drm/amd/display: Fix DCN32 DPSTREAMCLK_CNTL programming

[Why]
Each index in the DPSTREAMCLK_CNTL register
phyiscally maps 1-to-1 with HPO stream encoder
instance. On the other hand, each index in
DTBCLK_P_CNTL physically maps 1-to-1 with OTG
instance.

Current DCN32 DPSTREAMCLK_CLK programing assumes
that OTG instance always maps 1-to-1 with
HPO stream encoder instance. This is not always
guaranteed and can result in blackscreen.

[How]
Program the correct dpstreamclk instance with
the correct dtbclk_p source.

Reviewed-by: Ariel Bernstein <Eric.Bernstein@amd.com>
Acked-by: Brian Chang <Brian.Chang@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: Don't choose SubVP display if ActiveMargin > 0

[Description]
There can be SubVP scheduling issues if a SubVP
display is chosen has ActiveDramClockChangeLatency > 0.
Block this case for now, and enable Vactive case
(later) to handle this.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set ODM policy based on number of DSC slices

[Why & How]
Add addtional check in CalculateODMMode for cases where the ODM combine
is needed due to number of DSC slices.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Taimur Hassan <Syed.Hassan@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.200

This version brings along following fixes:
- Modify pipe split policy
- Fix odm 2:1 policy in 4k144 mode
- Correct HDMI ODM combine policy
- Change AUX NACK behavior
- Change runtime initialization for DCN32/321
- Fix cursor flicker in PSRSU
- Fix k1/k2 divider for virtual signal for DCN32
- Free phantom plane after removing the context
- Add interface to track PHY state
- Add SubVP scaling case
- Add log clock table for SMU
- Fix atomic_check check
- Fix SMU 13.0.0 driver_if header
- Fix doorbells allocation

Acked-by: Brian Chang <Brian.Chang@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/amdgpu: add missing pci_disable_device() in amdgpu_pmops_runtime_resume()

Add missing pci_disable_device() if amdgpu_device_resume() fails.

Fixes: 8e4d5d43cc6c ("drm/amdgpu: Handling of amdgpu_device_resume return value for graceful teardown")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix variable dereferenced before check

Fixes the following smatch warning:

drivers/gpu/drm/amd/amdgpu/../display/dc/dc_dmub_srv.c:311 dc_dmub_srv_p_state_delegate()
warn: variable dereferenced before check 'dc' (see line 309)

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: sunliming <sunliming@kylinos.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove the unneeded result variable

Return the value sdma_v5_2_start() directly instead of storing it in
another redundant variable.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Update mes_v11_api_def.h

New GFX11 MES FW adds the trap_en bit. For now hardcode to 1 (traps
enabled).

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

drm/amdgpu: disable FRU access on special SIENNA CICHLID card

Below driver load error will be printed, not friendly to end user.

amdgpu: ATOM BIOS: 113-D603GLXE-077
[drm] FRU: Failed to get size field
[drm:amdgpu_fru_get_product_info [amdgpu]] *ERROR* Failed to read FRU Manufacturer, ret:-5

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

drm/amd: remove possible condition with no effect (if == else)

This patch fix cocci warning:
drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c:1816:6-8:
WARNING: possible condition with no effect (if == else).

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

drm/amd: remove possible condition with no effect (if == else)

This patch fix cocci warning:
drivers/gpu/drm/amd/display/dc/core/dc.c:3335:2-4: WARNING:
possible condition with no effect (if == else).

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

drm/amd: fix potential memory leak

This patch fix potential memory leak (clk_src) when function run
into last return NULL.

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

drm/amd: fix potential memory leak

This patch fix potential memory leak (clk_src) when function run
into last return NULL.

s/free/kfree/ - Alex

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

drm/amdgpu: mmVM_L2_CNTL3 register not initialized correctly

The mmVM_L2_CNTL3 register is not assigned an initial value

Signed-off-by: Qu Huang <jinsdb@126.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm: amd: amdgpu: ACPI: Add comment about ACPI_FADT_LOW_POWER_S0

According to the ACPI specification [1], the ACPI_FADT_LOW_POWER_S0
flag merely means that it is better to use low-power S0 idle on the
given platform than S3 (provided that the latter is supported) and it
doesn't preclude using either of them (which of them will be used
depends on the choices made by user space).

However, on some systems that flag is used to indicate whether or not
to enable special firmware mechanics allowing the system to save more
energy when suspended to idle.  If that flag is unset, doing so is
generally risky.

Accordingly, add a comment to explain the ACPI_FADT_LOW_POWER_S0 check
in amdgpu_acpi_is_s0ix_active(), the purpose of which is otherwise
somewhat unclear.

Link: https://uefi.org/specs/ACPI/6.4/05_ACPI_Software_Programming_Model/ACPI_Software_Programming_Model.html#fixed-acpi-description-table-fadt
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: use time_after(a,b) to replace "a>b"

time_after() deals with timer wrapping correctly.

