation.

[How]
Mode validation is skipped only if in edid override. For real remote
sink, clock limit check should be done for HDMI remote sink.

Have HDMI related remote sink going through mode validation to
elimiate modes which pixel clock exceeds BW limitation.

Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change to correct unit on audio rate

[Why]
Formula uses kHz in their formula while our driver operates with Hz.

[How]
Divide audio rate by 1000 on the initial variable that is entered into formula.

Signed-off-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Acked-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Avoid set zero in the requested clk

[Why]
Sometimes CRTCs can be disabled due to display unplugging or temporarily
transition in the userspace; in these circumstances, DCE tries to set
the minimum clock threshold. When we have this situation, the function
bw_calcs is invoked with number_of_displays set to zero, making DCE set
dispclk_khz and sclk_khz to zero. For these reasons, we have seen some
ATOM bios errors that look like:

[drm:atom_op_jump [amdgpu]] *ERROR* atombios stuck in loop for more than
5secs aborting
[drm:amdgpu_atom_execute_table_locked [amdgpu]] *ERROR* atombios stuck
executing EA8A (len 761, WS 0, PS 0) @ 0xEABA

[How]
This error happens due to an attempt to optimize the bandwidth using the
sclk, and the dispclk clock set to zero. Technically we handle this in
the function dce112_set_clock, but we are not considering the case that
this value is set to zero. This commit fixes this issue by ensuring that
we never set a minimum value below the minimum clock threshold.

Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: align frag_end to covered address space

align frag_end to the next pd when there are no
page table entries on the current pde.
This fixes invalidation of larger address space areas
where some page tables are allocated and other aren't.

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

drm/amdgpu: fix NULL pointer dereference for Renoir

Commit c1cf79ca5ced46 ("drm/amdgpu: use IP discovery table for renoir")
introduced a NULL pointer dereference when booting with
amdgpu.discovery=0, because it removed the call of vega10_reg_base_init()
for that case.

Fix this by calling that funcion if amdgpu_discovery == 0 in addition to
the case that amdgpu_discovery_reg_base_init() failed.

Fixes: c1cf79ca5ced46 ("drm/amdgpu: use IP discovery table for renoir")
Signed-off-by: Dirk Gouders <dirk@gouders.net>
Cc: Hawking Zhang <Hawking.Zhang@amd.com>
Cc: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: add interrupt work handler for smu11 parts

We need to schedule the smu AC/DC interrupt ack to avoid
potentially sleeping if the smu message mutex is contended.

Fixes: e1188aacad1730 ("drm/amdgpu/smu11: add support for SMU AC/DC interrupts")
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: add interrupt work function

So we can schedule work from interrupts.  This might include
long tasks or things that could sleep.

Fixes: e1188aacad1730 ("drm/amdgpu/smu11: add support for SMU AC/DC interrupts")
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable GDDR6 save-restore support for navy_flounder

add mp0 11_0_11 for navy_flounder to the mem training
supported list, otherwise the modeprobe would fail
on navy_flounder with latest vbios.

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

drm/amdgpu: support indirect access reg outside of mmio bar (v2)

support both direct and indirect accessor in unified
helper functions.

v2: Retire indirect mmio access via mm_index/data

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: switch to indirect reg access helper

Switch WREG32/RREG32_PCIE to use indirect reg access
helper for soc15 and onwards

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add helper function for indirect reg access (v3)

Add helper function in order to remove RREG32/WREG32
in current pcie_rreg/wreg function for soc15 and
onwards adapters.
PCIE_INDEX/DATA pairs are used to access regsiters
outside of mmio bar in the helper functions.
The new helper functions help remove the recursion
of amdgpu_mm_rreg/wreg from pcie_rreg/wreg and
provide the oppotunity to centralize direct and
indirect access in a single function.

v2: Fixed typo and refine the comments

v3: Remove unnecessary volatile local variable

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdkfd: Surface files in Sysfs to allow users to get number of
compute units that are in use.

[Why]
Allow user to know how many compute units (CU) are in use at any given
moment.

[How]
Surface files in Sysfs that allow user to determine the number of compute
units that are in use for a given process. One Sysfs file is used per
device.

Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Reviewed-By: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: Define and implement a function that collects number of
waves that are in flight.

[Why]
Allow user to know how many compute units (CU) are in use at any given
moment.

[How]
Read registers of SQ that give number of waves that are in flight
of various queues. Use this information to determine number of CU's
in use.

Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Reviewed-By: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: disable gfxoff temporarily for navy_flounder

gfxoff is temporarily disabled for navy_flounder, since
at present the feature caused some tdr when performing
display operations.

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

drm/amdgpu: drop duplicated ecc check for vega10 (v5)

The same ECC check has been executed in amdgpu_ras_init for vega10,
prior to gmc_v9_0_late_init.

v2: drop all atombios helper callings
v3: use bit operation
v4: correct inline comment, remove parity check statement
v5: squash in build fix

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

drm/amd/display: add pipe reassignment prevention code to dcn3

Add code to gracefuly handle any pipe reassignment
occuring on dcn3 hardware. This should only happen when new
surfaces are used for an update rather than old ones updated.

Fixes: 69fc1f4b976cea ("amd/drm/display: avoid dcn3 on flip opp change for slave pipes")
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use function pointer for gfxhub functions

gfxhub functions are now called from function pointers,
instead of from asic-specific functions.

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/amd/amdgpu: Prepare implementation to support reporting of CU usage

[Why]
Allow user to know number of compute units (CU) that are in use at any
given moment.

[How]
Read registers of SQ that give number of waves that are in flight
of various queues. Use this information to determine number of CU's
in use.

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

drm/amd/amdgpu: Clean up header file of symbols that are defined to be static

[Why]
Header file exports functions get_gpu_clock_counter(), get_cu_info() and
select_se_sh() that are defined to be static

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

drm/amdgpu: Use SKU instead of DID for FRU check v2

The VG20 DIDs 66a0, 66a1 and 66a4 are used for various SKUs that may or may
not have the FRU EEPROM on it. Parse the VBIOS to check for server SKU
variants (D131 or D134) until a more general solution can be determined.

v2: Remove string-based logic, correct the VBIOS string comment

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

drm/amdgpu/swsmu/smu12: fix force clock handling for mclk

The state array is in the reverse order compared to other asics
(high to low rather than low to high).

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1313
Reviewed-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove duplicate call to rn_vbios_smu_get_smu_version()

Commit 78fe9f63947a2b ("drm/amd/display: Remove DISPCLK Limit Floor for Certain SMU Versions")
added a call to rn_vbios_smu_get_smu_version() to set clk_mgr->smu_ver.
That field is initialized prior to the if-statement, already.

Fixes: 78fe9f63947a2b (drm/amd/display: Remove DISPCLK Limit Floor for Certain SMU Versions)
Signed-off-by: Dirk Gouders <dirk@gouders.net>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Sung Lee <sung.lee@amd.com>
Cc: Yongqiang Sun <yongqiang.sun@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: add one sysfs file to support the feature to modify gfx clock on Raven/Raven2/Picasso APU.

This patch is to add one sysfs file -- "pp_od_clk_voltage" for
Raven/Raven2/Picasso APU, which is only used by dGPU like VEGA10.
This sysfs file supports the feature to modify gfx engine clock(Mhz units), it can
be used to configure the min value and the max value for gfx clock limited in the
safe range.

Command guide:
echo "s level clock" > pp_od_clk_voltage
s - adjust teh sclk level
level - 0 or 1, "0" represents the min value, "1" represents the max value
clock - the clock value(Mhz units), like 400, 800 or 1200, the value must be within the
                OD_RANGE limits.
Example:
$ cat pp_od_clk_voltage
OD_SCLK:
0:        200Mhz
1:       1400Mhz
OD_RANGE:
SCLK:     200MHz       1400MHz

$ echo "s 0 600" > pp_od_clk_voltage
$ echo "s 1 1000" > pp_od_clk_voltage
$ cat pp_od_clk_voltage
OD_SCLK:
0:        600Mhz
1:       1000Mhz
OD_RANGE:
SCLK:     200MHz       1400MHz

Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new trace event for page table update

This patch adds a new trace event to track the PTE update
events. This specific event will provide information like:
- start and end of virtual memory mapping
- HW engine flags for the map
- physical address for mapping

This will be particularly useful for memory profiling tools
(like RMV) which are monitoring the page table update events.

V2: Added physical address lookup logic in trace point
V3: switch to use __dynamic_array
    added nptes int the TPprint arguments list
    added page size in the arg list
V4: Addressed Christian's review comments
    add start/end instead of seg
    use incr instead of page_sz to be accurate
V5: Addressed Christian's review comments:
    add pid and vm context information in the event
V6: Re-sequence the variables (put pid and ctx_id first)

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

drm/amdgpu: fix incorrect comment

It should be one copy-paste typo.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: restore proper ref count in amdgpu_display_crtc_set_config

A recent attempt to fix a ref count leak in
amdgpu_display_crtc_set_config() turned out to be doing too much and
"fixed" an intended decrease as if it were a leak. Undo that part to
restore the proper balance. This is the very nature of this function
to increase or decrease the power reference count depending on the
situation.

