drm/amdgpu: defer cmd/fence/fw buffers destroy on hw_init failure

As the cleanup jobs performed in pre_fini may still need these
buffers. NULL pointer dereference will be triggered without them.

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/amdgpu: add more debug friendly prompts

Large piece of codes share one error prompt. That
is not friendly for debugging.

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/amdgpu: error out on mode1 reset failure

The error return value should be correctly reflected.

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/amdgpu: trivial typo fix

"error" was not correctly spelled.

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

drm/amdgpu: remove per obj debugfs write

there is ras_control node which can do its job.

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

drm/amdgpu: Fix amdgpu ras to ta enums conversion

Add helpes to transalte the two enums. And it will catch bugs
easily.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use macro instead of enum for flags

better to use macro.

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

drm/amdgpu: Fix some sanity check

ras context might be NULL, so move con->h_data after check !con
also fix sizeof wrong type while at it.

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

drm/amd/powerplay: Fix double unlock bug in smu_sys_set_pp_table()

We already unlocked a few lines earlier so this code unlocks twice on
the success path.

Fixes: 289921b03fe5e ("drm/amd/powerplay: implement sysfs of pp_table for smu11 (v2)")
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: Off by one in vega20_get_smu_msg_index()

The > should be >= so that we don't read one element beyond the end of
the vega20_message_map[] array.

Fixes: 78031c2c4dcd ("drm/amd/powerplay: implement smu vega20_message_map for vega20")
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: delete some dead code

The "size" variable is unsigned.  We never pass invalid sizes to this
function and we already used it as an array offset earlier so it's
too late to check here.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: fix build when DCN KCONFIG is not set

Leads to an undefined symbol otherwise.

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

drm/amdgpu: revert "XGMI pstate switch initial support"

This reverts commit 9b638f9751308ae3ae8f28e0c6e9decffd97f5f9.

Adding this to the mapping is complete nonsense and the whole
implementation looks racy. This patch wasn't thoughtfully reviewed
and should be reverted for now.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Liu, Shaoyun <Shaoyun.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use the new VM backend for PTEs

And remove the existing code when it is unused.

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

drm/amdgpu: use the new VM backend for PDEs

And remove the existing code when it is unused.

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

drm/amdgpu: new VM update backends

Separate out all functions for SDMA and CPU based page table
updates into separate backends.

This way we can keep most of the complexity of those from the
core VM code.

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

drm/amdgpu: reserve less memory for PDE updates

Allocating 16KB was way to much, just use 2KB as a start for now.

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

drm/amdgpu: move and rename amdgpu_pte_update_params

Move the update parameter into the VM header and rename them.

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

drm/amdgpu: always set and check dma addresses in the VM code

Clean that up a bit and allow to always have the DMA addresses around.

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

drm/amdgpu: remove some unused VM defines

Not needed any more.

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

drm/amdgpu: add one rlc version into gfxoff blacklist

RLC #53815 ucode has the noise issue on 4k playback while gfxoff enabled.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Tested-by: Tom St Denis <tom.stdenis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable gfxoff again on raven series (v2)

This patch enables gfxoff and stutter mode again, since we take more testing on
raven series. For raven2 and picasso, we can enable it directly. And for raven,
we need check the RLC/SMC ucode version cannot be less than #531/0x1e45.

v2: add smc version checking for raven.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1)
Tested-by: Likun Gao <Likun.Gao@amd.com> (v2)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: fix spelling mistake "unknow" -> "unknown"

There is a spelling mistake in pr_warn message; fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: Zero initialize num_of_levels in vega20_set_single_dpm_table

When building with -Wsometimes-uninitialized, Clang warns:

drivers/gpu/drm/amd/amdgpu/../powerplay/vega20_ppt.c:456:2: warning:
variable 'num_of_levels' is used uninitialized whenever '?:' condition
is false [-Wsometimes-uninitialized]
        smu_read_smc_arg(smu, &num_of_levels);
        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../powerplay/inc/amdgpu_smu.h:608:3: note:
expanded from macro 'smu_read_smc_arg'
        ((smu)->funcs->read_smc_arg? (smu)->funcs->read_smc_arg((smu), (arg)) : 0)
         ^~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../powerplay/vega20_ppt.c:457:7: note:
uninitialized use occurs here
        if (!num_of_levels) {
             ^~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/../powerplay/vega20_ppt.c:456:2: note: remove
the '?:' if its condition is always true
        smu_read_smc_arg(smu, &num_of_levels);
        ^
drivers/gpu/drm/amd/amdgpu/../powerplay/inc/amdgpu_smu.h:608:3: note:
expanded from macro 'smu_read_smc_arg'
        ((smu)->funcs->read_smc_arg? (smu)->funcs->read_smc_arg((smu), (arg)) : 0)
         ^
drivers/gpu/drm/amd/amdgpu/../powerplay/vega20_ppt.c:446:27: note:
initialize the variable 'num_of_levels' to silence this warning
        uint32_t i, num_of_levels, clk;
                                 ^
                                  = 0
1 warning generated.

