drm/amdgpu: Merge get_reserved_allocation to get_vbios_allocations.

Some ASICs need reserved memory for firmware or other components,
which is not allowed to be used by driver.
amdgpu_gmc_get_reserved_allocation is to handle additional areas.
To avoid any missing calling,
merged amdgpu_gmc_get_reserved_allocation to
amdgpu_gmc_get_vbios_allocations.

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

drm/amdkfd: evict svm bo worker handle error

Migrate vram to ram may fail to find the vma if process is exiting
and vma is removed, evict svm bo worker sets prange->svm_bo to NULL
and warn svm_bo ref count != 1 only if migrating vram to ram
successfully.

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

drm/amdgpu/vcn: fix vcn ring test failure in igt reload test

[why]
On Renoir, vcn ring test failed on the second time insmod in the reload
test. After investigation, it proves that vcn only can disable dpg under
dpg unpause mode (dpg unpause mode is default for dec only, dpg pause
mode is for dec/enc).

[how]
unpause dpg in dpg stopping procedure.

Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Tianci Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: only allow secure submission on rings which support that

Only GFX ring, SDMA ring and VCN decode ring support secure submission
at the moment.

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

drm/amdgpu: fixed the warnings reported by kernel test robot

The reported warnings are as follows:
  1.warning:no-previous-prototype-for-amdgpu_hdp_ras_fini.
  2.warning:no-previous-prototype-for-amdgpu_mmhub_ras_fini.

Amdgpu_hdp_ras_fini and amdgpu_mmhub_ras_fini are unused
in the code, they are the only functions in amdgpu_hdp.c
and amdgpu_mmhub.c. After removing these two functions,
both amdgpu_hdp.c and amdgpu_mmhub.c are empty, so these
two files can be deleted to fix the warning.

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

drm/amd/display: 3.2.177

    This version brings along following fixes:
    - Added I2C escape to support query exist
    - Functionality to get pipe from stream context
    - debugfs entry fixes
    - Block zstate when more than one plane enabled
    - Update to engine ddc
    - Reading lante status during LT fixes
    - Null point filters
    - [FW Promotion] Release 0.0.108.0

Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.108.0

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@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 save/restore PANEL_PWRSEQ_REF_DIV2

[why]
DCN31 has this in zstate save/restore sequence.
need for non_zstate supported ASIC

[how]
add this PANEL_PWRSEQ_REF_DIV2 to existing panel_cntl_hw_init structure.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Wait for hubp read line for Pollock

[Why]
Underflow occurred while hubp ret pipe read is idle and the
second pipe is powered up and added. Flickering and underflow
are only observed on Pollock.

[How]
Check the hubp ret pipe read prior to unlock pipes.

Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Becle Lee <becle.lee@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add null pointer filter

[Why & How]
Add null pointer filter for logical integrity.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Jing Zhou <Jing.Zhou@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Retry when fail reading lane status during LT

[Why]
In dp_get_lane_status_and_lane_adjust(), we call core_link_read_dpcd()
to get lane status but we ignore the return value of the function. If
there is problem with AUX and we eventually can only get partial data
from monitor, we might misuse the data to jude the link training result.
Which could have us trapped in the dc_link_dp_perform_link_training()
since we thought that we can still give a try on another link setting.

[How]
In dp_get_lane_status_and_lane_adjust(), take return value of
core_link_read_dpcd() into account. If we fail to read all requested
data, we treat lanes status as previous one and repeat LT steps again.

Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@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: Add new enum for EDID status

[Why]
We must support a new type of partial edid return in the future

[How]
Add interface and case hander for partial edid

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Dale Zhao <dale.zhao@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Block zstate when more than one plane enabled

[Why]
Currently Z10 save/restore only covers 1 plane case. If we enter
Z10 with 2 planes enabled, 1 of the 2 planes will not be restored.
causing black screen.

[How]
Disable Zstate when more than 1 plane is enabled. Right now Z9 enable
but Z10 disable is not a supported config, so we disable both.

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update engine ddc

[Why]
Crash caused by a ddc update failure

[How]
Update engine ddc before release engine.

Reviewed-by: Wyatt Wood <Wyatt.Wood@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Jing Zhou <Jing.Zhou@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix a few parts in debugfs entry

[Why & How]
1. To leverage and simplify the igt tool. Change the value of
retrain when calling preferred training settings.
Now, writing value to link settings triggers link training immediately.

2. Add dc lock when triggering link training in edp_ilr_write()

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@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: Add function to get the pipe from the stream context

[Why]
We need a helper function in dc to grab the pipe from the stream
context.

[How]
Add it.

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@amd.com>
Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add I2C escape to support query device exist.

