Tom St Denis [Fri, 20 Aug 2021 18:03:12 +0000 (14:03 -0400)]
drm/amd/amdgpu: New debugfs interface for MMIO registers (v5)
This new debugfs interface uses an IOCTL interface in order to pass
along state information like SRBM and GRBM bank switching. This
new interface also allows a full 32-bit MMIO address range which
the previous didn't. With this new design we have room to grow
the flexibility of the file as need be.
(v2): Move read/write to .read/.write, fix style, add comment
for IOCTL data structure
(v3): C style comments
(v4): use u32 in struct and remove offset variable
(v5): Drop flag clearing in op function, use 0xFFFFFFFF for broadcast
instead of 0x3FF, use mutex for op/ioctl.
Nirmoy Das [Wed, 25 Aug 2021 15:11:31 +0000 (17:11 +0200)]
drm/amdgpu: detach ring priority from gfx priority
Currently AMDGPU_RING_PRIO_MAX is redefinition of a
max gfx hwip priority, this won't work well when we will
have a hwip with different set of priorities than gfx.
Also, HW ring priorities are different from ring priorities.
Create a global enum for ring priority levels which each
HWIP can use to define its own priority levels.
Nirmoy Das [Tue, 24 Aug 2021 18:39:18 +0000 (20:39 +0200)]
drm/amdgpu: rework context priority handling
To get a hardware queue priority for a context, we are currently
mapping AMDGPU_CTX_PRIORITY_* to DRM_SCHED_PRIORITY_* and then
to hardware queue priority, which is not the right way to do that
as DRM_SCHED_PRIORITY_* is software scheduler's priority and it is
independent from a hardware queue priority.
Use userspace provided context priority, AMDGPU_CTX_PRIORITY_* to
map a context to proper hardware queue priority.
Aric Cyr [Sun, 15 Aug 2021 00:28:55 +0000 (20:28 -0400)]
drm/amd/display: 3.2.150
This version brings along following fixes:
- FW promotion 0.0.80
- Add missing ABM register offsets
- Fix in swizzle mode mapping
- Emulated sink support for freesync
- Improvoments in max target bpp
Aurabindo Pillai [Wed, 11 Aug 2021 18:38:44 +0000 (14:38 -0400)]
drm/amd/display: Add emulated sink support for updating FS
[Why]
When forced modes are used during certain IGT tests,
without a real connector, dc_sink would be null when
standard modes are added by the driver. Calling the
function to update freesync capabilities at this
point will result in an error being printed
[How]
Use emulated sink when available. If both the normal
and emulated sink are not available, set all freesync
parameters to 0.
Josip Pavic [Mon, 9 Aug 2021 19:13:49 +0000 (15:13 -0400)]
drm/amd/display: add missing ABM register offsets
[Why]
Some ABM registers don't exist on DCN 3.01, so are
missing from its register offset list. However,
this list was copied to later versions of DCN that
do have these registers. As a result, they're
inaccessible from the driver on those DCN versions
even though they exist.
[How]
Add the missing ABM register offsets to DCN 3.02+
Guchun Chen [Fri, 27 Aug 2021 10:31:41 +0000 (18:31 +0800)]
drm/amdgpu: stop scheduler when calling hw_fini (v2)
This gurantees no more work on the ring can be submitted
to hardware in suspend/resume case, otherwise a potential
race will occur and the ring will get no chance to stay
empty before suspend.
v2: Call drm_sched_resubmit_job before drm_sched_start to
restart jobs from the pending list.
Michael Strauss [Mon, 16 Aug 2021 20:15:36 +0000 (16:15 -0400)]
drm/amd/display: Initialize lt_settings on instantiation
[WHY]
lt_settings' pointers remain uninitialized but nonzero if display fails
to light up with no DPCD/EDID info populated, leading to a hang on access
Koba Ko [Mon, 30 Aug 2021 02:02:00 +0000 (10:02 +0800)]
drm/amdgpu: Disable PCIE_DPM on Intel RKL Platform
Due to high latency in PCIE clock switching on RKL platforms,
switching the PCIE clock dynamically at runtime can lead to HDMI/DP
audio problems. On newer asics this is handled in the SMU firmware.
