]> asedeno.scripts.mit.edu Git - linux.git/log
linux.git
6 years agodrm/amd/powerplay: enable/disable gfxoff through smu
Huang Rui [Fri, 2 Mar 2018 07:18:54 +0000 (15:18 +0800)]
drm/amd/powerplay: enable/disable gfxoff through smu

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: add gfx off control function
Huang Rui [Fri, 2 Mar 2018 07:10:52 +0000 (15:10 +0800)]
drm/amd/powerplay: add gfx off control function

gfx_off_control is used to be called for sending enabling/disabling gfxoff
message.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: set gfxoff disabled by default
Huang Rui [Fri, 2 Mar 2018 06:40:53 +0000 (14:40 +0800)]
drm/amdgpu: set gfxoff disabled by default

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add gfxoff feature mask
Huang Rui [Fri, 2 Mar 2018 06:16:06 +0000 (14:16 +0800)]
drm/amdgpu: add gfxoff feature mask

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: move PP_FEATURE_MASK to amd_shared header
Huang Rui [Tue, 13 Mar 2018 07:13:46 +0000 (15:13 +0800)]
drm/amdgpu: move PP_FEATURE_MASK to amd_shared header

It will be used not only for powerplay but also on amdgpu part in future
patches. So move it into amd_shared header file.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: send CGPG smc message if PG is enabled for raven
Huang Rui [Thu, 14 Dec 2017 05:38:13 +0000 (13:38 +0800)]
drm/amd/powerplay: send CGPG smc message if PG is enabled for raven

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add setting powergating method for gfx9
Huang Rui [Wed, 6 Dec 2017 01:23:50 +0000 (09:23 +0800)]
drm/amdgpu: add setting powergating method for gfx9

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: revise init_rlc_save_restore_list behavior to support latest register_lis...
Huang Rui [Thu, 21 Dec 2017 08:13:02 +0000 (16:13 +0800)]
drm/amdgpu: revise init_rlc_save_restore_list behavior to support latest register_list_format/register_restore table

RLC save/restore list will be used on CGPG and GFXOFF function, it loads two bin
table of register_list_format/register_restore in RLC firmware.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: cleanup init power gating function
Huang Rui [Thu, 21 Dec 2017 07:48:27 +0000 (15:48 +0800)]
drm/amdgpu: cleanup init power gating function

Remove gfx_v9_0_enable_sck_slow_down_on_power_up/down and CP power gating
enabling functions because they only need to be called on setting power gating
behavior. We keep it in set_powergating callback to enable/disable PG in
late_init.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: enter rlc safe mode before set cgpg
Huang Rui [Thu, 21 Dec 2017 07:03:31 +0000 (15:03 +0800)]
drm/amdgpu: enter rlc safe mode before set cgpg

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add save restore list cntl gpm and srm firmware support
Huang Rui [Mon, 22 Jan 2018 12:48:14 +0000 (20:48 +0800)]
drm/amdgpu: add save restore list cntl gpm and srm firmware support

RLC save/restore list cntl/gpm_mem/srm_mem ucodes are used for CGPG and gfxoff
function.

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add new rlc firmware header format v2.1
Huang Rui [Mon, 22 Jan 2018 09:51:35 +0000 (17:51 +0800)]
drm/amdgpu: add new rlc firmware header format v2.1

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: update psp gfx if header
Huang Rui [Tue, 5 Dec 2017 10:48:48 +0000 (18:48 +0800)]
drm/amdgpu: update psp gfx if header

Signed-off-by: Huang Rui <ray.huang@amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: add registry key to disable ACG
Kenneth Feng [Fri, 20 Apr 2018 05:55:39 +0000 (13:55 +0800)]
drm/amd/powerplay: add registry key to disable ACG

For the dummy ACG fuses,need to disable ACG, otherwise
corruption will be caused.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: print DMA-buf status in debugfs
Christian König [Sun, 25 Mar 2018 08:10:25 +0000 (10:10 +0200)]
drm/amdgpu: print DMA-buf status in debugfs

Just note if a BO was imported/exported.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: header file interface to SMU update
Kenneth Feng [Tue, 17 Apr 2018 13:49:51 +0000 (21:49 +0800)]
drm/amd/powerplay: header file interface to SMU update

update vega12 smu interface.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: simplify bo_va list when vm bo update (v2)
Junwei Zhang [Thu, 19 Apr 2018 05:17:26 +0000 (13:17 +0800)]
drm/amdgpu: simplify bo_va list when vm bo update (v2)

v2: fix compiling warning

Signed-off-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: init gfx9 aperture settings
Flora Cui [Wed, 18 Apr 2018 09:12:19 +0000 (17:12 +0800)]
drm/amdgpu: init gfx9 aperture settings

fix settings.

