RE: [PATCH v2] drm/ttm: Should to return the evict error

[AMD Official Use Only - AMD Internal Distribution Only] From: amd-gfx On Behalf Of Deng, Emily Sent: Monday, May 26, 2025 9:51 AM To: Chen, Xiaogang ; amd-gfx@lists.freedesktop.org Subject: RE: [PATCH v2] drm/ttm: Should to return the evict error [AMD Official Use Only - AMD Internal Distrib

RE: [PATCH v2] drm/amdkfd: Move the process suspend and resume out of full access

[AMD Official Use Only - AMD Internal Distribution Only] >-Original Message- >From: Emily Deng >Sent: Tuesday, May 27, 2025 6:50 PM >To: amd-gfx@lists.freedesktop.org >Cc: Deng, Emily >Subject: [PATCH v2] drm/amdkfd: Move the process suspend and resume out of full >access > >For the susp

[PATCH 09/19] drm/amdgpu/ring: add helper for padding the ring

Add a helper to pad the ring to a multiple of the fetch size. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 29 +--- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 1 + 2 files changed, 22 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/a

[PATCH 15/19] drm/amdgpu/gfx9: re-emit unprocessed state on kcq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 13 +++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index d

[PATCH 14/19] drm/amdgpu/gfx12: re-emit unprocessed state on kgq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c | 12 ++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index

[PATCH 12/19] drm/amdgpu/gfx10: re-emit unprocessed state on kgq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 19 --- 1 file changed, 12 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.

[PATCH 01/19] drm/amdgpu/gfx10: enable legacy enforce isolation

Enable legacy enforce isolation (just serialize kernel GC submissions). This way we can reset a ring and only affect the the process currently using that ring. This mirrors what windows does. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 9 + 1 file changed, 9

[PATCH 10/19] drm/amdgpu: pad ring in amdgpu_ib_schedule

We'll want to include the padding in the wptr count for resets. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 802743efa3b39..3e05e

[PATCH 05/19] drm/amdgpu/gfx8: drop reset_kgq

It doesn't work reliably and we have soft recover and full adapter reset so drop this. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 71 --- 1 file changed, 71 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/am

[PATCH 17/19] drm/amdgpu/gfx10: re-emit unprocessed state on kcq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 29 +- 1 file changed, 10 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gf

[PATCH 19/19] drm/amdgpu/gfx12: re-emit unprocessed state on kcq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c | 27 ++ 1 file changed, 10 insertions(+), 17 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gf

[PATCH 08/19] drm/amdgpu: add AMDGPU_QUEUE_RESET_TIMEOUT

Add a new define for queue reset timeout. This will be used for the IB tests used when validating ring resets. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgp

[PATCH V3 00/19] Reset improvements for GC10+

This set improves per queue reset support for GC10+. When we reset the queue, the queue is lost so we need to re-emit the unprocessed state from subsequent submissions. To that end, in order to make sure we actually restore unprocessed state, we need to enable legacy enforce isolation so that we ca

[PATCH 02/19] drm/amdgpu/gfx11: enable legacy enforce isolation

Enable legacy enforce isolation (just serialize kernel GC submissions). This way we can reset a ring and only affect the the process currently using that ring. This mirrors what windows does. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 9 + 1 file changed, 9

[PATCH 03/19] drm/amdgpu/gfx12: enable legacy enforce isolation

Enable legacy enforce isolation (just serialize kernel GC submissions). This way we can reset a ring and only affect the the process currently using that ring. This mirrors what windows does. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c | 9 + 1 file changed, 9

[PATCH 06/19] drm/amdgpu/gfx9: drop reset_kgq

It doesn't work reliably and we have soft recover and full adapter reset so drop this. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 46 --- 1 file changed, 46 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/am

[PATCH 13/19] drm/amdgpu/gfx11: re-emit unprocessed state on kgq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 11 +-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 3

