- 30 Jun, 2023 11 commits
-
-
Mukul Joshi authored
For GFX 9.4.3, interrupt handling needs to be updated for: - Interrupt cookie will have a NodeId field. Each KFD node needs to check the NodeId before processing the interrupt. - For CPX mode, there are additional checks of client ID needed to process the interrupt. - Add NodeId to the process drain interrupt. Signed-off-by:
Mukul Joshi <mukul.joshi@amd.com> Reviewed-by:
Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by:
Alex Deucher <alexander.deucher@amd.com>
-
Christian König authored
Since adding gang submit we need to take the gang size into account while reserving fences. Signed-off-by:
Christian König <christian.koenig@amd.com> Fixes: 4624459c ("drm/amdgpu: add gang submit frontend v6") Reviewed-by:
-
Hamza Mahfooz authored
Currently, it is possible for us to access memory that we shouldn't. Since, we acquire (possibly dangling) pointers to dirty rectangles before doing a bounds check to make sure we can actually accommodate the number of dirty rectangles userspace has requested to fill. This issue is especially evident if a compositor requests both MPO and damage clips at the same time, in which case I have observed a soft-hang. So, to avoid this issue, perform the bounds check before filling a single dirty rectangle and WARN() about it, if it is ever attempted in fill_dc_dirty_rect(). Cc: stable@vger.kernel.org # 6.1+ Fixes: 30ebe415 ("drm/amd/display: add FB_DAMAGE_CLIPS support") Reviewed-by:
Leo Li <sunpeng.li@amd.com> Signed-off-by:
Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by:
-
Lijo Lazar authored
Publish energy data in 15.625mJ unit for SMU v13.0.6. The same unit is used in Aldebaran also. Signed-off-by:
Lijo Lazar <lijo.lazar@amd.com> Reviewed-by:
Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by:
-
Lijo Lazar authored
Modify it such that it doesn't change the instance mask parameter. Signed-off-by:
Victor Skvortsov <victor.skvortsov@amd.com> Signed-off-by:
-
Mangesh Gadre authored
sdma halt/unhalt is performed by psp when frontdoor loading used,so this can be skipped. v2: Instead of removing halt/unhalt completely, driver will do it only during backdoor load. Signed-off-by:Mangesh Gadre <Mangesh.Gadre@amd.com> Reviewed-by:
Asad Kamal <asad.kamal@amd.com> Signed-off-by:
-
Tao Zhou authored
No RAS irq is allowed. Signed-off-by:
Tao Zhou <tao.zhou1@amd.com> Reviewed-by:
-
Kenneth Feng authored
add abnormal fan detection for smu 13.0.0 Signed-off-by:
Kenneth Feng <kenneth.feng@amd.com> Reviewed-by:
Evan Quan <evan.quan@amd.com> Signed-off-by:
-
Xiaogang Chen authored
Since we allow kfd and graphic operate on same GPU VM to have interoperation between them GPU VM may have been used by graphic vm operations before kfd turns a GPU VM into a compute VM. Remove vm clean checking at amdgpu_vm_make_compute. Signed-off-by:
Xiaogang Chen <xiaogang.chen@amd.com> Reviewed-by:
-
Mario Limonciello authored
A number of users have reported that there are random hangs occurring caused by PSR-SU specifically on panels that contain the parade 0803 TCON. Users have been able to work around the issue by disabling PSR entirely. To avoid these hangs, disable PSR-SU when this TCON is found. Cc: stable@vger.kernel.org Cc: Sean Wang <sean.ns.wang@amd.com> Cc: Marc Rossi <Marc.Rossi@amd.com> Cc: Hamza Mahfooz <Hamza.Mahfooz@amd.com> Suggested-by:
Tsung-hua (Ryan) Lin <Tsung-hua.Lin@amd.com> Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2443Signed-off-by:
Mario Limonciello <mario.limonciello@amd.com> Reviewed-by:
-
Mukul Joshi authored
Enable GWS capable queue creation for forward progress gaurantee on GFX 9.4.3. Signed-off-by:
-
- 23 Jun, 2023 29 commits
-
-
Evan Quan authored
It is proved that insufficient LC_L1_INACTIVITY setting can cause audio noise on some platform. With the LC_L1_INACTIVITY increased to 4ms, the issue can be resolved. Signed-off-by:
-
Aric Cyr authored