For SMU7-based asics, disable PCIE clock switching to avoid the issue.
AMD provide a parameter to disable PICE_DPM.
modprobe amdgpu ppfeaturemask=0xfff7bffb
It's better to contorl PCIE_DPM in amd gpu driver,
switch PCI_DPM by determining intel RKL platform for SMU7-based asics.
Fixes: 1a31474cdb48 ("drm/amd/pm: workaround for audio noise issue")
Ref: https://lists.freedesktop.org/archives/amd-gfx/2021-August/067413.html Signed-off-by: Koba Ko <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
Kees Cook [Fri, 27 Aug 2021 03:16:47 +0000 (20:16 -0700)]
drm/amd/pm: And destination bounds checking to struct copy
In preparation for FORTIFY_SOURCE performing compile-time and run-time
field bounds checking for memcpy(), memmove(), and memset(), avoid
intentionally writing across neighboring fields.
The "Board Parameters" members of the structs:
struct atom_smc_dpm_info_v4_5
struct atom_smc_dpm_info_v4_6
struct atom_smc_dpm_info_v4_7
struct atom_smc_dpm_info_v4_10
are written to the corresponding members of the corresponding PPTable_t
variables, but they lack destination size bounds checking, which means
the compiler cannot verify at compile time that this is an intended and
safe memcpy().
Since the header files are effectively immutable[1] and a struct_group()
cannot be used, nor a common struct referenced by both sides of the
memcpy() arguments, add a new helper, amdgpu_memcpy_trailing(), to
perform the bounds checking at compile time. Replace the open-coded
memcpy()s with amdgpu_memcpy_trailing() which includes enough context
for the bounds checking.
Evan Quan [Tue, 24 Aug 2021 03:32:44 +0000 (11:32 +0800)]
drm/amdgpu: reenable BACO support for 699F:C7 polaris12 SKU
This reverts the commit below:
"drm/amdgpu: disable BACO support for 699F:C7 polaris12 SKU temporarily".
As the S3 hang issue has been fixed by another commit:
"drm/amdgpu: add missing cleanups for Polaris12 UVD/VCE on suspend".
Luben Tuikov [Wed, 25 Aug 2021 17:50:20 +0000 (13:50 -0400)]
drm/amdgpu: Process any VBIOS RAS EEPROM address
We can now process any RAS EEPROM address from
VBIOS. Generalize so as to compute the top three
bits of the 19-bit EEPROM address, from any byte
returned as the "i2c address" from VBIOS.
Luben Tuikov [Tue, 24 Aug 2021 15:01:32 +0000 (11:01 -0400)]
drm/amdgpu: Fixes to returning VBIOS RAS EEPROM address
1) Generalize the function--if the user didn't set
i2c_address, still return true/false to
indicate whether VBIOS contains the RAS EEPROM
address. This function shouldn't evaluate
whether the user set the i2c_address pointer or
not.
2) Don't touch the caller's i2c_address, unless
you have to--this function shouldn't have side
effects.
3) Correctly set the function comment as a
kernel-doc comment.
Jing Yangyang [Tue, 24 Aug 2021 05:52:32 +0000 (22:52 -0700)]
drm:dcn31: fix boolreturn.cocci warnings
./drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.c:112:9-10:WARNING:
return of 0/1 in function 'dcn31_is_panel_backlight_on'
with return type bool
./drivers/gpu/drm/amd/display/dc/dcn31/dcn31_panel_cntl.c:122:9-10:WARNING:
return of 0/1 in function 'dcn31_is_panel_powered_on'
with return type bool
Return statements in functions returning bool should use true/false
instead of 1/0.