Signed-off-by: Flora Cui <Flora.Cui@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Fix NULL point check error in smu_set_watermarks_for_clocks_ranges
Rex Zhu [Thu, 19 Apr 2018 04:40:15 +0000 (12:40 +0800)]
drm/amd/pp: Fix NULL point check error in smu_set_watermarks_for_clocks_ranges

It is caused by
'commit d6c9a7dc86cd ("drm/amd/pp: Move common code to smu_helper.c")'

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix list not initialized
Chunming Zhou [Wed, 18 Apr 2018 10:35:09 +0000 (18:35 +0800)]
drm/amdgpu: fix list not initialized

Otherwise, cpu stuck for 22s with kernel panic.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: optionally do a writeback but don't invalidate TC for IB fences
Marek Olšák [Tue, 3 Apr 2018 17:05:03 +0000 (13:05 -0400)]
drm/amdgpu: optionally do a writeback but don't invalidate TC for IB fences

There is a new IB flag that enables this new behavior.
Full invalidation is unnecessary for RELEASE_MEM and doesn't make sense
when draw calls from two adjacent gfx IBs run in parallel. This will be
the new default for Mesa.

v2: bump the version

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: handle domain mask checking v2
Chunming Zhou [Tue, 17 Apr 2018 10:34:40 +0000 (18:34 +0800)]
drm/amdgpu: handle domain mask checking v2

if domain is illegal, we should return error.
v2:
  remove duplicated domain checking.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: set preferred_domain independent of fallback handling
Chunming Zhou [Tue, 17 Apr 2018 03:52:53 +0000 (11:52 +0800)]
drm/amdgpu: set preferred_domain independent of fallback handling

When GEM needs to fallback to GTT for VRAM BOs we still want the
preferred domain to be untouched so that the BO has a cance to move back
to VRAM in the future.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: limit reg_write_reg_wait workaround to SRIOV v2
Christian König [Tue, 17 Apr 2018 12:47:42 +0000 (14:47 +0200)]
drm/amdgpu: limit reg_write_reg_wait workaround to SRIOV v2

Turned out that this locks up some bare metal Vega10.

v2: fix stupid typo

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/scheduler: move last_sched fence updating prior to job popping (v2)
Pixel Ding [Wed, 18 Apr 2018 08:37:40 +0000 (04:37 -0400)]
drm/scheduler: move last_sched fence updating prior to job popping (v2)

Make sure main thread won't update last_sched fence when entity
is cleanup.

Fix a racing issue which is caused by putting last_sched fence
twice. Running vulkaninfo in tight loop can produce this issue
as seeing wild fence pointer.

v2: squash in build fix (Christian)

Signed-off-by: Pixel Ding <Pixel.Ding@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Monk Liu <Monk.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/scheduler: always put last_sched fence in entity_fini
Pixel Ding [Wed, 18 Apr 2018 08:33:26 +0000 (04:33 -0400)]
drm/scheduler: always put last_sched fence in entity_fini

Fix the potential memleak since scheduler main thread always
hold one last_sched fence.

Signed-off-by: Pixel Ding <Pixel.Ding@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: print the vbios version in the debugfs firmware info
Alex Deucher [Tue, 17 Apr 2018 13:55:44 +0000 (08:55 -0500)]
drm/amdgpu: print the vbios version in the debugfs firmware info

Useful for info gathering about what firmwares are in use in
the driver.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Fix bug voltage can't be OD separately on VI
Rex Zhu [Tue, 17 Apr 2018 09:26:26 +0000 (17:26 +0800)]
drm/amd/pp: Fix bug voltage can't be OD separately on VI

Make sure to update the MCLK and SCLK flags when setting the VDDC
flags due to dependencies.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Remove dead interface
Rex Zhu [Fri, 13 Apr 2018 08:16:49 +0000 (16:16 +0800)]
drm/amd/pp: Remove dead interface

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: Reserved vram for smu to save debug info.
Rex Zhu [Fri, 13 Apr 2018 08:13:41 +0000 (16:13 +0800)]
drm/amdgpu: Reserved vram for smu to save debug info.

v2: check reserved vram size before allocate.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: use amdgpu_bo_param for amdgpu_bo_create v2
Chunming Zhou [Mon, 16 Apr 2018 10:27:50 +0000 (18:27 +0800)]
drm/amdgpu: use amdgpu_bo_param for amdgpu_bo_create v2

After that, we can easily add new parameter when need.

v2:
a) rebase.
b) Initialize struct amdgpu_bo_param, future new
member could only be used in some one case, but all member
should have its own initial value.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com> (v1)
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Cc: christian.koenig@amd.com
Cc: Felix.Kuehling@amd.com
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add amdgpu_bo_param
Chunming Zhou [Mon, 16 Apr 2018 09:57:19 +0000 (17:57 +0800)]
drm/amdgpu: add amdgpu_bo_param

amdgpu_bo_create has too many parameters, and used in
too many places. Collect them to one structure.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Use dig enable to determine fast boot optimization.
Yongqiang Sun [Mon, 9 Apr 2018 20:15:20 +0000 (16:15 -0400)]
drm/amd/display: Use dig enable to determine fast boot optimization.