[How]
1. Search OEM I2C info from BIOS and compare with input parameter.
2. If BIOS doesn't record it, just try to read one byte.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Agustin Gutierrez <agustin.gutierrez@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>

drm/amdgpu: Move reset domain init before calling RREG32

amdgpu_detect_virtualization reads register, amdgpu_device_rreg access
adev->reset_domain->sem if kernel defined CONFIG_LOCKDEP, below is the
random boot hang backtrace on Vega10. It may get random NULL pointer
access backtrace if amdgpu_sriov_runtime is true too.

Move amdgpu_reset_create_reset_domain before calling to RREG32.

 BUG: kernel NULL pointer dereference, address:
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: 0000 [#1] PREEMPT SMP NOPTI
 Workqueue: events work_for_cpu_fn
 RIP: 0010:down_read_trylock+0x13/0xf0
 Call Trace:
  <TASK>
  amdgpu_device_skip_hw_access+0x38/0x80 [amdgpu]
  amdgpu_device_rreg+0x1b/0x170 [amdgpu]
  amdgpu_detect_virtualization+0x73/0x100 [amdgpu]
  amdgpu_device_init.cold.60+0xbe/0x16b1 [amdgpu]
  ? pci_bus_read_config_word+0x43/0x70
  amdgpu_driver_load_kms+0x15/0x120 [amdgpu]
  amdgpu_pci_probe+0x1a1/0x3a0 [amdgpu]

Fixes: d0fb18b535679a ("drm/amdgpu: Move reset sem into reset_domain")
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: fix gfx hang on renoir in IGT reload test

[why]
CP hangs in igt reloading test on renoir, more precisely, hangs on the
second time insmod.

[how]
mode2 reset can make it recover, and mode2 reset only effects gfx core,
dcn and the screen will not be impacted.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Tianci.Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: only check for _PR3 on dGPUs

We don't support runtime pm on APUs.  They support more
dynamic power savings using clock and powergating.

Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Tested-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: use pm mutex to protect ecc info table

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

drm/amdgpu: drop xmgi23 error query/reset support

xgmi_ras is only initialized when host to GPU interface
is PCIE. in such case, xgmi23 is disabled and protected
by security firmware. Host access will results to
security violation

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

drm/amdgpu: fix aldebaran xgmi topology for vf

VFs must also distinguish whether or not the TA supports full duplex or
half duplex link records in order to report the correct xGMI topology.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Shaoyun Liu <shaoyun.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: enable scatter/gather display for DCN 3.1.6

Enable scatter/gather display for DCN 3.1.6.

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

drm/amd/display: Add pstate verification and recovery for DCN31

[Why]
To debug when p-state is being blocked and avoid PMFW hangs when
it does occur.

[How]
Re-use the DCN10 hardware sequencer by adding a new interface for
verifying p-state high on the hubbub. The interface is mostly the
same as the DCN10 interface, but the bit definitions have changed for
the debug bus.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: CRIU export dmabuf handles for GTT BOs

Export dmabuf handles for GTT BOs so that their contents can be accessed
using SDMA during checkpoint/restore.

v2: Squash in fix from David to set dmabuf handle to invalid for BOs
that cannot be accessed using SDMA during checkpoint/restore.

Signed-off-by: David Yat Sin <david.yatsin@amd.com>
Reviewed-by : Rajneesh Bhardwaj <rajneesh.bhardwaj@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: CRIU Refactor restore BO function

Refactor CRIU restore BO to reduce identation.

Signed-off-by: David Yat Sin <david.yatsin@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: CRIU remove sync and TLB flush on restore

When the process is getting restored, the queues are not mapped yet, so
there is no VMID assigned for this process and no TLBs to flush.

Signed-off-by: David Yat Sin <david.yatsin@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dp kvm can't light up

[why]
The DP KVM failed to light up, since the lttpr_mode is not reset to
default value when failed to read LTTPR capabilities, and the
variable max_link_rate retains a initial value zero, this cause variable
link_rate be assigned to an error value zero, consquently pixel_clock
get wrong value zero, and kvm can't light up.

[how]
Reset lttpr_mode to default value when failed to read LTTPR
capabilities, so that the link_rate fallbacks to the minimum rate that
supported by link encoder and sink.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Tianci.Yin <tianci.yin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Send message when resp status is 0xFC

When PMFW is really busy, it will respond with 0xFC. However, it doesn't
change the response register state when it becomes free. Driver should
retry and proceed to send message if the response status is 0xFC.