Dave Airlie [Sun, 29 Aug 2021 23:04:47 +0000 (09:04 +1000)]
Merge tag 'drm-intel-next-fixes-2021-08-26' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
drm/i915 fixes for v5.15-rc1:
- Disable underrun recovery with eDP MSO panels on ADL-P
- Use designated initializers for init/exit table
- Fix some error pointer usages
drm/amd/display: Move AllowDRAMSelfRefreshOrDRAMClockChangeInVblank to bounding box
[Why]
This is a global parameter, not a per pipe parameter and it's useful
for experimenting with the prefetch schedule to be adjustable from
the SOC bb.
[How]
Add a parameter to the SOC bb, default is the existing policy for
all DCN. Fill it in when filling SOC bb parameters.
Revert the policy to use MinDCFClk at the same time since that's not
going to give us P-State in most cases on the spreadsheet.
Sean Keely [Fri, 20 Aug 2021 02:26:14 +0000 (21:26 -0500)]
drm/amdkfd: Account for SH/SE count when setting up cu masks.
On systems with multiple SH per SE compute_static_thread_mgmt_se#
is split into independent masks, one for each SH, in the upper and
lower 16 bits. We need to detect this and apply cu masking to each
SH. The cu mask bits are assigned first to each SE, then to
alternate SHs, then finally to higher CU id. This ensures that
the maximum number of SPIs are engaged as early as possible while
balancing CU assignment to each SH.
v2: Use max SH/SE rather than max SH in cu_per_sh.
v3: Fix comment blocks, ensure se_mask is initially zero filled,
and correctly assign se.sh.cu positions to unset bits in cu_mask.
Dave Airlie [Thu, 26 Aug 2021 03:05:10 +0000 (13:05 +1000)]
Merge tag 'drm/tegra/for-5.15-rc1' of ssh://git.freedesktop.org/git/tegra/linux into drm-next
drm/tegra: Changes for v5.15-rc1
The bulk of these changes is a more modern ABI that can be efficiently
used on newer SoCs as well as older ones. The userspace parts for this
are available here:
things are still slow in etnaviv land. Just one hardware support
addition for the GPU found on the NXP Layerscape LS1028A SoC from
Michael and the GEM mmap cleanup from Thomas.
amdgpu_bo_get_preferred_pin_domain is used for page tables
creation, which is not involved with page pinning. And it is used in
more cases than display scanout, modify its documentation as well.
Evan Quan [Tue, 17 Aug 2021 06:06:21 +0000 (14:06 +0800)]
drm/amdgpu: add missing cleanups for Polaris12 UVD/VCE on suspend
Perform proper cleanups on UVD/VCE suspend: powergate enablement,
clockgating enablement and dpm disablement. This can fix some hangs
observed on suspending when UVD/VCE still using(e.g. issue
"pm-suspend" when video is still playing).
Philip Yang [Sun, 15 Aug 2021 18:42:33 +0000 (14:42 -0400)]
drm/amdkfd: check access permisson to restore retry fault
Check range access permission to restore GPU retry fault, if GPU retry
fault on address which belongs to VMA, and VMA has no read or write
permission requested by GPU, failed to restore the address. The vm fault
event will pass back to user space.
Shashank Sharma [Fri, 20 Aug 2021 04:50:02 +0000 (10:20 +0530)]
drm/amdgpu/OLAND: clip the ref divider max value
This patch limits the ref_div_max value to 100, during the
calculation of PLL feedback reference divider. With current
value (128), the produced fb_ref_div value generates unstable
output at particular frequencies. Radeon driver limits this
value at 100.
On Oland, when we try to setup mode 2048x1280@60 (a bit weird,
I know), it demands a clock of 221270 Khz. It's been observed
that the PLL calculations using values 128 and 100 are vastly
different, and look like this:
With ref_div_max value clipped at 100, AMDGPU driver can also
drive videmode 2048x1280@60 (221Mhz) and produce proper output
without any blanking and distortion on the screen.