Consequences of this bug is that the power reference would
eventually get down to 0 while the display was still in use,
resulting in that display switching off unexpectedly.

Signed-off-by: Jean Delvare <jdelvare@suse.de>
Fixes: e008fa6fb415 ("drm/amdgpu: fix ref count leak in amdgpu_display_crtc_set_config")
Cc: stable@vger.kernel.org
Cc: Navid Emamdoost <navid.emamdoost@gmail.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: make two symbols static

This addresses the following sparse warning:

drivers/gpu/drm/amd/amdgpu/../display/dc/dce110/dce110_hw_sequencer.c:2740:6:
warning: symbol 'dce110_set_cursor_position' was not declared. Should it
be static?
drivers/gpu/drm/amd/amdgpu/../display/dc/dce110/dce110_hw_sequencer.c:2785:6:
warning: symbol 'dce110_set_cursor_attribute' was not declared. Should
it be static?

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

drm/amd/display: make get_color_space_type() static

This addresses the following sparse warning:

drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_hw_sequencer.c:180:26:
warning: symbol 'get_color_space_type' was not declared. Should it be
static?

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

amd/drm/display: avoid dcn3 on flip opp change for slave pipes

At the moment on flip opp reassignment does not work in all cases
for non root pipes.
This change simply makes sure we prefer pipes not used previously
when splitting in dcn3.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Reviewed-by: Eric Bernstein <eric.bernstein@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disable stream if pixel clock changed with link active

[Why]
Vbios uses preferred timing to turn on edp but OS could use other
timing.  If change pixel clock when link active, there is unexpected
garbage on monitor.

[How]
Once pixel clock changed, the driver needs to disable stream.

Signed-off-by: Chiawen Huang <chiawen.huang@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.35

[Header Changes]
   - Definition for retaining ABM settings during disable
   - Addition of some new AUX interface definitions
   - Addition of some outbox definitions

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

drm/amd/display: Revert check for flip pending before locking pipes

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

drm/amd/display: Add debug param to force dio disable

[WHY]
At the moment, some tests are failing because cur_link_settings is
invalid. As a workaround, add an option to force dio disable.

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

drm/amd/display: Calc DLG from dummy p-state if full p-state unsupported

[Why]
Currently, when full p-state changes are not supported, DLG parameters
are calculated for no p-state support at all. However, we are required
to always support dummy p-state changes, so we should instead calculate
DLG based on dummy p-state latency when full p-state is unsupported.
This behaviour already exists for DCN2.

[How]
 - move DLG calculation inside WM calculation
 - if p-state unsupported, do not recalculate for set A, instead copy from
set C, and perform DLG calculation with dummy p-state latency

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

drm/amd/display: disable stream if pixel clock changed with link active

[Why]
Vbios uses preferred timing to turn on edp but OS could use other
timing. If change pixel clock when link active, there is unexpected
garbage on monitor.

[How]
Once pixel clock changed, the driver needs to disable stream.

Signed-off-by: Chiawen Huang <chiawen.huang@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Ensure all debug bits are passed to fw

[Why]
Some debug bits are not being copied from driver to fw.

[How]
Copy debug bits properly.

Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add dp_set_dsc_pps_info_packet to virtual stream encoder

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update NV1x SR latency values

[Why]
HW team measurement requires updating values

[How]
Update bounding box values

Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: fix screen flicker seen on Navi14 with 2*4K monitors

Revert the guilty change introduced by the commit below:
drm/amd/pm: postpone SOCCLK/UCLK enablement after DAL initialization(V2)

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/amdgpu: fix a warning in amdgpu_ras.c (v2)

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c: In function ‘amdgpu_ras_fs_init’:
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:1284:2: warning: ignoring return value of ‘sysfs_create_group’, declared with attribute warn_unused_result [-Wunused-result]
 1284 |  sysfs_create_group(&adev->dev->kobj, &group);
      |  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

v2: just print an error for sysfs group creation failure

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

drm/amdgpu: clean up ras sysfs creation (v2)

Merge ras sysfs creation together by calling sysfs_create_group
once, as sysfs_update_group may not work properly as expected.

v2: improve commit message

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Dennis Li <Dennis.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: stop data_exchange work thread before reset

In FLR routine, init_data_exchange is called at reset_sriov
while fini_data_exchange is not. This will duplicating work
thread.