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

drm/amdgpu: message smu to update bad channel info

It should notice SMU to update bad channel info when detected
uncorrectable error in UMC block

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

drm/amd/pm: add send bad channel info function

support message SMU update bad channel info to update HBM bad channel
info in OOB table

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

drm/amd/display: move FPU code from dcn10 to dml/dcn10 folder

FPU operations in dcn10 was already moved to dml folder via calcs code.
However, dcn1_0_ip and dcn_1_0_soc with FPU componentd remains on dcn10.
Following previous changes to isolate FPU, this patch creates dcn10_fpu
files to isolate FPU-specific code and moves those structs to it.

Signed-off-by: Melissa Wen <mwen@igalia.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: move FPU operations from dcn21 to dml/dcn20 folder

dml/dcn20_fpu file centralizes all DCN2x functions that require FPU access.
Therefore, this patch moves FPU-related code from dcn21 to dcn20_fpu. These
include:
- dcn21_populate_dml_pipes_from_context()
- dcn21_validate_bandwidth_fp() and related:
  - dcn21_calculate_wm(),
  - patch_bounding_box(),
  - calculate_wm_set_for_vlevel()
- renaming update_bw_bounding_box() to dcn21_update_bw_bounding_box(), move
to dcn20_fpu with related static function construct_low_pstate_lvl()

Also, make dcn21_fast_validate_bw() public in dcn21_resource as it is called
by dcn21_validate_bandwidth_fp() now in dcn20_fpu.

Reuse dcn20_fpu_adjust_dppclk() in dcn21_fast_validate_bw() as it isolates
the same FPU operation.

Include dchubbub.h as it is required in dcn21_populate_dml_pipes_from_context()

Signed-off-by: Melissa Wen <mwen@igalia.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: move FPU-related code from dcn20 to dml folder

Move parts of dcn20 code that uses FPU to dml folder. It aims to isolate
FPU operations as described by series:

drm/amd/display: Introduce FPU directory inside DC
https://patchwork.freedesktop.org/series/93042/

This patch moves the following functions from dcn20_resource to
dml/dcn20_fpu and calls of public functions in dcn20_resource are
wrapped by DC_FP_START/END():

- void dcn20_populate_dml_writeback_from_context
- static bool is_dtbclk_required()
- static enum dcn_zstate_support_state()
- void dcn20_calculate_dlg_params()
- static void swizzle_to_dml_params()
- int dcn20_populate_dml_pipes_from_context()
- void dcn20_calculate_wm()
- void dcn20_cap_soc_clocks()
- void dcn20_update_bounding_box()
- void dcn20_patch_bounding_box()
- bool dcn20_validate_bandwidth_fp()

This movement also affects dcn21/30/31, as dcn20_calculate_dlg_params()
is used by them. For this reason, I included dcn20_fpu headers in
dcn20_resource headers to make dcn20_calculate_dlg_params() visible to
dcn21/30/31.

Three new functions are created to isolate well-delimited FPU
operations:

- void dcn20_fpu_set_wb_arb_params(): set cli_watermark,
  pstate_watermark and time_per_pixel from wb_arb_params (struct
mcif_arb_params), since those uses FPU operations on double types:
WritebackUrgentWatermark, WritebackDRAMClockChangeWatermark, '16.0'.
- void dcn20_fpu_set_wm_ranges(): set min_fill_clk_mhz and
  max_fill_clk_mhz involves FPU calcs on dram_speed_mts (double type);
- void dcn20_fpu_adjust_dppclk(): adjust operation on RequiredDPPCLK
  that is a double.

Signed-off-by: Melissa Wen <mwen@igalia.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix compile error from TO_CLK_MGR_INTERNAL

[Why]

On LNX, TO_CLK_MGR_INTERNAL() only works to get clk_mgr_internal from a
given clk_mgr. In clk_mgr_dcn316 struct, the clk_mgr_internal is already
a memeber by the alias 'base'

[How]

Use &clk_mgr->base instead.

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix double free during GPU reset on DC streams

[Why]
The issue only occurs during the GPU reset code path.

We first backup the current state prior to commiting 0 streams
internally from DM to DC. This state backup contains valid link
encoder assignments.

DC will clear the link encoder assignments as part of current state
(but not the backup, since it was a copied before the commit) and
free the extra stream reference it held.

DC requires that the link encoder assignments remain cleared/invalid
prior to commiting. Since the backup still has valid assignments we
call the interface post reset to clear them. This routine also
releases the extra reference that the link encoder interface held -
resulting in a double free (and eventually a NULL pointer dereference).

[How]
We'll have to do a full DC commit anyway after GPU reset because
the stream count previously went to 0.

We don't need to retain the assignment that we had backed up, so
just copy off of the now clean current state assignment after the
reset has occcurred with the new link_enc_cfg_copy() interface.

Fixes: 6d63fcc2a334 ("drm/amd/display: Reset link encoder assignments for GPU reset")

Reviewed-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add NULL check

[Why]
Virtualization enters blue screen of death (BSoD)
due to NULL res_pool object when accessing DSC
encoder capability.