Linux doesn't know lid state, better to check dig enable
value from register.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: add missing colorspace for set black color
Yue Hin Lau [Mon, 9 Apr 2018 18:46:32 +0000 (14:46 -0400)]
drm/amd/display: add missing colorspace for set black color

Signed-off-by: Yue Hin Lau <Yuehin.Lau@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: add calculated clock logging to DTN
Dmytro Laktyushkin [Wed, 4 Apr 2018 20:03:38 +0000 (16:03 -0400)]
drm/amd/display: add calculated clock logging to DTN

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: add rq/dlg/ttu to dtn log
Dmytro Laktyushkin [Thu, 29 Mar 2018 12:43:02 +0000 (08:43 -0400)]
drm/amd/display: add rq/dlg/ttu to dtn log

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Check SCRATCH reg to determine S3 resume.
Yongqiang Sun [Sat, 7 Apr 2018 01:38:10 +0000 (21:38 -0400)]
drm/amd/display: Check SCRATCH reg to determine S3 resume.

Use lid state only to determine fast boot optimization is not enough.
For S3/Resume, due to bios isn't involved in boot, eDP wasn't
light up, while lid state is open, if do fast boot optimization,
eDP panel will skip enable link and result in black screen after boot.
And becasue of bios isn't involved, no matter UEFI or Legacy boot,
BIOS_SCRATCH_3 value should be 0, use this to determine the case.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: HDMI has no sound after Panel power off/on
Charlene Liu [Sat, 7 Apr 2018 03:03:12 +0000 (23:03 -0400)]
drm/amd/display: HDMI has no sound after Panel power off/on

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Move DCC support functions into dchubbub
Eric Bernstein [Thu, 5 Apr 2018 21:09:20 +0000 (17:09 -0400)]
drm/amd/display: Move DCC support functions into dchubbub

Added dchububu.h header file for common enum/struct definitions.
Added new interface functions get_dcc_compression_cap,
dcc_support_swizzle, dcc_support_pixel_format.

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Do not create memory allocation if stats not enabled
Anthony Koo [Fri, 6 Apr 2018 16:07:19 +0000 (12:07 -0400)]
drm/amd/display: Do not create memory allocation if stats not enabled

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Check lid state to determine fast boot optimization.
Yongqiang Sun [Wed, 4 Apr 2018 21:27:18 +0000 (17:27 -0400)]
drm/amd/display: Check lid state to determine fast boot optimization.

For legacy enable boot up with lid closed, eDP information couldn't be
read correctly via SBIOS_SCRATCH_3 results in eDP cannot be light up
properly when open lid.
Check lid state instead can resolve the issue.

Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: change dml init to use default structs
Dmytro Laktyushkin [Fri, 23 Mar 2018 19:25:43 +0000 (15:25 -0400)]
drm/amd/display: change dml init to use default structs

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: fix segfault on insufficient TG during validation
Dmytro Laktyushkin [Thu, 29 Mar 2018 20:39:10 +0000 (16:39 -0400)]
drm/amd/display: fix segfault on insufficient TG during validation

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Fix regamma not affecting full-intensity color values
Leo (Sunpeng) Li [Tue, 3 Apr 2018 20:07:16 +0000 (16:07 -0400)]
drm/amd/display: Fix regamma not affecting full-intensity color values

Hardware understands the regamma LUT as a piecewise linear function,
with points spaced exponentially along the range. We previously
programmed the LUT for range [2^-10, 2^0). This causes (normalized)
color values of 1 (=2^0) to miss the programmed LUT, and fall onto the
end region.

For DCE, the end region is extrapolated using a single (base, slope)
pair, using the max y-value from the last point in the curve as base.
This presents a problem, since this value affects all three color
channels. Scaling down the intensity of say - the blue regamma curve -
will not affect it's end region. This is especially noticiable when
using RedShift. It scales down the blue and green channels, but leaves
full-intensity colors unshifted.

Therefore, extend the range to cover [2^-10, 2^1) by programming another
hardware segment, containing only one point. That way, we won't be
hitting the end region.

Note that things are a bit different for DCN, since the end region can
be set per-channel.

Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Move dp_pixel_encoding_type to stream_encoder include
Eric Bernstein [Tue, 3 Apr 2018 15:23:11 +0000 (11:23 -0400)]
drm/amd/display: Move dp_pixel_encoding_type to stream_encoder include

Signed-off-by: Eric Bernstein <eric.bernstein@amd.com>
Reviewed-by: Nikola Cornij <Nikola.Cornij@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: fix brightness level after resume from suspend
Roman Li [Thu, 29 Mar 2018 14:56:17 +0000 (10:56 -0400)]
drm/amd/display: fix brightness level after resume from suspend

Adding missing call to cache current backlight values.
Otherwise the brightness resets to default value on resume.

Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: dal 3.1.42
Eric Yang [Tue, 3 Apr 2018 15:36:14 +0000 (11:36 -0400)]
drm/amd/display: dal 3.1.42

Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Don't program bypass on linear regamma LUT
Harry Wentland [Thu, 12 Apr 2018 20:37:09 +0000 (16:37 -0400)]
drm/amd/display: Don't program bypass on linear regamma LUT

Even though this is required for degamma since DCE HW only supports a
couple predefined LUTs we can just program the LUT directly for regamma.

This fixes dark screens which occurs when we program regamma to bypass
while degamma is using srgb LUT.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: defer test IBs on the rings at boot (V3)
Shirish S [Mon, 16 Apr 2018 06:47:57 +0000 (12:17 +0530)]
drm/amdgpu: defer test IBs on the rings at boot (V3)

amdgpu_ib_ring_tests() runs test IB's on rings at boot
contributes to ~500 ms of amdgpu driver's boot time.

This patch defers it and ensures that its executed
in amdgpu_info_ioctl() if it wasn't scheduled.

V2: Use queue_delayed_work() & flush_delayed_work().
V3: removed usage of separate wq, ensure ib tests is
    run before enabling clockgating.

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Adding a function to store cc6 data in SMU10
Mikita Lipski [Wed, 11 Apr 2018 20:25:26 +0000 (16:25 -0400)]
drm/amd/pp: Adding a function to store cc6 data in SMU10

Filling the smu10_store_cc6_data based on the implementation
of Windows Powerplay.

There is an uncertainty with one of the parameters passed to the function
pstate_switch_disable - is not a part of smu10 private data structure.
So in the function its just ignored.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Adding set_watermarks_for_clocks_ranges for SMU10
Mikita Lipski [Tue, 10 Apr 2018 17:45:00 +0000 (13:45 -0400)]
drm/amd/pp: Adding set_watermarks_for_clocks_ranges for SMU10

The function is never implemented for raven on linux.
It follows similair implementation as on windows.

SMU still needs to notify SMC and copy WM table, which is added
here. But on other Asics such as Vega this step is not implemented.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: re-validate per VM BOs if required v2
Christian König [Mon, 19 Mar 2018 10:49:14 +0000 (11:49 +0100)]
drm/amdgpu: re-validate per VM BOs if required v2

If a per VM BO ends up in a allowed domain it never moves back into the
prefered domain.

v2: move the extra handling into amdgpu_vm_bo_update when we exit the
    state machine. Make memory type handling generic.

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>
6 years agodrm/amdgpu: revert "Don't change preferred domian when fallback GTT v6"
Christian König [Tue, 10 Apr 2018 11:42:38 +0000 (13:42 +0200)]
drm/amdgpu: revert "Don't change preferred domian when fallback GTT v6"

This reverts commit 7d1ca1325260a9e9329b10a21e3692e6f188936f.

Makes fallback handling to complicated. This is just a feature for the
GEM interface and shouldn't leak into the core BO create function.

The intended change to preserve the preferred domains is implemented in
a follow up patch.

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>
6 years agodrm/amdgpu: revert "add new bo flag that indicates BOs don't need fallback (v2)"
Christian König [Tue, 10 Apr 2018 11:42:29 +0000 (13:42 +0200)]
drm/amdgpu: revert "add new bo flag that indicates BOs don't need fallback (v2)"

This reverts commit 6f51d28bfe8e1a676de5cd877639245bed3cc818.

Makes fallback handling to complicated. This is just a feature for the
GEM interface and shouldn't leak into the core BO create function.

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>
6 years agodrm/amd/pp: Move common code to smu_helper.c
Rex Zhu [Sun, 8 Apr 2018 08:57:55 +0000 (16:57 +0800)]
drm/amd/pp: Move common code to smu_helper.c

Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Clear smu response register before send smu message
Rex Zhu [Wed, 11 Apr 2018 10:11:49 +0000 (18:11 +0800)]
drm/amd/pp: Clear smu response register before send smu message

smu firmware do not update response register immediately under
some delay tasks, we may read out the original value.

so need to clear the register before send smu message.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Remove struct pp_gpu_power
Rex Zhu [Wed, 4 Apr 2018 07:37:35 +0000 (15:37 +0800)]
drm/amd/pp: Remove struct pp_gpu_power

Currently smu only calculate average gpu power in real time.

for vddc/vddci/max power,
User need to set start time and end time, firmware can calculate
the average vddc/vddci/max power. but the type of return values
is not unified. For Vi, return type is uint.
For vega, return type is float.

so this struct can't be suitable for all asics.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Refine get_gpu_power for VI
Rex Zhu [Wed, 4 Apr 2018 06:17:09 +0000 (14:17 +0800)]
drm/amd/pp: Refine get_gpu_power for VI

pkgpwr is the average gpu power of 100ms. it is calculated by
firmware in real time.

