Alvin Lee [Sat, 30 Apr 2022 00:41:10 +0000 (20:41 -0400)]
drm/amd/display: Implement DTBCLK ref switching on dcn32
[WHY & HOW]
Implements DTB ref clock switching with reg key default to OFF.
Refactors dccg DTBCLK logic to not store redundant state information
dccg. Also removes duplicated functions that should be inherited from
other dcn versions.
Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Previously we used to send FCLK P-state enable messages upon each call
to update_clocks based on dml output. This resulted in increased message
transactions between DC and PMFW.
[How]
Update the code to check safe_to_lower status and send the message based
on dml input only on boot. This reduces message transactions. Also
remove other unwanted code based on current code status.
Signed-off-by: Chaitanya Dhere <chaitanya.dhere@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jun Lei [Sun, 20 Feb 2022 19:33:56 +0000 (14:33 -0500)]
drm/amd/display: Introduce new update_clocks logic
[why]
DCN has sidebands to control some clocks, it is useful for clk_mgr to
always update the clocks it explicitly controls rather than skip them
because it enables more configurations to work without SMU
[how]
only skip handling clocks where SMU manages the frequency for clocks
with DENTIST sideband (DISP/DPP), only skip the voltage request when SMU
not available, but otherwise proceed normally
Signed-off-by: Jun Lei <Jun.Lei@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Jun Lei [Sun, 20 Feb 2022 18:58:51 +0000 (13:58 -0500)]
drm/amd/display: add new pixel rate programming
[why]
New dividers in DCCG need to be programmed depending
on encoder/stream type since pixels per clock in
OTG/DIO is different
DIO also needs additional programming depending on
pixels per clock
Signed-off-by: Jun Lei <Jun.Lei@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Samson Tam [Thu, 7 Apr 2022 16:04:42 +0000 (12:04 -0400)]
drm/amd/display: do not override CURSOR_REQ_MODE when SubVP is not enabled
[Why]
HUBP_UNBOUNDED_REQ_MODE and CURSOR_REQ_MODE are normally set together.
In hubp32_prepare_subvp_buffering() call, CURSOR_REQ_MODE is set based on
whether SubVP is enabled or not. For non MPO case, both REQ_MODE
registers are set to 1. But since SubVP is not enabled, then
CURSOR_REQ_MODE is set to 0, overriding the previous value.
[How]
Do not set CURSOR_REQ_MODE to 0 if SubVP is not enabled. This
will allow CURSOR_REQ_MODE to stay as 1 in the non MPO case.
Add note to follow up and check case for single pipe MPO and
SubVP enabled as this would cause both REQ_MODE registers to be
set to 0 but SubVP enabled would override CURSOR_REQ_MODE to 1.
Signed-off-by: Samson Tam <Samson.Tam@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Deucher [Thu, 26 May 2022 16:04:38 +0000 (12:04 -0400)]
drm/amdgpu/swsmu: use new register offsets for smu_cmn.c
Use the per asic offsets so the we don't have to have
asic specific logic in the common code.
Reviewed-by: Luben Tuikov <luben.tuikov@amd.com> Reviewed-by: Yang Wang <kevinyang.wang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
VBIOS default clock value was halved, so the hardcoded dtb value should be
halved as well.
dtb clock should come from SMU eventually, but now dtb clock switching is not
fully supported yet in SMU.
Halve the dtb hardcoded value for now to have UHBR10 light up. Will rely on
SMU for dtb clock switching once available. The w/a is for DCN32 only, DCN321
should adopt the original value.
Xiaojian Du [Fri, 20 May 2022 13:33:29 +0000 (21:33 +0800)]
drm/amdgpu: fix sdma doorbell issue on SDMA v6.0 and NBIO v7.7
This patch will fix sdma doorbell issue on SDMA v6.0 and NBIO v7.7.0.
NBIO v7.7.0 uses a new reg function -- Common SDMA to allow a common
doorbell range for all SDMA queues, this is different to the old
NBIO version. This patch will add configuration for CSDMA and enable
SDMA doorbell function.
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com> Reviewed-by: Tim Huang <Tim.Huang@amd.com> Reviewed-by: Huang Rui <ray.huang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Xiaojian Du [Fri, 20 May 2022 14:27:57 +0000 (22:27 +0800)]
drm/amdgpu: add CSDMA reg offsets for NBIO v7.7.0
This patch will add CSDMA reg offsets for NBIO v7.7.0
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com> Acked-by: Roman Li <roman.li@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: Ensure that DMCUB fw in use is loaded by DC and not VBIOS
[Why?]