[How]
Add NULL check to avoid blue screen of death.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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: 3.2.176

drm/amd/display: 3.2.176

    This version brings along following fixes:
    - move FPU operations from dcn21 to dml/dcn20 folder
    - move FPU-related code from dcn20 to dml folder
    - Fix compile error from TO_CLK_MGR_INTERNAL
    - Fix double free during GPU reset on DC streams
    - Add NULL check
    - [FW Promotion] Release 0.0.107.0
    - enable dcn315/316 s0i2 support
    - handle DP2.0 RX with UHBR20 but not UHBR13.5 support
    - disable HPD SW timer for passive dongle type 1 only
    - add gamut coefficient set A and B
    - merge two duplicated clock_source_create
    - Add link dp trace support
    - move FPU associated DCN303 code to DML folder
    - Release AUX engine after failed acquire
    - Add minimal pipe split transition state
    - Clean up fixed VS PHY test w/a function
    - fix the clock source contruct for dcn315
    - cleaning up smu_if to add future flexibility
    - fix deep color ratio
    - add debug option to bypass ssinfo from bios for dcn315

Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.107.0

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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: enable dcn315/316 s0i2 support

[why]
enable the support in driver, let the control by CMOS

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: handle DP2.0 RX with UHBR20 but not UHBR13.5 support

[why]
DP2.0 defines DP2.0 RX capability using bit maps. This means there is
a corner case the DP2.0 RX can declare UHBR20 capability but not UHBR13.5 support.
In this case we should skip UHBR13.5 selection and continue to our next option
in decide liink settings.

Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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 HPD SW timer for passive dongle type 1 only

[why]
Need to provide this workaround
only for type 1 passive dongle

[how]
Detect if dongle is type 1 or 2.
And use it to determine if w/a is needed.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Sung Joon Kim <sungjoon.kim@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add gamut coefficient set A and B

[Why]
Screen shake on DCN3 night light SDR BR3.

[How]
Change the logic to use double buffer reisgter on gamut settings.

Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Jingwen Zhu <jingwen.zhu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: merge two duplicated clock_source_create

[why]
dcn31x could use dcn31 sepcific which contains deep_color_ratio for dmub

Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add link dp trace support

[Why]
It is nice to have link training and link loss status logging in
our trace, and this patch add the all necessary accessors in dc
side.

[How]
1. Define dp_trace struct and create new file link_dp_trace.c and
link_dp_trace.h;
2. Add dp link training status update in dc_link_dp.c;

Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: move FPU associated DCN303 code to DML folder

[Why & How]
As part of the FPU isolation work documented in
https://patchwork.freedesktop.org/series/93042/, isolate
code that uses FPU in DCN303 to DML, where all FPU code
should locate.

Co-authored-by: Jasdeep Dhillon <jasdeep.dhillon@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Jasdeep Dhillon <jasdeep.dhillon@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Release AUX engine after failed acquire

[Why]
There is a sequence in which aux arbitration
doesn't work correctly. Driver is left with
aux access after it times out waiting for access.
In future dmub fw is never granted aux access
and is stuck in a while loop.

[How]
Cancel aux request from driver after timing out.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add minimal pipe split transition state

[WHY?] When adding/removng a plane to some configurations,
unsupported pipe programming can occur when moving to a new plane.

[HOW?]
Add a safe transistion state before programming new configuration.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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: Clean up fixed VS PHY test w/a function

[Why]
Unify naming for fixed VS workarounds.

[How]
Rename function to match naming convention.
Refactor code to remove unused function.

Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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: fix the clock source contruct for dcn315

[why]
Driver sends invalid deep color ratio to DMUB. Update it to the right
one by using another clock source construct.

Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: cleaning up smu_if to add future flexibility

why and how:
need to hook in new smu interfaces
cleaning up code that used old variables

Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Leung, Martin <Martin.Leung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix deep color ratio

Fix enum mapping for deep color ratio

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Hansen Dsouza <Hansen.Dsouza@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add debug option to bypass ssinfo from bios for dcn315

[Why & How]
Add debug option to bypass ssinfo from BIOS for dcn315.

Reviewed-by: Park, Chris <Chris.Park@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: change registers in error checking for smu 13.0.5

smu 13.0.5 use new registers for smu msg and param.

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

drm/amdgpu: Disable baco dummy mode

On aldebaran, BACO dummy mode may be enabled during reset. Disable
it during resume.