1. we can send smu message PPSMC_MSG_GetCurrPkgPwr to read currentpkgpwr directly.

2. On Fiji/tonga/bonaire/hawwii, without PPSMC_MSG_GetCurrPkgPwr support.
   Send PPSMC_MSG_PmStatusLogStart/Sample to let smu write currentpkgpwr
   to ixSMU_PM_STATUS_94. driver can read pkgpwr from ixSMU_PM_STATUS_94.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agoRevert "drm/amd/powerply: fix power reading on Fiji"
Rex Zhu [Wed, 4 Apr 2018 06:11:45 +0000 (14:11 +0800)]
Revert "drm/amd/powerply: fix power reading on Fiji"

we don't have limit of [50ms, 4sec] sampling period.
smu calculate average gpu power in real time.
we can read average gpu power through smu message or
read special register.

This reverts commit 462d8dcc9fec0d89f1ff6a1f93f1d4f670878c71.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/gpu-sched: fix force APP kill hang(v4)
Emily Deng [Mon, 16 Apr 2018 02:07:02 +0000 (10:07 +0800)]
drm/gpu-sched: fix force APP kill hang(v4)

issue:
there are VMC page fault occurred if force APP kill during
3dmark test, the cause is in entity_fini we manually signal
all those jobs in entity's queue which confuse the sync/dep
mechanism:

1)page fault occurred in sdma's clear job which operate on
shadow buffer, and shadow buffer's Gart table is cleaned by
ttm_bo_release since the fence in its reservation was fake signaled
by entity_fini() under the case of SIGKILL received.

2)page fault occurred in gfx' job because during the lifetime
of gfx job we manually fake signal all jobs from its entity
in entity_fini(), thus the unmapping/clear PTE job depend on those
result fence is satisfied and sdma start clearing the PTE and lead
to GFX page fault.

fix:
1)should at least wait all jobs already scheduled complete in entity_fini()
if SIGKILL is the case.

2)if a fence signaled and try to clear some entity's dependency, should
set this entity guilty to prevent its job really run since the dependency
is fake signaled.

v2:
splitting drm_sched_entity_fini() into two functions:
1)The first one is does the waiting, removes the entity from the
runqueue and returns an error when the process was killed.
2)The second one then goes over the entity, install it as
completion signal for the remaining jobs and signals all jobs
with an error code.

v3:
1)Replace the fini1 and fini2 with better name
2)Call the first part before the VM teardown in
amdgpu_driver_postclose_kms() and the second part
after the VM teardown
3)Keep the original function drm_sched_entity_fini to
refine the code.

v4:
1)Rename entity->finished to entity->last_scheduled;
2)Rename drm_sched_entity_fini_job_cb() to
drm_sched_entity_kill_jobs_cb();
3)Pass NULL to drm_sched_entity_fini_job_cb() if -ENOENT;
4)Replace the type of entity->fini_status with "int";
5)Remove the check about entity->finished.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: Free VGA stolen memory as soon as possible.
Andrey Grodzovsky [Fri, 6 Apr 2018 19:54:10 +0000 (14:54 -0500)]
drm/amdgpu: Free VGA stolen memory as soon as possible.

Reserved VRAM is used to avoid overriding pre OS FB.
Once our display stack takes over we don't need the reserved
VRAM anymore.

v2:
Remove comment, we know actually why we need to reserve the stolen VRAM.
Fix return type for amdgpu_ttm_late_init.
v3:
Return 0 in amdgpu_bo_late_init, rebase on changes to previous patch
v4: rebase
v5:
For GMC9 reserve always just 9M and keep the stolem memory around
until GART table curruption on S3 resume is resolved.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/gmc: steal the appropriate amount of vram for fw hand-over (v3)
Alex Deucher [Fri, 6 Apr 2018 19:54:09 +0000 (14:54 -0500)]
drm/amdgpu/gmc: steal the appropriate amount of vram for fw hand-over (v3)

Steal 9 MB for vga emulation and fb if vga is enabled, otherwise,
steal enough to cover the current display size as set by the vbios.