This version brings along the following: - DCN314 fixes - DCN32x fixes - New fast update sequence enablement - DC mode clock switching enablement for DCN32x - DP link loss fix - New debugfs entry to set MST link settings Acked-by:
Aric Cyr <aric.cyr@amd.com> Signed-off-by:
-
Alvin Lee authored
- DC mode clock switch interface was previously only executed for DCN303. Enable it for DCN32x so that the interface is called correctly - Assign function pointers for DCN32x that are used in the dc mode interface - Update the dc mode interface to work generically for each ASIC - In update_clocks, make sure to consider softmax if we're in DC mode Reviewed-by:
Jun Lei <jun.lei@amd.com> Acked-by:
Alvin Lee <alvin.lee2@amd.com> Signed-off-by:
-
Ilya Bakoulin authored
[Why] We don't check 128b132b-specific bits in LANE_ALIGN_STATUS_UPDATED DPCD registers when parsing link loss status, which can cause us to miss a link loss notification from some sinks. [How] Add a 128b132b-specific status bit check. Cc: stable@vger.kernel.org # 6.3+ Reviewed-by:
Wenjing Liu <wenjing.liu@amd.com> Acked-by:
Ilya Bakoulin <ilya.bakoulin@amd.com> Signed-off-by:
-
Dmytro Laktyushkin authored
Correct setting is div by 2 for odm. Seamless odm transitions are enabled with enable_dp_dig_pixel_rate_div_policy debug flag. Fixes: a2c7356f ("drm/amd/display: fix pixel rate update sequence") Reviewed-by:
Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by:
Dmytro Laktyushkin <dmytro.laktyushkin@amd.com> Signed-off-by:
-
Fangzhi Zuo authored
When using debugfs to change MST link settings, we need to wait until the next stream update to apply the preferred link setting. So, trigger a hotplug event right after the preferred link setting is applied. Reviewed-by:
Wayne Lin <wayne.lin@amd.com> Acked-by:
Fangzhi Zuo <jerry.zuo@amd.com> Signed-off-by:
-
Leo Chen authored
[Why & How] Having seamless boot on while forcing debug option ODM combine 2 to 1 will cause some corruptions because of some missing programmings. Cc: stable@vger.kernel.org # 6.1+ Reviewed-by:
Leo Chen <sancchen@amd.com> Signed-off-by:
-
Austin Zheng authored
Why: Limiting clocks to DC mode max results in some display modes to no longer be supported How: Disable the path that limits the clock values Fixes: 3b718dca ("drm/amd/display: Filter out AC mode frequencies on DC mode systems") Reviewed-by:
Martin Leung <martin.leung@amd.com> Acked-by:
Austin Zheng <austin.zheng@amd.com> Signed-off-by:
-
Alvin Lee authored
The new fast update sequence is now supported on some ASICs. So, enable it by default for all applicable ASICs. Reviewed-by:
Samson Tam <samson.tam@amd.com> Acked-by:
-
Alvin Lee authored
Clear update flags so the next flip does not have any redundant programming (if a subsequent flip does not have a stream or plane update, the update flags are not cleared). Fixes: 0baae624 ("drm/amd/display: Refactor fast update to use new HWSS build sequence") Reviewed-by:
-
Alvin Lee authored
Condition for programming manually trigger used the wrong pipe (always used top pipe instead of the one we are iterating through). Fixes: 0baae624 ("drm/amd/display: Refactor fast update to use new HWSS build sequence") Reviewed-by:
-
Sridevi Arvindekar authored
[Why] We are currently missing some ABM registers. [How] Add the missing registers to dce_abm.h. Reviewed-by:
Chris Park <chris.park@amd.com> Acked-by:
Sridevi Arvindekar <sridevi.arvindekar@amd.com> Signed-off-by:
-
Austin Zheng authored
Why: Certain display configs resulted in underflow How: Add an entry containing all max DC clock timings Reviewed-by:
-
Sung-huai Wang authored
[Why & How] We have to check if stream is properly initialized before calling find_matching_pll(), otherwise we might end up trying to deferecence a NULL pointer. Cc: stable@vger.kernel.org # 6.1+ Reviewed-by:
Sung-huai Wang <danny.wang@amd.com> Signed-off-by:
-
Daniel Miess authored
This partially reverts commit de231189 ("drm/amd/display: Fix possible underflow for displays with large vblank"). [Why] The increased value of VBlankNomDefaultUS causes underflow at the desktop of an IP KVM setup [How] Change the value from 800 back to 668 Reviewed-by:
Daniel Miess <daniel.miess@amd.com> Signed-off-by:
-
Daniel Miess authored
This reverts commit e383b127. Controling hubp power gating using the DMCUB isn't stable so we are reverting this change to move control back into the driver. Cc: stable@vger.kernel.org # 6.3+ Reviewed-by:
-
Daniel Miess authored
[Why] RCO is causing error messages on some DCN314 systems [How] Force disable RCO for DCN314 Fixes: 17fbdbda ("drm/amd/display: Enable dcn314 DPP RCO") Reviewed-by:
-
Daniel Miess authored
[Why] Power gating is causing error messages on some DCN314 systems [How] Force disable power gating for DCN314 Fixes: 4cc1cebe ("drm/amd/display: Re-enable DPP/HUBP Power Gating") Reviewed-by:
-
Zhigang Luo authored
For SRIOV VF, no TMR needed. Reviewed-by:
Zhigang Luo <Zhigang.Luo@amd.com> Signed-off-by:
-
Nathan Chancellor authored
After commit 8020f0f9 ("drm/amd/amdgpu: enable W=1 for amdgpu"), there is an instance of -Wunused-const-variable when CONFIG_DEBUG_FS is disabled: drivers/gpu/drm/amd/amdgpu/../pm/amdgpu_pm.c:38:34: error: unused variable 'clocks' [-Werror,-Wunused-const-variable] 38 | static const struct cg_flag_name clocks[] = { | ^ 1 error generated. clocks is only used when CONFIG_DEBUG_FS is set, so move the definition into the CONFIG_DEBUG_FS block right above its only usage to clear up the warning. Signed-off-by:
Nathan Chancellor <nathan@kernel.org> Signed-off-by:
-
Evan Quan authored
Also, correct the comment for NAVI10_PCIE__LC_L1_INACTIVITY_TBT_DEFAULT as 0x0000000E stands for 400ms instead of 4ms. Signed-off-by:
-
Samuel Pitoiset authored
Per VM BOs must be marked as moved or otherwise their ranges are not updated on use which might be necessary when the replace operation splits mappings. This fixes random GPU hangs when replacing sparse mappings from the userspace, while OP_MAP/OP_UNMAP works fine because always valid BOs are correctly handled there. Cc: stable@vger.kernel.org Signed-off-by:
Samuel Pitoiset <samuel.pitoiset@gmail.com> Reviewed-by:
-
Alex Sierra authored
This flag determines whether the host possesses coherent access to the memory of the device. Signed-off-by:
Alex Sierra <alex.sierra@amd.com> Reviewed-by:
-
Lijo Lazar authored
Not all devices carry VBIOS version information. Add the device attribute only if supported. Signed-off-by:
Le Ma <le.ma@amd.com> Signed-off-by:
-
Alex Deucher authored
LPDDR5 channels are 32 bit rather than 64, report the width properly in the log. v2: Only LPDDR5 are 32 bits per channel. DDR5 is 64 bits per channel Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2468Acked-by:
-
Nathan Chancellor authored
After commit 8020f0f9 ("drm/amd/amdgpu: enable W=1 for amdgpu"), there is an instance of -Wunused-const-variable when CONFIG_DEBUG_FS is disabled: drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c:110:37: error: unused variable 'amdgpu_rap_debugfs_ops' [-Werror,-Wunused-const-variable] 110 | static const struct file_operations amdgpu_rap_debugfs_ops = { | ^ 1 error generated. There is no reason for the body of this function to be guarded when CONFIG_DEBUG_FS is disabled, as debugfs_create_file() is a stub that just returns an error pointer in that situation. Remove the preprocessor guards so that the variable never appears unused, while not changing anything at run time. Signed-off-by:
-
Mingtong Bao authored
fix the following coccicheck warning: drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c:1657:14-18: Unneeded variable: "size". Signed-off-by:
Mingtong Bao <baomingtong001@208suo.com> Signed-off-by:
-
Lijo Lazar authored
Its value is required for finding the memory id of xcp. Fixes: d26ea1b3 ("drm/amdgpu: Add xcp manager num_xcp_per_mem_partition") Signed-off-by:
-
Jiadong Zhu authored
Only low priority rings are using chunks to save the offset. Bypass the mark offset callings from high priority rings. Signed-off-by:
Jiadong Zhu <Jiadong.Zhu@amd.com> Acked-by:
-