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

Merge tag 'drm-intel-next-fixes-2022-03-10' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

- Reduce overzealous alignment constraints for GGTT
- Add missing mdev attribute "name" for GVT
- Async flip fixes (Ville)
- Static checker fix (Ville)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YimvoLOZ8RPKrITv@jlahtine-mobl.ger.corp.intel.com

Merge tag 'amd-drm-next-5.18-2022-03-09' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-5.18-2022-03-09:

amdgpu:
- Misc code cleanups
- Misc display fixes
- PSR display fixes
- More RAS cleanup
- Hotplug fix
- Bump minor version for hotplug tests
- SR-IOV fixes
- GC 10.3.7 updates
- Remove some firmwares which are no longer used
- Mode2 reset refactor
- Aldebaran fixes
- Add VCN fwlog feature for VCN debugging
- CS code cleanup
- Fix clang warning
- Fix CS clean up rebase breakage

amdkfd:
- SVM fixes
- SMI event fixes and cleanups
- vmid_pasid mapping fix for gfx10.3

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220309224439.2178877-1-alexander.deucher@amd.com

Merge tag 'drm-msm-next-2022-03-08' of https://gitlab.freedesktop.org/drm/msm into drm-next

Follow-up pull req for v5.18 to pull in some important fixes.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGvwHFHEd+9df-0aBOCfmw+ULvTS3f18sJuq_cvGKLDSjw@mail.gmail.com

drm/amdgpu: fix a wrong ib reference

It should be p->job->ibs[j] instead of p->job->ibs[i] here.

Fixes: cdc7893fc93f19 ("drm/amdgpu: use job and ib structures directly in CS parsers")
Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: bail out early if no get_atc_vmid_pasid_mapping_info

it makes no sense to continue with an undefined vmid.

Fixes: c8b0507f40deea ("drm/amdkfd: judge get_atc_vmid_pasid_mapping_info before call")
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reported-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'exynos-drm-next-v5.18' of git://git./linux/kernel/git/daeinki/drm-exynos into drm-next

New feature
- Add BGR pixel format support for FIMD device. As for this,
  this patch uses undocumented register, WIN_RGB_ORDER, but
  it is safe because product kernels have been using same
  register.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Inki Dae <inki.dae@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220304085220.324245-1-inki.dae@samsung.com

drm/i915/gtt: reduce overzealous alignment constraints for GGTT

Currently this will enforce both 2M alignment and padding for any LMEM
pages inserted into the GGTT. However, this was only meant to be applied
to the compact-pt layout with the ppGTT. For the GGTT we can reduce the
alignment and padding to 64K.

Bspec: 45015
Fixes: 87bd701ee268 ("drm/i915: enforce min GTT alignment for discrete cards")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Robert Beckett <bob.beckett@collabora.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220303100229.839282-1-matthew.auld@intel.com
(cherry picked from commit c64fa77dd4609cb8cd53fbb73b02434ae8212c7a)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/msm/gpu: Fix crash on devices without devfreq support (v2)

Avoid going down devfreq paths on devices where devfreq is not
initialized.

v2: Change has_devfreq() logic [Dmitry]

Reported-by: Linux Kernel Functional Testing <lkft@linaro.org>
Reported-by: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Fixes: 6aa89ae1fb04 ("drm/msm/gpu: Cancel idle/boost work on suspend")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220308184844.1121029-1-robdclark@gmail.com

drm/msm/adreno: fix cast in adreno_get_param()

These casts need to happen before the shift.  The only time it would
matter would be if "rev.core" is >= 128.  In that case the sign bit
would be extended and we do not want that.

Fixes: afab9d91d872 ("drm/msm/adreno: Expose speedbin to userspace")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>
Link: https://lore.kernel.org/r/20220307133105.GA17534@kili
Signed-off-by: Rob Clark <robdclark@chromium.org>

Merge tag 'gvt-next-2022-03-07' of https://github.com/intel/gvt-linux into drm-intel-next-fixes

- add the missing attribute "name" in VFIO mdev hierarchy.

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
From: "Wang, Zhi A" <zhi.a.wang@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/e38f0725-6b22-8e49-b8f6-41986b0ece06@intel.com

drm/i915: Fix the async flip wm0/ddb optimization

The current implementation of the async flip wm0/ddb optimization
does not work at all. The biggest problem is that we skip the
whole intel_pipe_update_{start,end}() dance and thus never actually
complete the commit that is trying to do the wm/ddb change.

To fix this we need to move the do_async_flip flag to the crtc
state since we handle commits per-pipe, not per-plane.

Also since all planes can now be included in the first/last
"async flip" (which gets converted to a sync flip to do the
wm/ddb mangling) we need to be more careful when checking if
the plane state is async flip comptatible. Only planes doing
the async flip should be checked and other planes are perfectly
fine not adhereing to any async flip related limitations.

However for subsequent commits which are actually going do the
async flip in hardware we want to make sure no other planes
are in the state. That should never happen assuming we did our
job correctly, so we'll toss in a WARN to make sure we catch
any bugs here.

Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Fixes: c3639f3be480 ("drm/i915: Use wm0 only during async flips for DG2")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214105532.13049-4-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
(cherry picked from commit 2e08437160d1e8f2cd3f0d56d59e74423602116e)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Check async flip capability early on

Since the async flip state check is done very late and
thus it can see potentially all the planes in the state
(due to the wm/ddb optimization) we need to move the
"can the requested plane do async flips at all?" check
much earlier. For this purpose we introduce
intel_async_flip_check_uapi() that gets called early during
the atomic check.

And for good measure we'll throw in a couple of basic checks:
- is the crtc active?
- was a modeset flagged?
- is+was the plane enabled?
Though atm all of those should be guaranteed by the fact
that the async flip can only be requested through the legacy
page flip ioctl.

Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Fixes: c3639f3be480 ("drm/i915: Use wm0 only during async flips for DG2")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214105532.13049-3-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
(cherry picked from commit b0b2bed2a1305c8f977c6b7d5fa162773693a212)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Don't skip ddb allocation if data_rate==0

data_rate==0 no longer means a plane is disabled, it could
also mean we want to use the minimum ddb allocation for it.
Hence we can't bail out early during ddb allocation or
else we'll simply forget to allocate any ddb for such planes.

Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Fixes: 6a4d8cc6bbbf ("drm/i915: Don't allocate extra ddb during async flip for DG2")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214105532.13049-2-ville.syrjala@linux.intel.com
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
(cherry picked from commit 6475e106821babc0dc478a9cb3fc3973739c43fb)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Avoid negative shift due to bigjoiner_pipes==0

bigjoiner_pipes==0 leads bigjoiner_master_pipe() to
do BIT(ffs(0)-1) which is undefined behaviour. The code should
actually still work fine since the only place we provoke
that is intel_crtc_bigjoiner_slave_pipes() and it'll bitwise
AND the result with 0, so doesn't really matter what we get
out of bigjoiner_master_pipe(). But best not provoke undefined
behaviour anyway.

Reported-by: kernel test robot <oliver.sang@intel.com>
Fixes: a6e7a006f5d5 ("drm/i915: Change bigjoiner state tracking to use the pipe bitmask")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220223131315.18016-2-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>
(cherry picked from commit cccc71b552a1040ad3d738d7ec95570801fb0bf6)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/amdkfd: Add format attribute to kfd_smi_event_add

To enable compiler type-checked against the format string in callers.

All warnings (new ones prefixed by >>):

>> warning: function 'kfd_smi_event_add' might be a candidate for
'gnu_printf' format attribute [-Wsuggest-attribute=format]

Fixes: d58b8a99cbb8 ("drm/amdkfd: Add SMI add event helper")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/i915/gvt: add the missing mdev attribute "name"

The mdev attribute "name" is required by some middle software, e.g.
KubeVirt, an open source SW that manages VM on Kubernetes cluster uses
the mdev sysfs directory/file structure to discover mediated device in
nodes in the cluster.

v2:

- Fix the missing defination in gvt_type_attrs. (Zhenyu)

Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: Hui Chun Ong <hui.chun.ong@intel.com>
Cc: Terrence Xu <terrence.xu@intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Signed-off-by: Zhi Wang <zhi.wang.linux@gmail.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220222150532.9090-1-zhi.a.wang@intel.com
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>

drm/msm: Fix dirtyfb refcounting

refcount_t complains about 0->1 transitions, which isn't *quite* what we
wanted.  So use dirtyfb==1 to mean that the fb is not connected to any
output that requires dirtyfb flushing, so that we can keep the underflow
and overflow checking.

Fixes: 9e4dde28e9cd ("drm/msm: Avoid dirtyfb stalls on video mode displays (v2)")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220304202146.845566-1-robdclark@gmail.com

drm/msm/a6xx: Fix missing ARRAY_SIZE() check

Fixes: f6d62d091cfd ("drm/msm/a6xx: add support for Adreno 660 GPU")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220305173405.914989-1-robdclark@gmail.com

drm/msm: Add MSM_SUBMIT_FENCE_SN_IN

Add a way for userspace to specify the sequence number fence used to
track completion of the submit.  As the seqno fence is simply an
incrementing counter which is local to the submitqueue, it is easy for
userspace to know the next value.