If no memory is used (e.g., secondary or headless card), skip
stolen memory reserve.

v2: skip reservation if vram is limited, address Christian's comments
v3: squash in fix from Harry

Reviewed-and-Tested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> (v2)
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
6 years agodrm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu tlb flush
Alex Deucher [Tue, 27 Mar 2018 22:10:56 +0000 (17:10 -0500)]
drm/amdgpu/gmc9: use amdgpu_ring_emit_reg_write_reg_wait in gpu tlb flush

Use amdgpu_ring_emit_reg_write_reg_wait.  On engines that support it,
it provides a write and wait in a single packet which avoids a missed
ack if a world switch happens between the request and waiting for the
ack.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/vcn1: add emit_reg_write_reg_wait ring callback
Alex Deucher [Tue, 27 Mar 2018 22:06:52 +0000 (17:06 -0500)]
drm/amdgpu/vcn1: add emit_reg_write_reg_wait ring callback

This adds support for writing and reading back using the
helper since the engines doesn't have a oneshot packet.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/vce4: add emit_reg_write_reg_wait ring callback
Alex Deucher [Tue, 27 Mar 2018 22:06:33 +0000 (17:06 -0500)]
drm/amdgpu/vce4: add emit_reg_write_reg_wait ring callback

This adds support for writing and reading back using the
helper since the engines doesn't have a oneshot packet.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/uvd7: add emit_reg_write_reg_wait ring callback
Alex Deucher [Tue, 27 Mar 2018 22:05:19 +0000 (17:05 -0500)]
drm/amdgpu/uvd7: add emit_reg_write_reg_wait ring callback

This adds support for writing and reading back using the
helper since the engines doesn't have a oneshot packet.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/sdma4: add emit_reg_write_reg_wait ring callback (v2)
Alex Deucher [Tue, 27 Mar 2018 21:51:41 +0000 (16:51 -0500)]
drm/amdgpu/sdma4: add emit_reg_write_reg_wait ring callback (v2)

This adds support for writing and reading back in a single
oneshot packet.  This is needed to send a tlb invalidation
and wait for ack in a single operation.

v2: squash sdma hang fix into this patch

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
6 years agodrm/amdgpu/gfx9: add emit_reg_write_reg_wait ring callback (v2)
Alex Deucher [Tue, 27 Mar 2018 20:07:50 +0000 (15:07 -0500)]
drm/amdgpu/gfx9: add emit_reg_write_reg_wait ring callback (v2)

This adds support for writing and reading back in a single
oneshot packet.  This is needed to send a tlb invalidation
and wait for ack in a single operation.

v2: squash the gfx ring stall fix

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
6 years agodrm/amdgpu: add emit_reg_write_reg_wait ring callback
Alex Deucher [Tue, 27 Mar 2018 16:58:14 +0000 (11:58 -0500)]
drm/amdgpu: add emit_reg_write_reg_wait ring callback

This callback writes a value to a register and then reads
back another register and waits for a value in a single
operation.

Provide a helper function using two operations for engines
that don't support this opertion.

Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/powerplay: rename smu7_upload_mc_firmware
Alex Deucher [Wed, 11 Apr 2018 23:09:39 +0000 (18:09 -0500)]
drm/amdgpu/powerplay: rename smu7_upload_mc_firmware

It doesn't actually upload any firmware is just
checks the version.  The actual upload happens in
the gmc modules.

Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/powerplay: fix smu7_get_memory_type for fiji
Alex Deucher [Wed, 11 Apr 2018 22:57:13 +0000 (17:57 -0500)]
drm/amdgpu/powerplay: fix smu7_get_memory_type for fiji

Fiji uses a different register than other smu7 asics, but
we already have this info in the base driver so just
use that.

Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agoRevert "drm/amd/display: disable CRTCs with NULL FB on their primary plane (V2)"
Harry Wentland [Thu, 12 Apr 2018 14:51:52 +0000 (10:51 -0400)]
Revert "drm/amd/display: disable CRTCs with NULL FB on their primary plane (V2)"

This seems to cause flickering and lock-ups for a wide range of users.
Revert until we've found a proper fix for the flickering and lock-ups.

This reverts commit 36cc549d59864b7161f0e23d710c1c4d1b9cf022.

Cc: Shirish S <shirish.s@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agoRevert "drm/amd/display: fix dereferencing possible ERR_PTR()"
Harry Wentland [Thu, 12 Apr 2018 14:51:51 +0000 (10:51 -0400)]
Revert "drm/amd/display: fix dereferencing possible ERR_PTR()"

This reverts commit cd2d6c92a8e39d7e50a5af9fcc67d07e6a89e91d.

Cc: Shirish S <shirish.s@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu/gfx9: cache DB_DEBUG2 and make it available to userspace
Alex Deucher [Tue, 10 Apr 2018 15:15:26 +0000 (10:15 -0500)]
drm/amdgpu/gfx9: cache DB_DEBUG2 and make it available to userspace

Userspace needs to query this value to work around a hw bug in
certain cases.