The if statement it mentions as potentially problematic is currently
always true because the read_smc_arg callback is assigned at the
bottom of this file but Clang can't tell that. If the callback were
ever to disappear, num_of_levels would never be initialized. Just
zero initialize it to ensure that the intent behind this code
remains the same.

Link: https://github.com/ClangBuiltLinux/linux/issues/425
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: remove set but not used variable 'rdev'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/gpu/drm/radeon/radeon_ttm.c: In function 'radeon_move_vram_ram':
drivers/gpu/drm/radeon/radeon_ttm.c:254:24: warning:
 variable 'rdev' set but not used [-Wunused-but-set-variable]

drivers/gpu/drm/radeon/radeon_ttm.c: In function 'radeon_move_ram_vram':
drivers/gpu/drm/radeon/radeon_ttm.c:301:24: warning:
 variable 'rdev' set but not used [-Wunused-but-set-variable]

It's not used since a987fcaa805f ("ttm: Make parts of a struct ttm_bo_device
global.")

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

drm/amd/display: Only put primary planes into the mode_info->planes list

We want DRM planes to be initialized in the following order:

- primary planes
- overlay planes
- cursor planes

to support existing userspace expectations for plane z-ordering. This
means that we also need to register CRTCs after all planes have been
initialized since overlay planes can be placed on any CRTC.

So the only reason why we have the mode_info->planes list is to
remember the primary planes for use later when we need to register
the CRTC.

Overlay planes have no purpose being in this list. DRM will cleanup
any planes that we've registered for us, so the only planes that need to
be explicitly cleaned up are the ones that have failed to register.

By dropping the explicit free on every plane in the mode_info->planes
list this patch also fixes a double-free in the case where we fail to
initialize only some of the planes.

Cc: Leo Li <sunpeng.li@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add debugfs dpcd interface

[Why]
We need arbitrary read/write over DP AUX DPCD
for debugging

[How]
Three debugfs entries

Set the target address by writing to
"aux_dpcd_address"
(The first four bytes written are used)

Set the transaction size in bytes by writing to
"aux_dpcd_size"
(The first four bytes written are used)

Start a transaction by reading/writing
"aux_dpcd_data"

Do note: there is no concerrency protection at all
Accessing these entries in quick succession can lead
to strange behaviour

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Generic SDP message access in amdgpu

[Why]
We need to add DP SDP message test debugfs to make sdp message test
more convenient and efficient.

[How]
Add sdp_message debugfs entry in amdgpu.

Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Expose generic SDP message access interface

[Why]
We need to add DP SDP message test debugfs to make sdp message test
more convenient and efficient.

[How]
Add a DM accessible SDP interface for custom data.

Signed-off-by: Leo (Hanghong) Ma <hanghong.ma@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm pipe management

There are issues removing surfaces/streams when odm is active.
This is a step to fix that

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add missing opp programming for odm

A number of places opp programming was missing for odm second pipe.
This change fixes the oversight.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm combine pipe reset

We fail to reset the second odm combine pipe. This change fixes
odm pointer management.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix releasing planes when exiting odm

Releasing planes should not release the 2nd odm pipe right away,
this change leaves us with 2 pipes with null planes and same stream
when planes are released during odm.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.23

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

drm/amd/display: Fix VTEM InfoPacket programming

Refactor setting bit fields. Correcting the offset of MD0.
Initializing the InfoPacket header fields. Defining the field offsets
and masks.

Signed-off-by: Reza Amini <Reza.Amini@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Link train only when link is DP and backend is enabled

[Why]
In certain cases we do link training when we don't have a backend.

[How]
In dc_link_set_preferred_link_settings(), store preferred link settings
first and then verify that the link is DP and the link stream's backend is
enabled.  If either is false, then we will not do any link retraining.

Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add global master update lock interfaces

Add these interfacese for future use
setup_global_lock()
lock_global()
pipe_control_lock_global()

Signed-off-by: Wenjing Liu <Wenjing.Liu@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use dc_is_hdmi_signal() instead of ENUM

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Nevenko Stupar <Nevenko.Stupar@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: cache additional dpcd caps for HDR capability check

[Why]
Currently we are missing a few checks to see if HDR10 is allowed.
In particular we never check for the extended colorimetry bit (whether its
present or set to 1). Further we don't read in the dpcd block in DC that
would provide these bits.

[How]
- Added in DC code to read in the block containing the extended colorimetry
bit.

Signed-off-by: Harmanprit Tatla <Harmanprit.Tatla@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Update ABM crtc state on non-modeset

[Why]
Somewhere in the atomic check reshuffle ABM got lost.
ABM is a crtc property (copied from a connector property).
It can change without a modeset, just like underscan.

[How]
In the skip_modeset branch of atomic check crtc updates,
copy over the abm property.

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Programming correct VRR_EN bit in VTEM structure

[Why]
In HDMI plugfest, MTK report our EMP with VRR_EN bit = 0.
VRR_EN bit is EMP-MD0-bit 0. Currently driver set 1 to bit 3.

[How]
Programming correct VRR_EN bit in EMP-MD0-bit0.

Signed-off-by: Hugo Hu <hugo.hu@amd.com>
Reviewed-by: Reza Amini <Reza.Amini@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.22

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

drm/amd/display: extend EDID support to 1kb

[why]
There exist displays with EDIDs > 512 bytes, existing code
will cause us to ignore all extension blocks.

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix DP audio regression

Audio was unmuted for HDMI only, need to do so for DP as well.

Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add a hysteresis to BTR frame multiplier

[Why]
Flickering is observed on some displays when the number of inserted BTR
frames changes frequently.

[How]
Add in a margin of drift to prevent the inserted number of frames from
jumping around too frequently.

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix DP 422 VID_M half the rate issue.

[Description]
when programming VID_TIMING, we were using the original VESA timing for DP_VIDM/N.
for YCbCr420 or compressed YCbCr422, using half rate as  YCbCr444.

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

drm/amd/display: Fix setting DP_VID_N_MUL

[Why]
Need to set VID_N_MUL for 4:2:0 cases

[How]
Move setting to enc1_stream_encoder_dp_unblank and
ensure it is also set for non-4:2:0 cases.

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Move dm_read_reg_func to dc_helper.

Move out of header to dc_helper.c, in preparation for future
implementations.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use drm helper for resetting plane state

[Why]
To help prevent plane state not being set to the correct default
value if any new properties are added in the future.

[How]
Use the drm helper - which seems to be the common solution among other
DRM drivers.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: change generic_reg_wait to void.

we were only checking the return value in one place, thus changing
generic_reg_wait from int to void and reading the register instead of
getting it from generic_reg_wait, when we need the return value.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Reset alpha state for planes to the correct values

[Why]
The plane_reset callback is subclassed but hasn't been updated since
the drm helper got updated to include resetting alpha related state
(state->alpha and state->pixel_blend_mode). The overlay planes
exposed by amdgpu_dm were therefore being rendered as invisible by
default ever since supported was exposed for alpha blending properties
on overlays.

This caused regressions in igt@kms_plane_multiple@atomic-tiling-none
and igt@kms_plane@plane-position-covered-pipe tests.

[How]
Reset the plane state values to their correct values as defined in
the drm helper.

This fixes the IGT test regression.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: use more entries for the first paging queue

To aid recoverable page faults.

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

drm/amdgpu: free up the first paging queue v2

We need the first paging queue to handle page faults.

v2: handle any number of SDMA instances gracefully

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

drm/amdgpu: re-enable retry faults

Now that we have re-reoute faults to the other IH
ring we can enable retries again.

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

drm/amdkfd/sriov:Put the pre and post reset in exclusive mode v2

add amdgpu_amdkfd_pre_reset and amdgpu_amdkfd_post_reset inside amdgpu_device_reset_sriov.

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

drm/amd/display: Respect DRM framebuffer info for video surfaces

[Why]
Incorrect hardcoded assumptions are made regarding luma and chroma
alignment. The actual values set for the DRM framebuffer should be used
when programming the address.

[How]
Respect the given pitch for both luma and chroma planes - it's not like
we can force the alignment to anything else at this point anyway.