So call fini_data_exchange before reset for SRIOV

Signed-off-by: Tiecheng Zhou <Tiecheng.Zhou@amd.com>
Signed-off-by: Bokun Zhang <Bokun.Zhang@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Implement new guest side VF2PF message transaction (v2)

- Refactor the driver code to use amdgpu_virt_read_pf2vf_data
  and amdgpu_virt_write_vf2pf_data instead of writing all code in
  one function (which is the old amdgpu_virt_init_data_exchange)

- Adding a new transaction method for VF2PF message between host
  and guest driver. Guest side will periodically update VF2PF
  message in the framebuffer.

  In the new header, we include guest ucode information, guest
  framebuffer usage, and engine usage

- Clean up the old macros since they will cause compile error if
  the new transaction method is used

v2: squash in build fix

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

drm/amdgpu: Update VF2PF interface

- Update guest side VF2PF interface header file

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

drm/include: add PP_FEATURE_MASK comments (v3)

Documents PP_FEATURE_MASK enum.
Provides instructions on how to use ppfeaturemasks.

v2: improve enum definitions and add kernel command line parameters to
    ppfeaturemask instructions
v3: fix alignment issues

Signed-off-by: Ryan Taylor <ryan.taylor@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: disable sienna chichlid UMC RAS

disable UMC RAS in lieu of stability issues on certain sku

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

drm/amdgpu: add an auto setting to the noretry parameter

This allows us to set different defaults on a per asic basis.  This
way we can enable noretry on dGPUs where it can increase performance
in certain cases and disable it on chips where it can be problematic.

For now the default is 0 for all asics, but we may want to try and
enable it again for newer dGPUs.

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

drm/amdgpu: store noretry parameter per driver instance

This will allow us to have different defaults per asic
in a future patch.

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

drm/amdgpu/display: fix CFLAGS setup for DCN30

Properly handle clang and older versions of gcc.

Fixes: e77165bf7b02a3 ("drm/amd/display: Add DCN3 blocks to Makefile")
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove some useless code

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

drm/amd: Skip not used microcode loading in SRIOV

smc, sdma, sos, ta and asd fw is not used in SRIOV. Skip them to
accelerate sw_init for navi12.

v2: skip above fw in SRIOV for vega10 and sienna_cichlid
v3: directly skip psp fw loading in SRIOV
Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Reviewed-by: Emily.Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Skip use smc fw data in SRIOV

smc fw is not needed in SRIOV, thus driver should not try to get smc
fw data.

Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Reviewed-by: Emily.Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix return value check for hdcp_work

max_caps might be 0, thus hdcp_work might be ZERO_SIZE_PTR

Signed-off-by: Flora Cui <flora.cui@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove gpu_info fw support for sienna_cichlid etc.

Remove gpu_info fw support for sienna_cichlid etc., since the
information can be retrieved from discovery binary.

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

gpu/drm/radeon: fix spelling typo in comments

Modify the comment typo: "definately" -> "definitely".

Signed-off-by: Wang Qing <wangqing@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: optimize code runtime a bit

In the function dal_ddc_service_query_ddc_data,
get rid of dal_ddc_i2c_payloads_destroy, call
dal_vector_destruct() directly.
This change is to make the code run a bit fast.

Signed-off-by: Bernard Zhao <bernard@vivo.com>
Changes since V1:
*get rid of dal_ddc_i2c_payloads_destroy, call
dal_vector_destruct() directly.

Link for V1:
*https://lore.kernel.org/patchwork/patch/1309014/

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

drm/amd: fix typoes in comments

Change the comment typo: "programm" -> "program".

Signed-off-by: Bernard Zhao <bernard@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: fix typoes in comments

Change the comment typo: "programm" -> "program".

Signed-off-by: Bernard Zhao <bernard@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add initial kernel documentation for the amd_ip_block_type structure. v3

Added IP block section to amdgpu.rst.
Added more documentation to amd_ip_funcs.
Created documentation for amd_ip_block_type.

v2: Provides a more detailed DOC section on IP blocks
v3: Clarifies the IP block list. Adds info on IP block enumeration.

Signed-off-by: Ryan Taylor <ryan.taylor@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix hdp register access error

mmHDP_READ_CACHE_INVALIDATE register is in HDP not in NBIO

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

drm/amd/pm: update driver if file for sienna cichlid

Update driver if file for sienna cichlid.