Acked-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/ttm: keep a reference to transfer pipelined BOs
Christian König [Fri, 9 Mar 2018 12:39:47 +0000 (13:39 +0100)]
drm/ttm: keep a reference to transfer pipelined BOs

Make sure the transfered BO is never destroy before the transfer BO.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Remove useless smu7 running state check
Rex Zhu [Wed, 4 Apr 2018 10:41:08 +0000 (18:41 +0800)]
drm/amd/pp: Remove useless smu7 running state check

Only check smc running state before start smu.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Remove dead function in smu7_smumgr.c
Rex Zhu [Wed, 4 Apr 2018 10:33:15 +0000 (18:33 +0800)]
drm/amd/pp: Remove dead function in smu7_smumgr.c

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: initialzie the dpm intial enabled state
Kenneth Feng [Tue, 10 Apr 2018 09:05:36 +0000 (17:05 +0800)]
drm/amd/powerplay: initialzie the dpm intial enabled state

To expose the right dpm levels to the sysfs

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/powerplay: Get more than 8 level gfxclk states
Kenneth Feng [Wed, 4 Apr 2018 07:17:22 +0000 (15:17 +0800)]
drm/amd/powerplay: Get more than 8 level gfxclk states

To apply on Vega12 for more than 8 gfx dpm levels

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agoamd/powerplay: implement the vega12_force_clock_level interface
Kenneth Feng [Mon, 9 Apr 2018 06:53:51 +0000 (14:53 +0800)]
amd/powerplay: implement the vega12_force_clock_level interface

pp_dpm_sclk/pp_dpm_mclk in sysfs implemented to force
gfxclk/uclk dpm level for Vega12

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: Add APU support in vi_set_vce_clocks
Rex Zhu [Tue, 10 Apr 2018 09:49:56 +0000 (17:49 +0800)]
drm/amdgpu: Add APU support in vi_set_vce_clocks

1. fix set vce clocks failed on Cz/St
   which lead 1s delay when boot up.
2. remove the workaround in vce_v3_0.c

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
6 years agodrm/amdgpu: Add APU support in vi_set_uvd_clocks
Rex Zhu [Tue, 10 Apr 2018 09:17:22 +0000 (17:17 +0800)]
drm/amdgpu: Add APU support in vi_set_uvd_clocks

fix the issue set uvd clock failed on CZ/ST
which lead 1s delay when boot up.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
6 years agodrm/amd/pp: Remove unnecessary forward declaration
Rex Zhu [Tue, 10 Apr 2018 02:58:43 +0000 (10:58 +0800)]
drm/amd/pp: Remove unnecessary forward declaration

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Move same macro definitions to hwmgr.h
Rex Zhu [Tue, 10 Apr 2018 02:58:43 +0000 (10:58 +0800)]
drm/amd/pp: Move same macro definitions to hwmgr.h

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: remove dummy is_blanked() to optimise boot time
Shirish S [Wed, 28 Mar 2018 06:52:22 +0000 (12:22 +0530)]
drm/amd/display: remove dummy is_blanked() to optimise boot time

is_blanked() hook is a dummy one for underlay pipe, hence
when called, it loops for ~300ms at boot.

This patch removes this dummy call and adds missing checks.

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Don't spam debug messages
Harry Wentland [Mon, 9 Apr 2018 18:04:56 +0000 (14:04 -0400)]
drm/amd/display: Don't spam debug messages

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Remove PRE_VEGA flag
Harry Wentland [Mon, 9 Apr 2018 18:27:46 +0000 (14:27 -0400)]
drm/amd/display: Remove PRE_VEGA flag

We enabled this upstream by default now and no longer need the flag.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Fix 64-bit division in hwss_edp_power_control
Harry Wentland [Tue, 10 Apr 2018 20:08:44 +0000 (16:08 -0400)]
drm/amd/display: Fix 64-bit division in hwss_edp_power_control

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix and cleanup cpu visible VRAM handling
Christian König [Thu, 5 Apr 2018 14:42:03 +0000 (16:42 +0200)]
drm/amdgpu: fix and cleanup cpu visible VRAM handling

The detection if a BO was placed in CPU visible VRAM was incorrect.