Use the FB offset for the chroma planes directly. DRM already
provides this to us so there's no need to calculate it manually.

While we don't actually use the chroma surface size parameters on Raven,
these should have technically been fb->width / 2 and fb->height / 2
since the chroma plane is half size of the luma plane for NV12.

Leave a TODO indicating that those should be set based on the actual
surface format instead since this is only correct for YUV420 formats.

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

drm/amdgpu: Wait for newly allocated PTs to be idle

When page table are updated by the CPU, synchronize with the
allocation and initialization of newly allocated page tables.

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

drm/amdgpu: more descriptive message if HMM not enabled

If using old kernel config file, CONFIG_ZONE_DEVICE is not selected,
so CONFIG_HMM and CONFIG_HMM_MIRROR is not enabled, the current driver
error message "Failed to register MMU notifier" is not clear. Inform
user with more descriptive message on how to fix the missing kernel
config option.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109808

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: support userptr cross VMAs case with HMM

userptr may cross two VMAs if the forked child process (not call exec
after fork) malloc buffer, then free it, and then malloc larger size
buf, kerenl will create new VMA adjacent to old VMA which was cloned
from parent process, some pages of userptr are in the first VMA, the
rest pages are in the second VMA.

HMM expects range only have one VMA, loop over all VMAs in the address
range, create multiple ranges to handle this case. See
is_mergeable_anon_vma in mm/mmap.c for details.

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/amdkfd: support concurrent userptr update for HMM

Userptr restore may have concurrent userptr invalidation after
hmm_vma_fault adds the range to the hmm->ranges list, needs call
hmm_vma_range_done to remove the range from hmm->ranges list first,
then reschedule the restore worker. Otherwise hmm_vma_fault will add
same range to the list, this will cause loop in the list because
range->next point to range itself.

Add function untrack_invalid_user_pages to reduce code duplication.

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: stop evicting busy PDs/PTs

Otherwise we won't be able to cleanly handle page faults.

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

drm/amdgpu: wait for VM to become idle during flush

Make sure that not only the entities are flush, but that
we also wait for the HW to finish all processing.

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

drm/amdgpu: remove non-sense NULL ptr check

It's a bug having a dead pointer in the IDR, silently returning
is the worst we can do.

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

drm/amdgpu: remove chash

Remove the chash implementation for now since it isn't used any more.

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

drm/amdgpu: use ring/hash for fault handling on GMC9 v3

Further testing showed that the idea with the chash doesn't work as expected.
Especially we can't predict when we can remove the entries from the hash again.

So replace the chash with a ring buffer/hash mix where entries in the container
age automatically based on their timestamp.

v2: use ring buffer / hash mix
v3: check the timeout to make sure all entries age

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> (v2)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: limit the number of IVs processed at once

Only process a maximum of 32 IVs before writing back the RPTR. This improves
hw handling when we get close to an overflow in the ring buffer.

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

drm/amdgpu: enable IH ring 1&2 for Vega20 as well

That doesn't seem to have any negative effects.

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

drm/amdgpu: enable IH doorbell for ring 1&2 on Vega

The doorbells should already be reserved, just enable them.

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

drm/amdgpu: change Vega IH ring 1 config

Disable overflow and enable full drain. This makes fault handling on ring 1
much more reliable since we don't generate back pressure any more.

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

drm/amdgpu: Only clear dumb buffers if ring is enabled

The buffers should be cleared when possible but we also don't want
buffer creation to fail in the rare case where the ring isn't ready
during the call. This could happen during some suspend/resume sequences.

Cc: Christian König <ckoenig.leichtzumerken@gmail.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Clear VRAM for DRM dumb_create buffers

The dumb_create API isn't intended for high performance rendering
and it's more useful for userspace (ie. IGT) to have them precleared.

The bonus here is that we also won't needlessly leak whatever was
previously in VRAM, but it also probably wasn't sensitive if it was
going through this API.

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

drm/amdgpu: fix semicolon.cocci warnings

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:405:2-3: Unneeded semicolon
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:435:2-3: Unneeded semicolon

 Remove unneeded semicolon.

Generated by: scripts/coccinelle/misc/semicolon.cocci

CC: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new ras workflow control flags

add ras post init function.
Do some initialization after all IP have finished their late init.

Add new member flags which will control the ras work flow.
For now, vbios enable ras for us on boot. That might change in the
future.
So there should be a flag from vbios to tell us if ras is enabled or not
on boot. Looks like there is no such info now.