V1:
Got acks from: Acked-by: Alex Deucher <[email protected]> Acked-by: Christian König <[email protected]>
V2:
- Restricting the changes only for OLAND, just to avoid any regression
for other cards.
- Changed unsigned -> unsigned int to make checkpatch quiet.
V3: Apply the change on SI family (not only oland) (Christian)
drm/i915/selftest: Fix use of err in igt_reset_{fail, nop}_engine()
Clang warns:
In file included from drivers/gpu/drm/i915/gt/intel_reset.c:1514:
drivers/gpu/drm/i915/gt/selftest_hangcheck.c:465:62: warning: variable
'err' is uninitialized when used here [-Wuninitialized]
pr_err("[%s] Create context failed: %d!\n", engine->name, err);
^~~
...
drivers/gpu/drm/i915/gt/selftest_hangcheck.c:580:62: warning: variable
'err' is uninitialized when used here [-Wuninitialized]
pr_err("[%s] Create context failed: %d!\n", engine->name, err);
^~~
...
2 warnings generated.
This appears to be a copy and paste issue. Use ce directly using the %pe
specifier to pretty print the error code so that err is not used
uninitialized in these functions.
Matt Roper [Mon, 16 Aug 2021 20:41:12 +0000 (13:41 -0700)]
drm/i915/adl_p: Also disable underrun recovery with MSO
One of the cases that the bspec lists for when underrun recovery must be
disabled is "COG;" that note actually refers to eDP multi-segmented
operation (MSO). Let's ensure the this additional restriction is
honored by the driver.
Kees Cook [Tue, 17 Aug 2021 23:33:57 +0000 (16:33 -0700)]
drm/i915: Use designated initializers for init/exit table
The kernel builds with -Werror=designated-init, and __designated_init
is used by CONFIG_GCC_PLUGIN_RANDSTRUCT for automatically selected (all
function pointer) structures. Include the field names in the init/exit
table. Avoids warnings like:
drivers/gpu/drm/i915/i915_module.c:59:4: error: positional initialization of field in 'struct' declared with 'designated_init' attribute [-Werror=designated-init]
drm/exynos: Always initialize mapping in exynos_drm_register_dma()
In certain randconfigs, clang warns:
drivers/gpu/drm/exynos/exynos_drm_dma.c:121:19: warning: variable
'mapping' is uninitialized when used here [-Wuninitialized]
priv->mapping = mapping;
^~~~~~~
drivers/gpu/drm/exynos/exynos_drm_dma.c:111:16: note: initialize the
variable 'mapping' to silence this warning
void *mapping;
^
= NULL
1 warning generated.
This occurs when CONFIG_EXYNOS_IOMMU is enabled and both
CONFIG_ARM_DMA_USE_IOMMU and CONFIG_IOMMU_DMA are disabled, which makes
the code look like
void *mapping;
if (0)
mapping = arm_iommu_create_mapping()
else if (0)
mapping = iommu_get_domain_for_dev()
...
priv->mapping = mapping;
Add an else branch that initializes mapping to the -ENODEV error pointer
so that there is no more warning and the driver does not change during
runtime.
Michel Dänzer [Tue, 17 Aug 2021 08:23:25 +0000 (10:23 +0200)]
drm/amdgpu: Cancel delayed work when GFXOFF is disabled
schedule_delayed_work does not push back the work if it was already
scheduled before, so amdgpu_device_delay_enable_gfx_off ran ~100 ms
after the first time GFXOFF was disabled and re-enabled, even if GFXOFF
was disabled and re-enabled again during those 100 ms.
This resulted in frame drops / stutter with the upcoming mutter 41
release on Navi 14, due to constantly enabling GFXOFF in the HW and
disabling it again (for getting the GPU clock counter).