This is useful for native userspace drivers in a vm guest, as the guest
to host roundtrip can have high latency.  Assigning the fence seqno in
the guest userspace allows the guest to continue without waiting for
response from the host.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220224222321.60653-1-robdclark@gmail.com

drm/msm/a6xx: Zap counters across context switch

Any app controlled perfcntr collection (GL_AMD_performance_monitor, etc)
does not require counters to maintain state across context switches.  So
clear them if systemwide profiling is not active.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220304005317.776110-5-robdclark@gmail.com

drm/msm: Add SYSPROF param (v2)

Add a SYSPROF param for system profiling tools like Mesa's pps-producer
(perfetto) to control behavior related to system-wide performance
counter collection.  In particular, for profiling, one wants to ensure
that GPU context switches do not effect perfcounter state, and might
want to suppress suspend (which would cause counters to lose state).

v2: Swap the order in msm_file_private_set_sysprof() [sboyd] and
    initialize the sysprof_active refcount to one (because the under/
    overflow checking in refcount_t doesn't expect a 0->1 transition)
    meaning that values greater than 1 means sysprof is active.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220304005317.776110-4-robdclark@gmail.com

drm/msm: Add SET_PARAM ioctl

It was always expected to have a use for this some day, so we left a
placeholder.  Now we do.  (And I expect another use in the not too
distant future when we start allowing userspace to allocate GPU iova.)

Signed-off-by: Rob Clark <robdclark@chromium.org>
Link: https://lore.kernel.org/r/20220304005317.776110-3-robdclark@gmail.com

drm/msm: Update generated headers

Update headers from mesa commit:

  commit 7e63fa2bb13cf14b765ad06d046789ee1879b5ef
  Author:     Rob Clark <robclark@freedesktop.org>
  AuthorDate: Wed Mar 2 17:11:10 2022 -0800

      freedreno/registers: Add a couple regs we need for kernel

Signed-off-by: Rob Clark <robdclark@chromium.org>
      Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15221>

Signed-off-by: Rob Clark <robdclark@chromium.org>
[for display bits:]
Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
Link: https://lore.kernel.org/r/20220304005317.776110-2-robdclark@gmail.com

drm/amdgpu: initialize the vmid_wait with the stub fence

This way we don't need to check for NULL any more.

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

drm/amdgpu: properly embed the IBs into the job

We now have standard macros for that.

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

drm/amdgpu: use job and ib structures directly in CS parsers

Instead of providing the ib index provide the job and ib pointers directly to
the patch and parse functions for UVD and VCE.

Also move the set/get functions for IB values to the IB declerations.

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

drm/amdgpu: header cleanup

No function change, just move a bunch of definitions from amdgpu.h into
separate header files.

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

drm/amd/amdgpu: set disabled vcn to no_schduler

[Why]
after the reset domain introduced, the sched.ready will be init after
hw_init, which will overwrite the setup in vcn hw_init, and lead to
vcn ib test fail.

[How]
set disabled vcn to no_scheduler

Fixes: 5fd8518d187ed0 ("drm/amdgpu: Move scheduler init to after XGMI is ready")
Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Acked-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: install ctx entities with cmpxchg

Since we removed the context lock we need to make sure that not two threads
are trying to install an entity at the same time.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 461fa7b0ac565e ("drm/amdgpu: remove ctx->lock")
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: implement get_atc_vmid_pasid_mapping_info for gfx10.3

This patch implements get_atc_vmid_pasid_mapping_info for gfx10.3

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

drm/amdkfd: judge get_atc_vmid_pasid_mapping_info before call

Fix the NULL point issue:

[ 3076.255609] BUG: kernel NULL pointer dereference, address: 0000000000000000
[ 3076.255624] #PF: supervisor instruction fetch in kernel mode
[ 3076.255637] #PF: error_code(0x0010) - not-present page
[ 3076.255649] PGD 0 P4D 0
[ 3076.255660] Oops: 0010 [#1] SMP NOPTI
[ 3076.255669] CPU: 20 PID: 2415 Comm: kfdtest Tainted: G        W  OE     5.11.0-41-generic #45~20.04.1-Ubuntu
[ 3076.255691] Hardware name: AMD Splinter/Splinter-RPL, BIOS VS2326337N.FD 02/07/2022
[ 3076.255706] RIP: 0010:0x0
[ 3076.255718] Code: Unable to access opcode bytes at RIP 0xffffffffffffffd6.
[ 3076.255732] RSP: 0018:ffffb64283e3fc10 EFLAGS: 00010297
[ 3076.255744] RAX: 0000000000000000 RBX: 0000000000000008 RCX: 0000000000000027
[ 3076.255759] RDX: ffffb64283e3fc1e RSI: 0000000000000008 RDI: ffff8c7a87f60000
[ 3076.255776] RBP: ffffb64283e3fc78 R08: ffff8c7d88518ac0 R09: ffffb64283e3fa60
[ 3076.255791] R10: 0000000000000001 R11: 0000000000000001 R12: 000000000000000f
[ 3076.255805] R13: ffff8c7bdcea5800 R14: ffff8c7a9f3f3000 R15: ffff8c7a8696bc00
[ 3076.255820] FS:  0000000000000000(0000) GS:ffff8c7d88500000(0000) knlGS:0000000000000000
[ 3076.255839] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 3076.255851] CR2: ffffffffffffffd6 CR3: 0000000109e3c000 CR4: 0000000000750ee0
[ 3076.255866] PKRU: 55555554
[ 3076.255873] Call Trace:
[ 3076.255884]  dbgdev_wave_reset_wavefronts+0x72/0x160 [amdgpu]
[ 3076.256025]  process_termination_cpsch.cold+0x26/0x2f [amdgpu]
[ 3076.256182]  ? ktime_get_mono_fast_ns+0x4e/0xa0
[ 3076.256196]  kfd_process_dequeue_from_all_devices+0x49/0x70 [amdgpu]
[ 3076.256328]  kfd_process_notifier_release+0x187/0x2b0 [amdgpu]
[ 3076.256451]  ? mn_itree_inv_end+0xdc/0x110
[ 3076.256463]  __mmu_notifier_release+0x74/0x1f0
[ 3076.256474]  exit_mmap+0x170/0x200
[ 3076.256484]  ? __handle_mm_fault+0x677/0x920
[ 3076.256496]  ? _cond_resched+0x19/0x30
[ 3076.256507]  mmput+0x5d/0x130
[ 3076.256518]  do_exit+0x332/0xaf0
[ 3076.256526]  ? handle_mm_fault+0xd7/0x2b0
[ 3076.256537]  do_group_exit+0x43/0xa0
[ 3076.256548]  __x64_sys_exit_group+0x18/0x20
[ 3076.256559]  do_syscall_64+0x38/0x90
[ 3076.256569]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

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

drm/amdgpu/vcn: Add vcn firmware log

vcn fwlog is for debugging purpose only,
by default, it is disabled.

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

drm/amdgpu/vcn: Update fw shared data structure

Add fw log in fw shared data structure.

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

drm/amdgpu: Add DFC CAP support for aldebaran

Add DFC CAP support for aldebaran

Initialize cap microcode in psp_init_sriov_microcode,
the ta microcode will be  initialized in psp_vxx_init_microcode

Signed-off-by: David Yu <David.Yu@amd.com>
Reviewed-by: Shaoyun.liu <Shaoyun.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Set correct DMA mask for aldebaran

Aldebaran has 48-bit physical address support

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

drm/amdgpu: Refactor mode2 reset logic for v13.0.2

Use IP version and refactor reset logic to apply to a list of devices.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <Le.Ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.175

drm/amd/display: 3.2.175

    This version brings along following fixes:
    - Remove invalid RDPCS Programming in DAL
    - Make functional resource functions non-static
    - Reset VIC if HDMI_VIC is present
    - Add frame alternate 3D & restrict HW packed on dongles
    - Reg to turn on/off PSR Power seq in FSM
    - Modify plane removal sequence to avoid hangs
    - Pass HostVM enable flag into DCN3.1 DML
    - DC Validation failures
    - Program OPP before ODM
    - Refactor fixed VS w/a for PHY tests
    - Pass deep sleep disabled allow info to dmub fw
    - Refine the EDID override
    - [FW Promotion] Release 0.0.106.0
    - Add verify_link_cap back for hdmi

Acked-by: Alan Liu <HaoPing.Liu@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: add verify_link_cap back for hdmi

[why]
hdmi specific: add verify link cap after retrive link cap.

Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Charlene Liu <Charlene.Liu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.106.0

Acked-by: Alan Liu <HaoPing.Liu@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: refine the EDID override

[Why]
We already get the SBIOS EDID via ACPI on KMD,
but after that, we just use the monitor EDID to set it HDR caps

[How]
Make the SBIOS EDID override to read_edid()
That can change the read EDID caps from the right EDID

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: jinzh <jinzh@github.amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Pass deep sleep disabled allow info to dmub fw

[Why]
The deep sleep mode need to be disabled in some PSR scenario.

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Robin Chen <robin.chen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Refactor fixed VS w/a for PHY tests

[Why/How]
Refactor original w/a to unify naming and
simplify logic. This also re-enables the code
that was previously skipped due to the
disabling of the previous workaround logic.

Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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: Program OPP before ODM

[WHY]
HW expects OPP to be configured before ODM is enabled.
Failure to do so can cause errors.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@amd.com>
Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Pass HostVM enable flag into DCN3.1 DML

[WHY]
Calculations differ with HostVM enabled/disabled, causing underflow in
configs with high refresh displays + scaling due to lower available BW

[HOW]
Check riommu_active in order to pass correct HostVM enablement to DML

Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Alan Liu <HaoPing.Liu@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>