Signed-off-by: Yu Zhe <yuzhe@nfschina.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use dev_info to benefit mGPU case

'free PSP TMR buffer' happens in suspend, but sometimes
in mGPU config, it mixes with PSP resume log printing from
another GPU, which is confusing. So use dev_info instead of
DRM_INFO for printing.

[drm] PSP is resuming...
[drm] reserve 0xa00000 from 0x877e000000 for PSP TMR
amdgpu 0000:e3:00.0: amdgpu: GECC is enabled
amdgpu 0000:e3:00.0: amdgpu: SECUREDISPLAY: securedisplay ta ucode is not available
amdgpu 0000:e3:00.0: amdgpu: SMU is resuming...
amdgpu 0000:e3:00.0: amdgpu: smu driver if version = 0x00000040, smu fw if version = 0x00000041, smu fw program = 0, version = 0x003a5400 (58.84.0)
amdgpu 0000:e3:00.0: amdgpu: SMU driver if version not matched
amdgpu 0000:e3:00.0: amdgpu: dpm has been enabled
amdgpu 0000:e3:00.0: amdgpu: SMU is resumed successfully!
[drm] DMUB hardware initialized: version=0x02020014
[drm] free PSP TMR buffer
[drm] kiq ring mec 2 pipe 1 q 0

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

drm/amdgpu: use adev_to_drm to get drm device

adev_to_drm is used everywhere in amdgpu code, so modify
it to keep consistency.

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

drm/amdgpu: add MGCG perfmon setting for gfx11

Enable GFX11 MGCG perfmon setting.
V2: set rlc to saft mode before setting.

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

drm/amd/amdgpu: avoid soft reset check when gpu recovery disabled

Avoid soft reset, even ip hang check (ring/ib test) when gpu recovery
disabled.

v2: add missing "}"

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

drm/amdkfd: Fix isa version for the GC 10.3.7

Correct the isa version for handling KFD test.

Fixes: 7c4f4f197e0c ("drm/amdkfd: Add GC 10.3.6 and 10.3.7 KFD definitions")
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Aaron Liu <aaron.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix page table setup on Arcturus

When translate_further is enabled, page table depth needs to
be updated. This was missing on Arcturus MMHUB init. This was
causing address translations to fail for SDMA user-mode queues.

Fixes: 352e683b72e7 ("drm/amdgpu: Enable translate_further to extend UTCL2 reach")
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Allocate doorbells only when needed

Only allocate doorbells when the first queue is created on a GPU or the
doorbells need to be mapped into CPU or GPU virtual address space. This
avoids allocating doorbells unnecessarily and can allow more processes
to use KFD on multi-GPU systems.

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: update SMU 13.0.0 driver_if header

To fit the latest 78.53 PMFW.

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

drm/amd/display: Remove redundant check in atomic_check

[Why]
We have 2 back-to-back checks for skipping connectors.
Logically one of them will do the job.

[How]
Remove redundant check.

Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Roman Li <roman.li@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Adding log clock table from SMU

[Why & How]
Adding log for clock table from SMU helps with the debugging process.
Implemented using DC_LOG_SMU to output log.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Leo Chen <sancchen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Uncomment SubVP scaling case

[Description]
Uncomment scaling cmd assignment since
FW headers are now promoted.

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add interface to track PHY state

[Why]
Sometimes pixel clock needs to remain active after transmitter disable.

[How]
Use update_phy_state to track PHY state after stream
enable/disable and program pixel clock as needed.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Taimur Hassan <Syed.Hassan@amd.com>
Signed-off-by: Alvin Lee <alvin.lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Free phantom plane and stream properly

[Description]
Refcount is incremented on allocation and
when adding to the context. Therefore we must
release the phantom plane and stream after
removing from the context.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: program k1/k2 divider for virtual signal for DCN32

[Why&How]
When using IGT, kms_bw multi display tests trigger an assert since
we ignore virtual signal type. k1/k2 dividers should be correctly
programmed if VSYNC needs to be correct. Add the appropriate condition
to the if arm to fix this.

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Cursor flicker when entering PSRSU

[Why]
The DAL driver may transmit the wrong cursor position to PSRSU
DMUB driver when there are multiple planes.

[How]
Currently the driver apply the HW cursor on the top plane. So we
should only transmit the cursor position on the top plane to
PSRSU DMUB driver.

Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Robin Chen <po-tchen@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: change to runtime initialization for reg offsets for DCN321

DC was using compile time initialization of register addresses using
SR_* macros and their variants. These have been converted to use runtime
initialization.

The REG_STRUCT macro is a definition that is added to SR_* macros.
During initialization, this must be defined before SR_* macros are
invoked, which are in turn invoked through various IP initialization macros.

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: change to runtime initialization for reg offsets for DCN32

DC was using compile time initialization of register addresses using
SR_* macros and their variants. These have been converted to use runtime
initialization.

The REG_STRUCT macro is a definition that is added to SR_* macros.
During initialization, this must be defined before SR_* macros are
invoked, which are in turn invoked through various IP initialization macros.