Fix it and merge it with the correct detection in amdgpu_ttm.c

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: use ctx bytes_moved
Christian König [Thu, 5 Apr 2018 12:46:41 +0000 (14:46 +0200)]
drm/amdgpu: use ctx bytes_moved

Instead of the global (inaccurate) counter.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: fix the wrong readout engine clock in deep sleep
Evan Quan [Tue, 10 Apr 2018 05:05:49 +0000 (13:05 +0800)]
drm/amd/pp: fix the wrong readout engine clock in deep sleep

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: use soc15 common macros instead of vega10 specific
Evan Quan [Tue, 10 Apr 2018 04:32:16 +0000 (12:32 +0800)]
drm/amd/pp: use soc15 common macros instead of vega10 specific

pp_soc15.h is vega10 specific. Update powerplay code to use soc15 common
macros defined in soc15_common.h.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: add MP1 and THM hw ip base reg offset
Evan Quan [Tue, 10 Apr 2018 04:30:59 +0000 (12:30 +0800)]
drm/amdgpu: add MP1 and THM hw ip base reg offset

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/pp: Refine pp_atomfwctrl_get_vbios_bootup_values
Rex Zhu [Wed, 4 Apr 2018 04:36:57 +0000 (12:36 +0800)]
drm/amd/pp: Refine pp_atomfwctrl_get_vbios_bootup_values

In order to share pp_atomfwctrl_get_vbios_bootup_values
on asics with different BIOS_CLKID.
Not call function pp_atomfwctrl_get_clk_information_by_clkid in
pp_atomfwctrl_get_vbios_bootup_values.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix null pointer panic with direct fw loading on gpu reset
Huang Rui [Sun, 8 Apr 2018 06:39:18 +0000 (14:39 +0800)]
drm/amdgpu: fix null pointer panic with direct fw loading on gpu reset

When system uses fw direct loading, then psp context structure won't be
initiliazed. And it is also unable to execute mode reset.

[  434.601474] amdgpu 0000:0c:00.0: GPU reset begin!
[  434.694326] amdgpu 0000:0c:00.0: GPU reset
[  434.743152] BUG: unable to handle kernel NULL pointer dereference at
0000000000000058
[  434.838474] IP: psp_gpu_reset+0xc/0x30 [amdgpu]
[  434.893532] PGD 406ed9067
[  434.893533] P4D 406ed9067
[  434.926376] PUD 400b46067
[  434.959217] PMD 0
[  435.033379] Oops: 0000 [#1] SMP
[  435.072573] Modules linked in: amdgpu(OE) chash(OE) gpu_sched(OE) ttm(OE)
drm_kms_helper(OE) drm(OE) fb_sys_fops syscopyarea sysfillrect sysimgblt
rpcsec_gss_krb5 auth_rpcgss nfsv4 nfs lockd grace fscache snd_hda_codec_realtek
snd_hda_codec_generic snd_hda_codec_hdmi snd_hda_intel snd_hda_codec
snd_hda_core snd_hwdep snd_pcm edac_mce_amd snd_seq_midi snd_seq_midi_event
kvm_amd snd_rawmidi kvm irqbypass crct10dif_pclmul crc32_pclmul snd_seq
ghash_clmulni_intel snd_seq_device pcbc snd_timer eeepc_wmi aesni_intel snd
asus_wmi aes_x86_64 sparse_keymap crypto_simd glue_helper joydev soundcore
wmi_bmof cryptd video i2c_piix4 shpchp 8250_dw i2c_designware_platform mac_hid
i2c_designware_core sunrpc parport_pc ppdev lp parport autofs4 hid_generic igb
usbhid dca ptp mxm_wmi pps_core ahci hid i2c_algo_bit
[  435.931754]  libahci wmi

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/radeon: add PX quirk for Asus K73TK
Nico Sneck [Sat, 7 Apr 2018 15:13:04 +0000 (15:13 +0000)]
drm/radeon: add PX quirk for Asus K73TK

With this the dGPU turns on correctly.

Signed-off-by: Nico Sneck <nicosneck@hotmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
6 years agodrm/amd/display: Fix memleak on input transfer function
Leo (Sunpeng) Li [Wed, 4 Apr 2018 20:01:30 +0000 (16:01 -0400)]
drm/amd/display: Fix memleak on input transfer function

Input transfer function creation is now done when the plane is created.
This is done within the following change:

    Author: Anthony Koo <Anthony.Koo@amd.com>
        drm/amd/display: Have DC manage its own allocation of gamma

Therefore, we no longer need to create it when filling in the plane
attributes.

Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Fix dim display on DCE11
Leo (Sunpeng) Li [Thu, 29 Mar 2018 21:04:12 +0000 (17:04 -0400)]
drm/amd/display: Fix dim display on DCE11

Before programming the input gamma, check that we're not using the
identity correction.

Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Have DC manage its own allocation of gamma
Anthony Koo [Tue, 27 Mar 2018 20:43:56 +0000 (16:43 -0400)]
drm/amd/display: Have DC manage its own allocation of gamma

Creating plane will also allocate gamma and input TF
Creating stream will also allocate outputTF

Fix issue with gamma not applied
OS may call SetGamma before surface committed, so need to store
in target and apply later.

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Fix structure initialization of hdmi_info_packet
Anthony Koo [Wed, 28 Mar 2018 03:12:21 +0000 (23:12 -0400)]
drm/amd/display: Fix structure initialization of hdmi_info_packet

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