Other bits of the flags are reserved to control other parts of ras.

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

drm/amdgpu: let ras initialization a little noticeable

add drm info output if ras initialized successfully.
add ras atomfirmware sanity check.

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

drm/amdgpu: Fix lockdep warning more gracely

lockdep need a static key.
Previously we set ignore bit to avoid the warning.
Now call sysfs_attr_init to initialize the static key.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-and-Tested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix ras debugfs data parse

Unzero char is accepted by sscanf, so when data is structure but
unexpectedly return error invalid;

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new member hw_supported

Currently, it is not clear how ras is supported. Both software and
hardware can set the supported. That is confusing.

Fix it by adding new member hw_supported.

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

drm/amdgpu: Fix warning when lockdep is enabled

Set ignore bit to satisfy locpdep.

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

drm/amdgpu: Fix NULL pointer when ta is missing

Ta is optional, so check if ta firmware is loaded or not.

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

drm/amdgpu: fix ras parameter descriptions

The descriptions of modinfo wrongly show two parameters
for each feature(see below). This patch can fix this
incorrect outputs.

parm:           amdgpu_ras_enable:Enable RAS features on the GPU (0 = disable, 1 = enable, -1 = auto (default))
parm:           ras_enable:int
parm:           amdgpu_ras_mask:Mask of RAS features to enable (default 0xffffffff), only valid when ras_enable == 1
parm:           ras_mask:uint

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

drm/amdgpu: export both supported and enabled ras features

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

drm/amdgpu: lookup vbios table to check ecc capability

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

drm/amdgpu: query sram ecc/ecc availability from atombios

query sram ecc capability via amdgpu_atomfirmware_ecc_default_enabled
query ecc availability via amdgpu_atomfirmware_sram_ecc_supported

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

drm/amdgpu: add atomfirmware helper function to query sram ecc caps

sram ecc capability could be get from firmware_capability field in firmwareinfo table

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

drm/amdgpu: add atomfirmware helper function to query ecc status

ecc default status (enabled or disabled) could be get from umc_config field in umc_info table

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

drm/amdgpu: update atomfirmware header with ecc related members

add new umc_info structures and new firmware_capability defines

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

drm/amdgpu: handle ras resume

Suspend will put irq, so resume need get irq back.
And in the same time, skip other ras initialization.

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

drm/amdkfd: add RAS ECC event support (v3)

RAS ECC event will combine with GPU reset event, due to
ECC interrupts are caused by uncorrectable error that triggers
GPU reset.

v2: Fix misleading-indentation warning
v3: fix build with CONFIG_HSA_AMD disabled

Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: add RAS capabilities in topology for Vega20 (v2)

It is to collaborate with HSA_CAPABILITY in libhsakmt.

v2: squash in NULL pointer check

Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add human readable debugfs control support (v2)

Currently, the debugfs control node can't parse bash-like commands.
Now add such support for any tester that uses scripts.

v2: squash in fixes for input validation

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

drm/amdgpu: skip gpu reset when ras error occured

gpu reset is not stable on vega20 A1.

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

drm/amdgpu: add ioctl query for enabled ras features (v2)

Add a query for userspace to check which RAS features
are enabled.

v2: squash in warning fix

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

drm/amdgpu: Add a new flag to AMDGPU_CTX_OP_QUERY_STATE2

Add AMDGPU_CTX_QUERY2_FLAGS_RAS_CE/UE which indicate if any error happened
between previous query and this query.

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

drm/amdgpu: enable ras on gmc9

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

drm/amdgpu: enable ras on gfx9 (v2)

Register ecc interrupts and ecc interrupt handler on gfx9.
Add ras support on gfx9

v2: squash in warning fix

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable ras on sdma4

register IH, enable ras features on sdma.
create sysfs debugfs file for sdma.

Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: reserve bad pages during recovery

Mark vram pages with errors as bad and prevent the driver
from using them.

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

drm/amdgpu: add debugfs ctrl node

allow userspace enable/disable ras

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

drm/amdgpu: add amdgpu_ras.c to support ras (v2)

add obj management.
add feature control.
add debugfs infrastructure.
add sysfs infrastructure.
add IH infrastructure.
add recovery infrastructure.

It is a framework. Other IPs need call amdgpu_ras_xxx function instead of
psp_ras_xxx functions.

v2: squash in warning fixes

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

drm/amdgpu: add psp cmd submit timeout

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