[PATCH 18/19] drm/amdgpu/gfx11: re-emit unprocessed state on kcq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 28 ++ 1 file changed, 11 insertions(+), 17 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gf

[PATCH 11/19] drm/amdgpu: track ring state associated with a job

We need to know the wptr and sequence number associated with a job so that we can re-emit the unprocessed state after a ring reset. Pre-allocate storage space for the ring buffer contents and add a helper to save off the unprocessed state so that it can be re-emitted after the queue is reset. Sig

[PATCH 16/19] drm/amdgpu/gfx9.4.3: re-emit unprocessed state on kcq reset

Re-emit the unprocessed state after resetting the queue. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 15 --- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c

[PATCH 07/19] drm/amdgpu: rework queue reset scheduler interaction

From: Christian König Stopping the scheduler for queue reset is generally a good idea because it prevents any worker from touching the ring buffer. But using amdgpu_fence_driver_force_completion() before restarting it was a really bad idea because it marked fences as failed while the work was po

[PATCH 04/19] drm/amdgpu/gfx7: drop reset_kgq

It doesn't work reliably and we have soft recover and full adapter reset so drop this. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 71 --- 1 file changed, 71 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/am

Re: [PATCH 02/10] drm/amdgpu: rework queue reset scheduler interaction

On Tue, May 27, 2025 at 12:47 PM Rodrigo Siqueira wrote: > > On 05/22, Alex Deucher wrote: > > From: Christian König > > > > Stopping the scheduler for queue reset is generally a good idea because > > it prevents any worker from touching the ring buffer. > > > > But using amdgpu_fence_driver_forc

[PATCH 24/24] drm/amd/display: Promote DAL to 3.2.336

From: Taimur Hassan This version brings along following fixes: - Fix brightness relevant settings - Fix calling blanking stream twice - Extend dc mode validation types to support more scenarios - Update DMCUB loading sequence for DCN3.5 Acked-by: ChiaHsuan Chung Signed-off-by: Taimur Hassan Si

[PATCH 21/24] drm/amd/display: Use DC log instead of using DM error msg

From: Cruise Hung [Why & How] It sent an error msg when it failed to read the DP tunneling DPCD field. This should just be a warning msg. Use a DC log instead of a DM error msg. Reviewed-by: Wenjing Liu Signed-off-by: Cruise Hung Signed-off-by: Wayne Lin --- .../gpu/drm/amd/display/dc/link/p

[PATCH 23/24] drm/amd/display: Fix default DC and AC levels

From: Mario Limonciello [Why] DC and AC levels are advertised in a percentage, not a luminance. [How] Scale DC and AC levels to supported values. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4221 Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello Signed-off-by: Wayne Lin ---

[PATCH 22/24] drm/amd/display: Add debugging message for brightness caps

From: Mario Limonciello [Why] Default BIOS brightness caps are buried in ACPI. [How] Add extra dynamic debug that can show default brightness caps. Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 7

[PATCH 20/24] drm/amd/display: Avoid calling blank_stream() twice

From: Zhongwei Zhang [Why] We've made fix for garbage in dcn31_reset_back_end_for_pipe(), adding blank_stream() before disable_crtc(). And set_dpms_off() will call blank_stream() again. [How] Add flag to avoid calling blank_stream() twice. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Zhongw

[PATCH 19/24] drm/amd/display: replace fast_validate with enum dc_validate_mode

From: Yan Li [Why] The boolean fast_validate is used as an input parameter in multiple functions. To support more scenarios, we are replacing it with enum dc_validate_mode. [How] The enum dc_validate_mode introduces three possible values: 1) DC_VALIDATE_MODE_AND_PROGRAMMING: Apply the mode to

[PATCH 18/24] drm/amd/display: Update DMCUB loading sequence for DCN3.5

From: Nicholas Kazlauskas [Why] New sequence from HW for reset and firmware reloading has been provided that aims to stabilize the reload sequence in the case the firmware is hung or has outstanding requests. [How] Update the sequence to remove the DMUIF reset and the redundant writes in the rel