On wake from S3/S4, driver checks if DMUB is initialized. On S4 VBIOS loads
DMUB, and driver does not reload as it appears to be initialized already.
[How?]
Add a check for the DAL_FW bit to ensure that loaded FW is from driver and
not VBIOS.
Aurabindo Pillai [Mon, 21 Feb 2022 20:33:05 +0000 (15:33 -0500)]
drm/amd: Add GFX11 modifiers support to AMDGPU (v3)
GFX11 IP introduces new tiling mode. Various combinations of DCC
settings are possible and the most preferred settings must be exposed
for optimal use of the hardware.
add_gfx11_modifiers() is based on recommendation from Marek for the
preferred tiling modifier that are most efficient for the hardware.
v2: microtiling fix noticed by Marek
v3: keep Z tiling check
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Aurabindo Pillai [Thu, 24 Feb 2022 00:05:09 +0000 (19:05 -0500)]
drm/amd/display: Add DM support for DCN32/DCN321
Add Display Manager specific changes for DCN3.2.x. DM
handles the interaction between the core DC modesetting
code and the drm modesetting infrastructure.
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Mon, 30 May 2022 06:58:08 +0000 (14:58 +0800)]
drm/amdgpu: suppress the compile warning about 64 bit type
Suppress the compile warning below:
drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c:1292
gfx_v11_0_rlc_backdoor_autoload_copy_ucode() warn: should '1 << id' be a 64 bit type?
Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Evan Quan <evan.quan@amd.com> Reviewed-by: Guchun Chen <guchun.chen@amd.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Evan Quan [Mon, 30 May 2022 03:37:07 +0000 (11:37 +0800)]
drm/amd/pm: suppress compile warnings about possible unaligned accesses
Suppress the following compile warnings:
>> drivers/gpu/drm/amd/amdgpu/../pm/swsmu/inc/smu_v11_0_pptable.h:163:17:
warning: field smc_pptable within 'struct smu_11_0_powerplay_table' is
less aligned than 'PPTable_t' and is usually due to 'struct smu_11_0_powerplay_table'
being packed, which can lead to unaligned accesses [-Wunaligned-access]
PPTable_t smc_pptable; //PPTable_t in smu11_driver_if.h
^
1 warning generated.
--
>> drivers/gpu/drm/amd/amdgpu/../pm/swsmu/inc/smu_v11_0_7_pptable.h:193:17:
warning: field smc_pptable within 'struct smu_11_0_7_powerplay_table' is
less aligned than 'PPTable_t' and is usually due to 'struct smu_11_0_7_powerplay_table'
being packed, which can lead to unaligned accesses [-Wunaligned-access]
PPTable_t smc_pptable; //PPTable_t in smu11_driver_if.h
^
1 warning generated.
--
>> drivers/gpu/drm/amd/amdgpu/../pm/swsmu/inc/smu_v13_0_pptable.h:161:12:
warning: field smc_pptable within 'struct smu_13_0_powerplay_table' is less aligned than
'PPTable_t' and is usually due to 'struct smu_13_0_powerplay_table' being packed, which
can lead to unaligned accesses [-Wunaligned-access]
Signed-off-by: Evan Quan <evan.quan@amd.com> Reported-by: kernel test robot <lkp@intel.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Fri, 3 Jun 2022 13:19:34 +0000 (09:19 -0400)]
drm/amdkfd: Fix partial migration bugs
Migration range from system memory to VRAM, if system page can not be
locked or unmapped, we do partial migration and leave some pages in
system memory. Several bugs found to copy pages and update GPU mapping
for this situation:
1. copy to vram should use migrate->npage which is total pages of range
as npages, not migrate->cpages which is number of pages can be migrated.
2. After partial copy, set VRAM res cursor as j + 1, j is number of
system pages copied plus 1 page to skip copy.
3. copy to ram, should collect all continuous VRAM pages and copy
together.
4. Call amdgpu_vm_update_range, should pass in offset as bytes, not
as number of pages.
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> Cc: stable@vger.kernel.org
Lang Yu [Tue, 31 May 2022 01:19:43 +0000 (09:19 +0800)]
drm/amdkfd: add pinned BOs to kfd_bo_list
The kfd_bo_list is used to restore process BOs after
evictions. As page tables could be destroyed during
evictions, we should also update pinned BOs' page tables
during restoring to make sure they are valid.
So for pinned BOs,
1, Validate them and update their page tables.