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Change AUX NACK behavior

[Why]
Retrying on receiving a NACK can result in long overall EDID read times
in some cases.

[How]
Retry only on DEFER and return immediately on NACK.

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Ilya Bakoulin <Ilya.Bakoulin@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: HDMI ODM Combine Policy Correction

[WHY]
Reprogramming the stream despite no changes in ODM combine mode.
Reprogramming the stream would cause intermittent black screen on
display which could only be recovered through enable/disable sequence.

[HOW]
Fixed bug where we detected a change in ODM combine mode despite ODM
combine mode being disabled. Also removed code which required stream to
be reprogrammed once a change in ODM combine mode was noticed. Lastly we
do not support dynamic ODM switching for HDMI TMDS and FRL on DCN32,
therefore we never want to change its ODM policy.

Reviewed-by: Samson Tam <Samson.Tam@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Saaem Rizvi <SyedSaaem.Rizvi@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm 2:1 policy not being applied consistently in 4k144 modes

[Why]
odm 2:1 policy is splitting the pipes in 4k144.
then in subvp code, we merge the pipes. but since the
 configuration is unsupported, we keep the pipes split

[How]
for unsupported subvp configuration, redo the dml and
 pipe split calls

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Samson Tam <samson.tam@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: do not change pipe split policy for RV2

[Why]
RV2 do not change pipe split policy in the
minimal pipe split transition state.
This will unblock mode support on some
parts that limit to DPM0 for power reason.

[How]
Do not change pipe split policy in the
minimal pipe split transition state to
allow 4k multi display configs to be
supported at DPM0.

Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Derek Lai <Derek.Lai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.199

This verion brings along following fixes:
-Add scaling factor for SubVP
-Modify stop_dbg_mode return value
-Add gfx_off members and document
-Add GFXOFF function for vangogh
-Add GFXOFF stats to debug
-Fix codestyle problems
-Fix overflow on MIN_I64
-Fix Unneeded semicolon
-Fix comment typo
-Remove useless condition in amdgpu_job_stop_all_jobs_on_sched()
-Add decoder_iv_ts helper for ih_v6
-Add chip version to DCN32
-Avoid doing vm_init multiple time
-Modify size calculation in MALL
-Fix DSC for phantom pipes
-Update clock table policy for DCN314
-Modify header inclusion pattern
-Fix plug/unplug external monitor will playback MPO video
-Add debug parameter to retain default clock table
-Increase tlb flush timeout for sriov
-Fix compare intergers of different widths
-Add reserved dc_log_type
-Fix pixel clock programming

Acked-by: Brian Chang <Brian.Chang@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 0.0.130.0

- For SubVP add scaling factor to allow firmware to calculate
 accurate line to start programming

Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Brian Chang <Brian.Chang@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/amdgpu: skip set_topology_info for VF

Skip set_topology_info as xgmi TA will now block it
and host needs to program it.

Signed-off-by: Vignesh Chander <Vignesh.Chander@amd.com>
Reviewed-By : Shaoyun Liu <Shaoyun.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add sdma instance check for gfx11 CGCG

For some ASICs, like GFX IP v11.0.1, only have one SDMA instance,
so not need to configure SDMA1_RLC_CGCG_CTRL for this case.

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

drm/amd/display: remove unneeded defines from bios parser

Removes DEFINEs that should have been removed after they were
introduced to ObjectID.h by the commit abea57d70e90
("drm/amdgpu: Add BRACKET_LAYOUT_ENUMs to ObjectID.h")

Signed-off-by: Tales Aparecida <tales.aparecida@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: enable PCON support for dcn314

[Why]
DCN314 supports PCON.

[How]
Explicitly enable it in dcn314 resources.

Signed-off-by: Roman Li <roman.li@amd.com>
Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable NBIO IP v7.7.0 Clock Gating

Enable AMD_CG_SUPPORT_BIF_MGCG and AMD_CG_SUPPORT_BIF_LS support.

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

drm/amdgpu: add NBIO IP v7.7.0 Clock Gating support

Add BIF Clock Gating MGCG and LS support for NBIO IP v7.7.0.

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

drm/amdgpu: add TX_POWER_CTRL_1 macro definitions for NBIO IP v7.7.0

Add the BIF0_PCIE_TX_POWER_CTRL_1 register offset and mask macro
definitions for AMD_CG_SUPPORT_BIF_LS.