To fix this, call cancel_delayed_work_sync when the disable count
transitions from 0 to 1, and only schedule the delayed work on the
reverse transition, not if the disable count was already 0. This makes
sure the delayed work doesn't run at unexpected times, and allows it to
be lock-free.
v2:
* Use cancel_delayed_work_sync & mutex_trylock instead of
mod_delayed_work.
v3:
* Make amdgpu_device_delay_enable_gfx_off lock-free (Christian König)
v4:
* Fix race condition between amdgpu_gfx_off_ctrl incrementing
adev->gfx.gfx_off_req_count and amdgpu_device_delay_enable_gfx_off
checking for it to be 0 (Evan Quan)
Evan Quan [Fri, 20 Aug 2021 08:28:59 +0000 (16:28 +0800)]
drm/amd/pm: a quick fix for "divided by zero" error
Considering Arcturus is a dedicated ASIC for computing, it
will be more proper to drop the support for fan speed reading
and setting. That's on the TODO list.
YuBiao Wang [Tue, 17 Aug 2021 09:36:33 +0000 (17:36 +0800)]
drm/amd/amdgpu:flush ttm delayed work before cancel_sync
[Why]
In some cases when we unload driver, warning call trace
will show up in vram_mgr_fini which claims that LRU is not empty, caused
by the ttm bo inside delay deleted queue.
[How]
We should flush delayed work to make sure the delay deleting is done.
Zhan Liu [Fri, 13 Aug 2021 15:31:04 +0000 (08:31 -0700)]
drm/amd/display: Use DCN30 watermark calc for DCN301
[why]
dcn301_calculate_wm_and_dl() causes flickering when external monitor is
connected.
This issue has been fixed before by commit 0e4c0ae59d7e
("drm/amdgpu/display: drop dcn301_calculate_wm_and_dl for now"), however
part of the fix was gone after commit 2cbcb78c9ee5 ("Merge tag 'amd-drm-next-5.13-2021-03-23' of https://gitlab.freedesktop.org/agd5f/linux into drm-next").
[how]
Use dcn30_calculate_wm_and_dlg() instead as in the original fix.
Jonathan Kim [Tue, 3 Aug 2021 23:01:55 +0000 (19:01 -0400)]
drm/amdgpu: get extended xgmi topology data
The TA has a limit to the amount of data that can be retrieved from
GET_TOPOLOGY. For setups that exceed this limit, the xGMI topology
needs to be re-initialized and data needs to be re-fetched from the
extended link records by setting a flag in the shared command buffer.
The number of hops and the number of links must be accumulated by the
driver. Other data points are all fetched from the first request.
Because the TA has already exceeded its link record limit, it
cannot hold bidirectional information. Otherwise the driver would
have to do more than two fetches so the driver has to reflect the
topology information in the opposite direction.
Dave Airlie [Tue, 17 Aug 2021 00:45:26 +0000 (10:45 +1000)]
Merge tag 'drm-msm-next-2021-08-12' of https://gitlab.freedesktop.org/drm/msm into drm-next
This is the main pull for v5.15, after the early pull request with
drm/scheduler conversion:
* New a6xx GPU support: a680 and 7c3
* dsi: 7nm phi, sc7280 support, test pattern generator support
* mdp4 fixes for older hw like the nexus7
* displayport fixes
Dave Airlie [Tue, 17 Aug 2021 00:29:31 +0000 (10:29 +1000)]
Merge tag 'mediatek-drm-next-5.15' of https://git.kernel.org/pub/scm/linux/kernel/git/chunkuang.hu/linux into drm-next
Mediatek DRM Next for Linux 5.15
1. MT8133 AAL support, adjust rdma fifo threshold formula.
2. Implement mmap as GEM object function.
3. Add support for MT8167.
4. Test component initialization earlier in the function mtk_drm_crtc_create.
5. CMDQ refinement.
projects / linux.git / log