2, Don't add eviction fence for them.
v2:
- Don't handle pinned ones specially in BO validation.(Felix)
Signed-off-by: Lang Yu <Lang.Yu@amd.com> Acked-by: Christian König <christian.koenig@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Wed, 1 Jun 2022 23:02:45 +0000 (19:02 -0400)]
drm/amdgpu: Update PDEs flush TLB if PTB/PDB moved
Flush TLBs when existing PDEs are updated because a PTB or PDB moved,
but avoids unnecessary TLB flushes when new PDBs or PTBs are added to
the page table, which commonly happens when memory is mapped for the
first time.
Suggested-by: Christian König <christian.koenig@amd.com> Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Philip Yang [Thu, 26 May 2022 20:15:38 +0000 (16:15 -0400)]
drm/amdkfd: Use mmget_not_zero in MMU notifier
MMU notifier callback may pass in mm with mm->mm_users==0 when process
is exiting, use mmget_no_zero to avoid accessing invalid mm in deferred
list work after mm is gone.
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>
Roman Li [Wed, 25 May 2022 21:20:21 +0000 (17:20 -0400)]
drm/amdgpu: fix aper_base for APU
[Why]
Wrong fb offset results in dmub f/w errors and white screen.
[drm:dc_dmub_srv_wait_idle [amdgpu]] *ERROR* Error waiting for DMUB idle: status=3
[How]
Read aper_base from mmhub because GC is off by default
v2: use BAR for passthrough (Alex)
Signed-off-by: Roman Li <Roman.Li@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex Deucher [Thu, 26 May 2022 20:34:55 +0000 (16:34 -0400)]
drm/amdgpu: update VCN codec support for Yellow Carp
Supports AV1. Mesa already has support for this and
doesn't rely on the kernel caps for yellow carp, so
this was already working from an application perspective.
Fixes: 554398174d98 ("amdgpu/nv.c - Added video codec support for Yellow Carp")
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2002 Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org
Aric Cyr [Mon, 16 May 2022 01:52:38 +0000 (21:52 -0400)]
drm/amd/display: 3.2.187
This version brings along the following fixes:
* Changes to DP LT fallback behavior to more closely match the DP standard
* Added new interfaces for lut pipeline
* Restore ref_dtblck value when clk struct is cleared in init_clocks
* Fixes DMUB outbox trace in S4
* Fixes lingering DIO FIFO errors when DIO no longer enabled
* Reads Golden Settings Table from VBIOS
Ilya [Mon, 7 Feb 2022 22:39:10 +0000 (17:39 -0500)]
drm/amd/display: Fix possible infinite loop in DP LT fallback
[Why]
It's possible for some fallback scenarios to result in infinite looping
during link training.
[How]
This change modifies DP LT fallback behavior to more closely match the
DP standard. Keep track of the link rate during the EQ_FAIL fallback,
and use it as the maximum link rate for the CR sequence.
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Ilya <Ilya.Bakoulin@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alvin [Thu, 12 May 2022 20:49:16 +0000 (16:49 -0400)]
drm/amd/display: Don't clear ref_dtbclk value
[Description]
ref_dtbclk value is assigned in clk_mgr_construct,
but the clks struct is cleared in init_clocks.
Make sure to restore the value or we will get
0 value for ref_dtbclk in DCN31.
Reviewed-by: Chris Park <Chris.Park@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alvin Lee <Alvin.Lee2@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Hung, Cruise [Fri, 13 May 2022 01:16:42 +0000 (09:16 +0800)]
drm/amd/display: Fix DMUB outbox trace in S4 (#4465)
[Why]
DMUB Outbox0 read/write pointer not sync after resumed from S4.
And that caused old traces were sent to outbox.
[How]
Disable DMUB Outbox0 interrupt
and clear DMUB Outbox0 read/write pointer when resumes from S4.
And then enable Outbox0 interrupt before starts DMCUB.
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Cruise Hung <Cruise.Hung@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
hengzhou [Sat, 7 May 2022 01:43:08 +0000 (09:43 +0800)]
drm/amd/display: Wait DMCUB to idle state before reset.
[WHY]
Very low rate to cause memory access issue while resetting
DMCUB after the halt command was sent to it.
The process of stopping fw of DMCUB may be timeout, that means
it is not in idle state, such as the window frames may still be
kept in cache, so reset by force will cause MMHUB hang.
[HOW]
After the halt command was sent, keep checking the DMCUB state until
it is idle.