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

drm/radeon: add a force flush to delay work when radeon

Although radeon card fence and wait for gpu to finish processing current batch rings,
there is still a corner case that radeon lockup work queue may not be fully flushed,
and meanwhile the radeon_suspend_kms() function has called pci_set_power_state() to
put device in D3hot state.
Per PCI spec rev 4.0 on 5.3.1.4.1 D3hot State.
> Configuration and Message requests are the only TLPs accepted by a Function in
> the D3hot state. All other received Requests must be handled as Unsupported Requests,
> and all received Completions may optionally be handled as Unexpected Completions.
This issue will happen in following logs:
Unable to handle kernel paging request at virtual address 00008800e0008010
CPU 0 kworker/0:3(131): Oops 0
pc = [<ffffffff811bea5c>]  ra = [<ffffffff81240844>]  ps = 0000 Tainted: G        W
pc is at si_gpu_check_soft_reset+0x3c/0x240
ra is at si_dma_is_lockup+0x34/0xd0
v0 = 0000000000000000  t0 = fff08800e0008010  t1 = 0000000000010000
t2 = 0000000000008010  t3 = fff00007e3c00000  t4 = fff00007e3c00258
t5 = 000000000000ffff  t6 = 0000000000000001  t7 = fff00007ef078000
s0 = fff00007e3c016e8  s1 = fff00007e3c00000  s2 = fff00007e3c00018
s3 = fff00007e3c00000  s4 = fff00007fff59d80  s5 = 0000000000000000
s6 = fff00007ef07bd98
a0 = fff00007e3c00000  a1 = fff00007e3c016e8  a2 = 0000000000000008
a3 = 0000000000000001  a4 = 8f5c28f5c28f5c29  a5 = ffffffff810f4338
t8 = 0000000000000275  t9 = ffffffff809b66f8  t10 = ff6769c5d964b800
t11= 000000000000b886  pv = ffffffff811bea20  at = 0000000000000000
gp = ffffffff81d89690  sp = 00000000aa814126
Disabling lock debugging due to kernel taint
Trace:
[<ffffffff81240844>] si_dma_is_lockup+0x34/0xd0
[<ffffffff81119610>] radeon_fence_check_lockup+0xd0/0x290
[<ffffffff80977010>] process_one_work+0x280/0x550
[<ffffffff80977350>] worker_thread+0x70/0x7c0
[<ffffffff80977410>] worker_thread+0x130/0x7c0
[<ffffffff80982040>] kthread+0x200/0x210
[<ffffffff809772e0>] worker_thread+0x0/0x7c0
[<ffffffff80981f8c>] kthread+0x14c/0x210
[<ffffffff80911658>] ret_from_kernel_thread+0x18/0x20
[<ffffffff80981e40>] kthread+0x0/0x210
 Code: ad3e0008  43f0074a  ad7e0018  ad9e0020  8c3001e8  40230101
 <88210000> 4821ed21
So force lockup work queue flush to fix this problem.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Zhenneng Li <lizhenneng@kylinos.cn>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unused header

dml_wrapper* files were removed in commit 01b537eeb049
("drm/amd/display: Remove unused code"), as they are not used anywhere.
However, the header file wasn't removed, so remove the header as well.

Signed-off-by: Magali Lemes <magalilemes00@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Include missing header

The file amdgpu_dm_plane.c missed the header amdgpu_dm_plane.h, which
resulted on the following warning:

drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:1046:5:
warning: no previous prototype for 'fill_dc_scaling_info'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:1222:6:
warning: no previous prototype for 'handle_cursor_update'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:152:6:
warning: no previous prototype for 'modifier_has_dcc'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:1576:5:
warning: no previous prototype for 'amdgpu_dm_plane_init'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:157:10:
warning: no previous prototype for 'modifier_gfx9_swizzle_mode'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:752:5:
warning: no previous prototype for 'fill_plane_buffer_attributes'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:83:31:
warning: no previous prototype for 'amd_get_format_info'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:88:6:
warning: no previous prototype for 'fill_blending_from_plane_state'
[-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:992:5:
warning: no previous prototype for 'dm_plane_helper_check_state'
[-Wmissing-prototypes]

Therefore, include the missing header on the file and turn global functions
that are not used outside of the file into static functions.

Fixes: 5d945cbcd4b1 ("drm/amd/display: Create a file dedicated to planes")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Maíra Canal <mairacanal@riseup.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unneeded semicolon

Semicolon is not required after curly braces.

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=1918
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix i386 frame size warning

Addresses the following warning:
drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn30/display_mode_vba_30.c:3596:6: error: stack frame size (2092) exceeds limit (2048) in 'dml30_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
void dml30_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
     ^

UseMinimumDCFCLK() is eating away at
dml30_ModeSupportAndSystemConfigurationFull()'s stack space, so use a
pointer to struct vba_vars_st instead of passing lots of large arrays
as parameters by value.

Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Remove the additional kfd pre reset call for sriov

The additional call is caused by merge conflict

Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Check num_gfx_rings for gfx v9_0 rb setup.

No need to set up rb when no gfx rings.

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

drm/amdgpu: fix hive reference leak when adding xgmi device

Only amdgpu_get_xgmi_hive but no amdgpu_put_xgmi_hive
which will leak the hive reference.

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

drm/amdgpu: Move psp_xgmi_terminate call from amdgpu_xgmi_remove_device to psp_hw_fini

V1:
The amdgpu_xgmi_remove_device function will send unload command
to psp through psp ring to terminate xgmi, but psp ring has been
destroyed in psp_hw_fini.