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

drm/amd/pm: drop redundant watermarks bitmap setting

As this is already set inside the implementation of
smu_set_watermarks_table().

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

drm/amd/pm: decouple the watermark table setting from socclk/uclk dpms

As they have no real dependence. And for Navi1x, the socclk/uclk dpms
are enabled after DAL initialization.

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

drm/amd/pm: correct the pmfw version check for Navi14

Otherwise, that will be always true for Navi14.

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: update athub interrupt harvesting handle

GCEA/MMHUB EA error should not result to DF freeze, this is
fixed in next generation, but for some reasons the GCEA/MMHUB
EA error will result to DF freeze in previous generation,
diver should avoid to indicate GCEA/MMHUB EA error as hw fatal
error in kernel message by read GCEA/MMHUB err status registers.

Changed from V1:
    make query_ras_error_status function more general
    make read mmhub er status register more friendly

Changed from V2:
    move ras error status query function into do_recovery workqueue

Changed from V3:
    remove useless code from V2, print GCEA error status
    instance number

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

drm/amd/pm: Removed fixed clock in auto mode DPM

SMU10_UMD_PSTATE_PEAK_FCLK value should not be used to set the DPM.

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

drm/amd/powerplay: optimize the mclk dpm policy settings

Different mclk dpm policy will be applied based on the VRAM
width.

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

drm/amdgpu/gmc9: simplify the return expression of gmc_v9_0_suspend

Simplify the return expression.

Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: simplify the return expression of smu_hw_fini

Simplify the return expression.

Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: simplify the return expression

Simplify the return expression.

Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/mes: simplify the return expression of mes_v10_1_ring_init

Simplify the return expression.

Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FIX] update clock under two conditions

[Why]
Update clock only when non-seamless boot stream exists
creates regression on multiple scenerios.

[How]
Update clock in two conditions
1. Non-seamless boot stream exist.
2. Stream_count = 0

Fixes: 598c13b21e25 ("drm/amd/display: update clock when non-seamless boot stream exist")
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com>
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Qingqing Zhuo <Qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <stable@vger.kernel.org>

drm/amd/display: 3.2.104

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.34

[Header Changes]
       - Add new SCRATCH0 status bits for detecting restore state

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: TMDS Fallback transition

[Why]
HDMI requires fallback to TMDS by redetection
in order to switch PHY settings.
This avoids black out when link training fail
during mode setting, link quality update,
disable driver sequence.

[How]
Allow driver to redetect HDMI displays
based on retraining or fallback mechanism.

Signed-off-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Check for flip pending before locking pipes.

[Why]
When running a game/benchmark with v-sync disabled, disabling a plane
(which is v-sync) can cause an underflow. This is due to flips that are
pending before pipe locking being applied after locks are released and
pipes have been re-arranged or disconnected. This can potentially apply
a flip on the incorrect pipe.

[How]
Check that any pending flips are cleared before locking any pipes to
ensure flips are applied on the correct pipes.

Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Implement PSR wait for enable/disable

[Why]
For DMUB implementation of PSR, the 'wait' parameter,
used to determine if driver should wait for PSR enable/disable,
is not implemented correctly.

[How]
Implement wait for PSR enable/disable.

Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: allow DP RX to use more cr aux rd interval delay

[why]
Regression is caused by previous change with attempt to correct the
extended cr aux rd interval delay due to mis interpretation of the DP specs.
The change turns out not working well with certain RXs.
So we decided to keep the cr aux rd interval logic as before.

Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable DP YCbCr420 mode support for DCN10

[WHY]
In DCN10 when a panel with YCbCr420 capability is connected via
USB-C to HDMI active dongle, no YCbCr420 option is listed in
Radeon settings.

[HOW]
Enable DP YCbCr420 mode support for DCN10

Signed-off-by: Gary Li <gary.li@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Increase timeout for DP Disable

[WHY]
When disabling DP video, the current REG_WAIT timeout
of 50ms is too low for certain cases with very high
VSYNC intervals.

[HOW]
Increase the timeout to 102ms, so that
refresh rates as low as 10Hz can be handled properly.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <stable@vger.kernel.org>

drm/amd/display: Fix ODM policy implementation

[WHY]
Only the leftmost ODM pipe should be offset when scaling. A previous
code change was intended to implement this policy, but a section of code
was overlooked.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <stable@vger.kernel.org>

drm/amd/display: eDP intermittent black screen during PnP

[Why]
We dont's turn off backlight before power off eDP (VDD),
which is a violation of eDP specs.