Reviewed-by: Eric Yang <Eric.Yang2@amd.com> Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: hengzhou <Hengyong.Zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nicholas Kazlauskas [Fri, 6 May 2022 16:56:38 +0000 (12:56 -0400)]
drm/amd/display: Pass the new context into disable OTG WA
[Why]
When enabling an HPO stream for the first time after having previously
enabled a DIO stream there may be lingering DIO FIFO errors even though
the DIO is no longer enabled.
These can cause display clock change to hang if we don't apply the
OTG disable workaround since the ramping logic is tied to OTG on.
[How]
The workaround wasn't being applied in the sequence of:
1 DIO stream
0 streams
1 HPO stream
because current_state has no stream or planes in its context - and
it's only swapped after optimize has finished.
We should be using the incoming context instead to determine whether
this logic is needed or not.
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Leung, Martin [Tue, 10 May 2022 16:27:08 +0000 (12:27 -0400)]
drm/amd/display: revert Blank eDP on disable/enable drv
why and how:
Revert this change. It was causing a black screen with certain blocks
Reviewed-by: George Shen <George.Shen@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Leung, Martin <Martin.Leung@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Sherry Wang [Tue, 10 May 2022 10:42:18 +0000 (18:42 +0800)]
drm/amd/display: Read Golden Settings Table from VBIOS
[Why]
Dmub read AUX_DPHY_RX_CONTROL0 from Golden Setting Table,
but driver will set it to default value 0x103d1110, which
causes issue in some case
[How]
Remove the driver code, use the value set by dmub in
dp_aux_init
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com> Acked-by: Jasdeep Dhillon <jdhillon@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Sherry Wang <YAO.WANG1@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Dave Airlie [Wed, 1 Jun 2022 00:56:11 +0000 (10:56 +1000)]
Merge tag 'amd-drm-next-5.19-2022-05-26-2' of https://gitlab.freedesktop.org/agd5f/linux into drm-next
amd-drm-next-5.19-2022-05-26-2:
amdgpu:
- Update fdinfo to the common drm format
UAPI:
- Add VM_NOALLOC GPUVM attribute to prevent buffers for going into the MALL
Add AMDGPU_GEM_CREATE_DISCARDABLE flag to create buffers that can be discarded on eviction
Mesa code which uses these: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16466
Christian König [Tue, 10 May 2022 16:29:35 +0000 (18:29 +0200)]
drm/amdgpu: add drm-client-id to fdinfo v2
This is enough to get gputop working :)
v2: rebase and some addition cleanup
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Shashank Sharma <shashank.sharma@amd.com> (v1) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Wed, 11 May 2022 09:06:26 +0000 (11:06 +0200)]
drm/amdgpu: Convert to common fdinfo format v5
Convert fdinfo format to one documented in drm-usage-stats.rst.
It turned out that the existing implementation was actually completely
nonsense. The calculated percentages indeed represented the usage of the
engine, but with varying time slices.
So 10% usage for application A could mean something completely different
than 10% usage for application B.
Completely nuke that and just use the now standardized nanosecond
interface.
v2: drop the documentation change for now, nuke percentage calculation
v3: only account for each hw_ip, move the time_spend to the ctx mgr.
v4: move general ctx changes into separate patch, rework the fdinfo to
ctx_mgr interface so that all usages are calculated at once, drop
some unecessary and dangerous refcount dance.
v5: add one more comment how we calculate the time spend
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Shashank Sharma <shashank.sharma@amd.com> Cc: Daniel Vetter <daniel@ffwll.ch> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 6 May 2022 11:21:28 +0000 (13:21 +0200)]
drm/amdgpu: bump minor version number
Increase the minor version number to indicate that the new flags are
available.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Marek Olšák <marek.olsak@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 6 May 2022 11:11:41 +0000 (13:11 +0200)]
drm/amdgpu: add AMDGPU_VM_NOALLOC v2
Add the AMDGPU_VM_NOALLOC flag to let userspace control MALL allocation.
v2: also add the flag to the allowed flags.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1) Reviewed-by: Marek Olšák <marek.olsak@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian König [Fri, 6 May 2022 11:08:34 +0000 (13:08 +0200)]
drm/amdgpu: add AMDGPU_GEM_CREATE_DISCARDABLE
Add a AMDGPU_GEM_CREATE_DISCARDABLE flag to note that the content of a BO
doesn't needs to be preserved during eviction.
KFD was already using a similar functionality for SVM BOs so replace the
internal flag with the new UAPI.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Marek Olšák <marek.olsak@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
projects / linux.git / log