V2:
1. Change the commit title.
2. Restore amdgpu_xgmi_remove_device to its original calling location.
   Move psp_xgmi_terminate call from amdgpu_xgmi_remove_device to
   psp_hw_fini.

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

drm/amdgpu: enable GFXOFF allow control for GC IP v11.0.1

Enable GFXOFF allow control when set the GFX power gating.

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

drm/amdgpu: Document gfx_off members of struct amdgpu_gfx

Add comments to document gfx_off related members of struct amdgpu_gfx.

Signed-off-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Documentation/gpu: Document GFXOFF's count and residency

Add documentation explaining those two new files.

While here, add a note about the value type.

Signed-off-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Implement GFXOFF's entry count and residency for vangogh

Implement functions to get and set GFXOFF's entry count and residency
for vangogh.

Signed-off-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd: Add detailed GFXOFF stats to debugfs

Add debugfs interface to log GFXOFF statistics:

- Read amdgpu_gfxoff_count to get the total GFXOFF entry count at the
  time of query since system power-up

- Write 1 to amdgpu_gfxoff_residency to start logging, and 0 to stop.
  Read it to get average GFXOFF residency % multiplied by 100
  during the last logging interval.

Both features are designed to be keep the values persistent between
suspends.

Signed-off-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix minor codestyle problems

Fixes five checkpatch warnings:

CHECK: Please don't use multiple blank lines
+
+

ERROR: Macros with complex values should be enclosed in parentheses
+#define MAX_I64 \
+       (int64_t)((1ULL << 63) - 1)

WARNING: Missing a blank line after declarations
+       struct bw_fixed res;
+       ASSERT(value < BW_FIXED_MAX_I32 && value > BW_FIXED_MIN_I32);

ERROR: that open brace { should be on the previous line
+               do
+               {

ERROR: that open brace { should be on the previous line
+                       if (remainder >= arg2_value)
+                       {

Signed-off-by: Tales Aparecida <tales.aparecida@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix overflow on MIN_I64 definition

The definition of MIN_I64 in bw_fixed.c can cause gcc to whinge about
integer overflow, because it is treated as a positive value, which is
then negated. The temporary positive value is not necessarily
representable.

This causes the following warning:
../drivers/gpu/drm/amd/amdgpu/../display/dc/dml/calcs/bw_fixed.c:30:19:
warning: integer overflow in expression ‘-9223372036854775808’ of type
‘long long int’ results in ‘-9223372036854775808’ [-Woverflow]
  30 |         (int64_t)(-(1LL << 63))
     |                   ^

Writing out (-MAX_I64 - 1) works instead.

Signed-off-by: David Gow <davidgow@google.com>
Signed-off-by: Tales Aparecida <tales.aparecida@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Unneeded semicolon

There is no semicolon after '}' in line 510.

Signed-off-by: min tang <tangmin@cdjrlc.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix comment typo

The double `aligned' is duplicated in line 1070, remove one.

Signed-off-by: min tang <tangmin@cdjrlc.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove unused code

We have some old code associated with DML, which we had plans to use,
but at some point, we just moved away from it. This commit removes the
dml_wrapper* files since they are not used anymore.

Reported-by: Magali Lemes <magalilemes00@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use sjt mec fw on aldebaran for sriov

The second jump table is required on live migration or mulitple VF
configuration on Aldebaran. With this implemented, the first level
jump table(hw used) will be same, mec fw internal will use the
second level jump table jump to the real functionality implementation.
so the different VF can load different version of MEC as long as
they support sjt

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: include soc._clock_tmp[] into DC's scratch region

Currently, we are using soc._clock_tmp[] to temporarily store and modify
data from soc.clock_limits[] before copying it back into
soc.clock_limits[] (because modifying data directly in
soc.clock_limits[] causes unintended behavior). However, this approach
has a number of downsides, such as:

1. struct _vcs_dpi_soc_bounding_box_st's creation/destruction
   aren't well defined (which could mean more unintended
   behavior).
2. Throwing "temp" varibles in structs everywhere doesn't make
   for a particularly readable codebase.

For these reasons, we should get rid of soc._clock_tmp[] by defining a
struct scratch within struct dc that, contains within it all of the
temporary variables (including _clock_tmp[]) such that it is obvious how
they are intended to be used.

Co-authored-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: implement DSC pass-through support

Currently, we only attempt to setup DSC at the virtual DPCD port,
however many modern displays come with DSC support and MST hubs can now
support DSC pass-through. So, to more optimally make use of the
available bandwidth, use DSC pass-through when possible by adding DSC
pass-through enablement support into the DSC enable sequence.

Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: consider DSC pass-through during mode validation

Add a mode validation routine for DSC pass-through. Both the link from
source to hub, and the link from hub to monitor are checked, according
to the current link training result and full pbn returned by enum path
resource sideband message.