[How]
Power off eDP backlight before power off eDP

Signed-off-by: Peikang Zhang <peikang.zhang@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Bug in dce_is_panel_backlight_on()

[Why]
dce_is_panel_backlight_on() will return wrong value if
LVTMA_BLON_OVRD is 0

[How]
When LVTMA_BLON_OVRD is 0, read
LVTMA_PWRSEQ_TARGET_STATE instead

Signed-off-by: Peikang Zhang <peikang.zhang@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.103

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.33

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix incorrect backlight register offset for DCN

[Why]
Typo in backlight refactor introduced wrong register offset.

[How]
SR(BIOS_SCRATCH_2) to NBIO_SR(BIOS_SCRATCH_2).

Signed-off-by: David Galiffi <David.Galiffi@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: <stable@vger.kernel.org>

drm/amdkfd: Use kvmalloc instead of kmalloc for VCRAT

Since we're dynamically allocating the CPU VCRAT, use kvmalloc in case
the allocation size is huge.

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

drm/amdkfd: Fix kfd init stack dump

amdkfd is dumping a stack during initialization.
kfd_procfs_add_sysfs_stats is being called twice.  This removes one of
them.

Fixes: 4327bed2ff8e3d ("drm/amdkfd: Add process eviction counters to sysfs")
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Philip Cox <Philip.Cox@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix dead lock issue for vblank

Always start vblank timer, but only calls vblank function
when vblank is enabled.

This is used to fix the dead lock issue.
When drm_crtc_vblank_off want to disable vblank,
it first get event_lock, and then call hrtimer_cancel,
but hrtimer_cancel want to wait timer handler function finished.
Timer handler also want to aquire event_lock in drm_handle_vblank.

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

drm/amd/display: fix crash/reboot while accessing sysfs files

read/writes to aux_dpcd_* sysfs entries leads to system
reboot or hang.
Hence fix the handling of input data and reporting of errors
appropriately to the user space.

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

drm/amdkfd: Move process doorbell allocation into kfd device

Move doorbell allocation for a process into kfd device and
allocate doorbell space in each PDD during process creation.
Currently, KFD manages its own doorbell space but for some
devices, amdgpu would allocate the complete doorbell
space instead of leaving a chunk of doorbell space for KFD to
manage. In a system with mix of such devices, KFD would need
to request process doorbell space based on the type of device,
either from amdgpu or from its own doorbell space.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Calculate CPU VCRAT size dynamically (v2)

Instead of guessing at a sufficient size for the CPU VCRAT, base the
size on the number of online NUMA nodes.

v2: fix warning

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

drm/amdgpu: Fix handling of KFD initialization failures

Remember KFD module initializaton status in a global variable. Skip KFD
device probing when the module was not initialized. Other amdgpu_amdkfd
calls are then protected by the adev->kfd.dev check.

Also print a clear error message when KFD disables itself. Amdgpu
continues its initialization even when KFD failed.

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

drm/amd/pm: Skip smu_post_init in SRIOV

smu_post_init needs to enable SMU feature, while this require
virtualization off. Skip it since this feature is not used in SRIOV.

v2: move the check to the early stage of smu_post_init.

v3: fix typo

Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Reviewed-by: Emily.Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: apply dummy reads workaround for CDR enabled only

For CDR disabled case, the dummy reads workaround is not needed.

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/amdgpu: remove experimental flag from navi12

Navi12 has worked fine for a while now.

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

drm/amdkfd: Reduce eviction/restore message levels

Reduce the eviction and restore messages from INFO level to DEBUG level.

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

drm/amdkfd: Add process eviction counters to sysfs

Add per-process eviction counters to sysfs to keep track of
how many eviction events have happened for each process.

v2: rename the stats dir, and track all evictions per process, per device.
v3: Simplify the stats kobject handling and cleanup.
v4: more code cleanup

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

drm/amdkfd: Add some eveiction debugging code

Extending the module parameter debug_evictions to also print a stack
trace when the eviction code path is called.

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

drm/amdgpu/powerplay: hwmgr - modify the return value

Return value should be -EINVAL rather than EINVAL

Fixes: f83a9991648bb("drm/amd/powerplay: add Vega10 powerplay support (v5)")
Fixes: 2cac05dee6e30("drm/amd/powerplay: add the hw manager for vega12 (v4)")
Cc: Eric Huang <JinHuiEric.Huang@amd.com>
Cc: Evan Quan <evan.quan@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Xiaoliang Pang <dawning.pang@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>