Pick up the minimum value as the bandwidth bottleneck for the end to
end link bandwidth constraint, and check if the maximum DSC decompression
bandwidth can fit.

v2: squash in DSC CONFIG_DRM_AMD_DC_DCN compilation fix (Hamza)

Co-authored-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/dp_mst: add passthrough_aux to struct drm_dp_mst_port

Currently, there is no way to identify if DSC pass-through can be
enabled and what aux DSC pass-through requests ought to be sent to. So,
add a variable to struct drm_dp_mst_port that keeps track of the
aforementioned information.

Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Documentation/gpu/amdgpu/amdgpu_dm: add DM docs for pixel blend mode

AMD GPU display manager (DM) maps DRM pixel blend modes (None,
Pre-multiplied, Coverage) to MPC hw blocks through blend configuration
options. Describe relevant elements and how to set and test them to get
the expected DRM blend mode on DCN hw.

v2:
- add ref tag (Tales)

Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Tales Aparecida <tales.aparecida@gmail.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add doc entries for MPC blending configuration

Describe structs and enums used to set blend mode properties to MPC
blocks. Some pieces of information are already available as code
comments, and were just formatted. Others were collected and summarised
from discussions on AMD issue tracker[1][2].

[1] https://gitlab.freedesktop.org/drm/amd/-/issues/1734
[2] https://gitlab.freedesktop.org/drm/amd/-/issues/1769

v2:
- fix typos (Tales)
- add MPCC to MPC entry in the glossary

Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Tales Aparecida <tales.aparecida@gmail.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Documentation/amdgpu/display: add DC color caps info

Add details about color correction capabilities and explain a bit about
differences between DC hw generations and also how they are mapped
between DRM and DC interface. Two schemas for DCN 2.0 and 3.0 (converted
to svg from the original png) is included to illustrate it. They were
obtained from a discussion[1] in the amd-gfx mailing list.

[1] https://lore.kernel.org/amd-gfx/20220422142811.dm6vtk6v64jcwydk@mail.igalia.com/

v1:
- remove redundant comments (Harry)
- fix typos (Harry)

v2:
- reword introduction of color section
- add co-dev tag for Harry - who provided most of the info
- fix typos (Tales)
- describe missing struct parameters (Tales and Siqueira)

Co-developed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Tales Aparecida <tales.aparecida@gmail.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Documentation/amdgpu_dm: Add DM color correction documentation

AMDGPU DM maps DRM color management properties (degamma, ctm and gamma)
to DC color correction entities. Part of this mapping is already
documented as code comments and can be converted as kernel docs.

v2:
- rebase to amd-staging-drm-next
- fix typos (Tales)
- undo kernel-docs inside functions (Tales)

Signed-off-by: Melissa Wen <mwen@igalia.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tales Aparecida <tales.aparecida@gmail.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Document some of the DML structs

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Expand documentation for timing

The timing programming inside DCN is far from trivial, it has multiple
parameters associated with that, and the lack of documentation does not
help comprehend this already complicated topic. This commit tries to
improve this situation by expanding the documentation of dc_crtc_timing
and the VTG program function.

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Document pipe split policy

In the DCN code, we can find some references to three different pipe
split policies but no clear description. This commit adds some
kernel-doc that explains more about these options and a little bit of
their impact.

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add some extra kernel doc to amdgpu_dm

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add basic kernel doc to CRC code under DC

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add documentation to some of the cursor struct

Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: reduce reset time

In multi container use case, reset time is important, so skip ring
tests and cp halt wait during ip suspending for reset as they are
going to fail and cost more time on reset

v2: add a hang flag to indicate the reset comes from a job timeout,
skip ring test and cp halt wait in this case

v3: move hang flag to adev

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

drm/amdgpu: revert context to stop engine before mode2 reset

For some hang caused by slow tests, engine cannot be stopped which
may cause resume failure after reset. In this case, force halt
engine by reverting context addresses

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

drm/amdgpu: save and restore gc hub regs

Save and restore gfxhub regs as they will be reset during mode 2

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

drm/amdgpu: add debugfs amdgpu_reset_level

Introduce amdgpu_reset_level debugfs in order to help debug and
test specific type of reset. Also helps blocking unwanted type of
resets.

By default, mode2 reset will not be enabled

v2: make this debugfs in adev and use debugfs_create_u32

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

drm/amdgpu: let mode2 reset fallback to default when failure

- introduce AMDGPU_SKIP_MODE2_RESET flag
- let mode2 reset fallback to default reset method if failed

v2: move this part out from the asic specific part

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

drm/amdgpu: add mode2 reset for sienna_cichlid

To meet the requirement for multi container usecase which needs
a quicker reset and not causing VRAM lost, adding the Mode2
reset handler for sienna_cichlid.

v2: move skip mode2 flag part separately

v3: remove the use of asic_reset_res

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

Revert "drm/amd/amdgpu: add pipe1 hardware support"

This reverts commit 4c7631800e6bf0eced08dd7b4f793fcd972f597d.

Triggered GFX hangs with GNOME Wayland on Navi 21.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2117
Signed-off-by: Michel Dänzer <mdaenzer@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix use-after-free on amdgpu_bo_list mutex

If amdgpu_cs_vm_handling returns r != 0, then it will unlock the
bo_list_mutex inside the function amdgpu_cs_vm_handling and again on
amdgpu_cs_parser_fini. This problem results in the following
use-after-free problem:

[ 220.280990] ------------[ cut here ]------------
[ 220.281000] refcount_t: underflow; use-after-free.
[ 220.281019] WARNING: CPU: 1 PID: 3746 at lib/refcount.c:28 refcount_warn_saturate+0xba/0x110
[ 220.281029] ------------[ cut here ]------------
[ 220.281415] CPU: 1 PID: 3746 Comm: chrome:cs0 Tainted: G W L ------- --- 5.20.0-0.rc0.20220812git7ebfc85e2cd7.10.fc38.x86_64 #1
[ 220.281421] Hardware name: System manufacturer System Product Name/ROG STRIX X570-I GAMING, BIOS 4403 04/27/2022
[ 220.281426] RIP: 0010:refcount_warn_saturate+0xba/0x110
[ 220.281431] Code: 01 01 e8 79 4a 6f 00 0f 0b e9 42 47 a5 00 80 3d de
7e be 01 00 75 85 48 c7 c7 f8 98 8e 98 c6 05 ce 7e be 01 01 e8 56 4a
6f 00 <0f> 0b e9 1f 47 a5 00 80 3d b9 7e be 01 00 0f 85 5e ff ff ff 48
c7
[ 220.281437] RSP: 0018:ffffb4b0d18d7a80 EFLAGS: 00010282
[ 220.281443] RAX: 0000000000000026 RBX: 0000000000000003 RCX: 0000000000000000
[ 220.281448] RDX: 0000000000000001 RSI: ffffffff988d06dc RDI: 00000000ffffffff
[ 220.281452] RBP: 00000000ffffffff R08: 0000000000000000 R09: ffffb4b0d18d7930
[ 220.281457] R10: 0000000000000003 R11: ffffa0672e2fffe8 R12: ffffa058ca360400
[ 220.281461] R13: ffffa05846c50a18 R14: 00000000fffffe00 R15: 0000000000000003
[ 220.281465] FS: 00007f82683e06c0(0000) GS:ffffa066e2e00000(0000) knlGS:0000000000000000
[ 220.281470] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 220.281475] CR2: 00003590005cc000 CR3: 00000001fca46000 CR4: 0000000000350ee0
[ 220.281480] Call Trace:
[ 220.281485] <TASK>
[ 220.281490] amdgpu_cs_ioctl+0x4e2/0x2070 [amdgpu]
[ 220.281806] ? amdgpu_cs_find_mapping+0xe0/0xe0 [amdgpu]
[ 220.282028] drm_ioctl_kernel+0xa4/0x150
[ 220.282043] drm_ioctl+0x21f/0x420
[ 220.282053] ? amdgpu_cs_find_mapping+0xe0/0xe0 [amdgpu]
[ 220.282275] ? lock_release+0x14f/0x460
[ 220.282282] ? _raw_spin_unlock_irqrestore+0x30/0x60
[ 220.282290] ? _raw_spin_unlock_irqrestore+0x30/0x60
[ 220.282297] ? lockdep_hardirqs_on+0x7d/0x100
[ 220.282305] ? _raw_spin_unlock_irqrestore+0x40/0x60
[ 220.282317] amdgpu_drm_ioctl+0x4a/0x80 [amdgpu]
[ 220.282534] __x64_sys_ioctl+0x90/0xd0
[ 220.282545] do_syscall_64+0x5b/0x80
[ 220.282551] ? futex_wake+0x6c/0x150
[ 220.282568] ? lock_is_held_type+0xe8/0x140
[ 220.282580] ? do_syscall_64+0x67/0x80
[ 220.282585] ? lockdep_hardirqs_on+0x7d/0x100
[ 220.282592] ? do_syscall_64+0x67/0x80
[ 220.282597] ? do_syscall_64+0x67/0x80
[ 220.282602] ? lockdep_hardirqs_on+0x7d/0x100
[ 220.282609] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 220.282616] RIP: 0033:0x7f8282a4f8bf
[ 220.282639] Code: 00 48 89 44 24 18 31 c0 48 8d 44 24 60 c7 04 24 10
00 00 00 48 89 44 24 08 48 8d 44 24 20 48 89 44 24 10 b8 10 00 00 00
0f 05 <89> c2 3d 00 f0 ff ff 77 18 48 8b 44 24 18 64 48 2b 04 25 28 00
00
[ 220.282644] RSP: 002b:00007f82683df410 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[ 220.282651] RAX: ffffffffffffffda RBX: 00007f82683df588 RCX: 00007f8282a4f8bf
[ 220.282655] RDX: 00007f82683df4d0 RSI: 00000000c0186444 RDI: 0000000000000018
[ 220.282659] RBP: 00007f82683df4d0 R08: 00007f82683df5e0 R09: 00007f82683df4b0
[ 220.282663] R10: 00001d04000a0600 R11: 0000000000000246 R12: 00000000c0186444
[ 220.282667] R13: 0000000000000018 R14: 00007f82683df588 R15: 0000000000000003
[ 220.282689] </TASK>
[ 220.282693] irq event stamp: 6232311
[ 220.282697] hardirqs last enabled at (6232319): [<ffffffff9718cd7e>] __up_console_sem+0x5e/0x70
[ 220.282704] hardirqs last disabled at (6232326): [<ffffffff9718cd63>] __up_console_sem+0x43/0x70
[ 220.282709] softirqs last enabled at (6232072): [<ffffffff970ff669>] __irq_exit_rcu+0xf9/0x170
[ 220.282716] softirqs last disabled at (6232061): [<ffffffff970ff669>] __irq_exit_rcu+0xf9/0x170
[ 220.282722] ---[ end trace 0000000000000000 ]---

Therefore, remove the mutex_unlock from the amdgpu_cs_vm_handling
function, so that amdgpu_cs_submit and amdgpu_cs_parser_fini can handle
the unlock.

Fixes: 90af0ca047f3 ("drm/amdgpu: Protect the amdgpu_bo_list list with a mutex v2")
Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Maíra Canal <mairacanal@riseup.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix interrupt handling on ih_soft ring

There are no backing hardware registers for ih_soft ring.
As a result, don't try to access hardware registers for read
and write pointers when processing interrupts on the IH soft
ring.

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

drm/amdgpu: Add secure display TA load for Renoir

Add secure display TA load for Renoir

Signed-off-by: Shane Xiao <shane.xiao@amd.com>
Reviewed-by: Aaron Liu <aaron.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Include scaling factor for SubVP command

[Description]
For SubVP scaling cases, we must include the scaling
info as part of the cmd. This is required when converting
OTG line to HUBP line for the MALL_START_LINE programming.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: Return void from the stop_dbg_mode

There is no point in returning an int here. It only returns 0 which
the caller never uses. Therefore return void and remove the unnecessary
assignment.

Addresses-Coverity: 1504988 ("Unused value")
Fixes: 8da1170a16e4 ("drm/amdgpu: add VCN4 ip block support")
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Suggested-by: Ruijing Dong <ruijing.dong@amd.com>
Suggested-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Khalid Masum <khalid.masum.92@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove useless condition in amdgpu_job_stop_all_jobs_on_sched()

Local variable 'rq' is initialized by an address
of field of drm_sched_job, so it does not make
sense to compare 'rq' with NULL.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Andrey Strachuk <strochuk@ispras.ru>
Fixes: 7c6e68c777f1 ("drm/amdgpu: Avoid HW GPU reset for RAS.")
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Add decode_iv_ts helper for ih_v6 block

Was missing.  Add it.

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

drm/amd/display: add chip revision to DCN32

[Why & How]
Add GC_11_0_3_A0 as a chip revision to the DCN32 family

Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Samson Tam <samson.tam@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: avoid doing vm_init multiple time

[why]
this is to ensure that driver will not reprogram hvm_prefetch_req again if
it is done.

Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Brian Chang <Brian.Chang@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: Use pitch when calculating size to cache in MALL

[Description]
Use pitch when calculating size to cache in MALL

Reviewed-by: Samson Tam <Samson.Tam@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Don't set DSC for phantom pipes

[Description]
Don't set DSC bit for phantom pipes, not
required since phantom pipe don't have
any actual output

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update clock table policy for DCN314

[Why & How]
Depending on how the clock table is constructed from PMFW we can run
into issues where we don't think we have enough bandwidth available
due to FCLK too low - eg. when the FCLK table contains invalid entries
or a single entry.

We should always pick up the maximum clocks for each state as a final
state in this case to prevent validation from failing if the table is
malformed.

We should also contain sensible defaults in the case where values
are invalid.

Redfine the clock table structures by adding a 314 prefix to make
debugging these issues easier by avoiding symbol name clashes.

Overall this policy more closely aligns to how we did things for 315,
but because of how the voltage rail is setup we should favor keeping
DCFCLK low rather than DISPCLK or DPPCLK - so use the max for those
in every entry.

Reviewed-by: Daniel Miess <daniel.miess@amd.com>
Acked-by: Brian Chang <Brian.Chang@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: Modify header inclusion pattern

[Why]
Recent backport from opensource broke the Nightly tool build
that tests DC and DML for bugs and regressions. This was
because the backport had a header inclusion that was not
consistent with the AMD style of including headers was allowed
to be merged back in DML code that caused tool compilation
failures.

[How]
Modify the way in which the header file in included so that it
is consistent with AMD style of including headers. This then
automatically fixes the tool compilation process and also
helps maintain the code quality and consistency.

Reviewed-by: Alvin Lee <alvin.lee2@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Chaitanya Dhere <chaitanya.dhere@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>