[PATCH 17/24] drm/amd/display: Promote DAL to 3.2.335

From: Taimur Hassan This version brings along following fixes: - Fixes for DML21 - Support OLED SDR with AMD ABC - Indirect buffer transport for FAMS2 commands - Correct stream attributes setup timing - Correct non-OLED pre_T11_delay - Optime boot-up consuming time - Add support for 2nd sharpenin

[PATCH 16/24] drm/amd/display: [FW Promotion] Release 0.1.12.0

From: Taimur Hassan Add dmub command to support LSDMA Acked-by: ChiaHsuan Chung Signed-off-by: Taimur Hassan Signed-off-by: Wayne Lin --- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 156 ++ 1 file changed, 156 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dmub/i

[PATCH 15/24] drm/amd/display: Move vmalloc include to header file

From: Ray Wu [Why & How] Move vmalloc.h include code to header file. Reviewed-by: ChiaHsuan Chung Signed-off-by: Ray Wu Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c | 2 -- drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c| 2 -- drivers/gpu

[PATCH 14/24] drm/amd/display: Add support for 2nd sharpening range

From: Samson Tam [Why & How] Add support for 2nd sharpening range for cases where we want override existing DCN sharpening range Reviewed-by: Ilya Bakoulin Signed-off-by: Samson Tam Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dc/dc.h | 6 ++ .../gpu/drm/amd/displa

[PATCH 13/24] drm/amd/display: Do not bypass chroma scaling in 1:1 case

From: Navid Assadian [Why] When doing 2:1 downscaling on a YUV sub-sampled format, the chroma scaling ratio is 1:1. Since chroma has cositing, it is needed to do scaling on the chroma plane(s) and not to bypass chroma scaling. [How] Do not set the chroma taps to one when the chroma ratio is iden

[PATCH 12/24] drm/amd/display: Add DML path for FAMS methods

From: Oleh Kuzhylnyi [Why] DML needs a path for FAMS methods. [How] Apply instance of fams2_stream_sub_params_v2 structure with a FAMS placeholder for DML. Reviewed-by: Alvin Lee Signed-off-by: Oleh Kuzhylnyi Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c |

[PATCH 11/24] drm/amd/display: Add disconnect case on dongle check

From: Jingwen Zhu [why] In the case of an external monitor disconnection, the kernel mode will attempt to post new timing validation with two path counts (eDP + external monitor removed to virtual). [how] Skip validating color depth and pixel encoding in the scenario involving a DP to HDMI activ

[PATCH 10/24] drm/amd/display: Avoid trying AUX transactions on disconnected ports

[Why & How] Observe that we try to access DPCD 0x600h of disconnected DP ports. In order not to wasting time on retrying these ports, call dpcd_write_rx_power_ctrl() after checking its connection status. Reviewed-by: Aurabindo Pillai Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dc/l

[PATCH 09/24] drm/amd/display: Correct non-OLED pre_T11_delay.

From: Zhongwei Zhang [Why] Only OLED panels require non-zero pre_T11_delay defaultly. Others should be controlled by power sequence. [How] For non OLED, pre_T11_delay delay in code should be zero. Also post_T7_delay. Cc: Mario Limonciello Cc: Alex Deucher Cc: sta...@vger.kernel.org Reviewed-b

[PATCH 08/24] drm/amd/display: Call setup_stream_attribute after stream enc clk is ungated

From: Michael Strauss [WHY] If symclk RCO is enabled, stream encoder may not be receiving an ungated clock by the time we attempt to set stream attributes when setting dpms on. Since the clock is gated, register writes to the stream encoder fail. [HOW] Move set_stream_attribute call into enable_

[PATCH 07/24] drm/amd/display: Drop unnecessary `amdgpu` prefix

From: Mario Limonciello [Why] The `drm_*()` print macros will handle including the driver in the print already. The extra print of the word `amdgpu` is unnecessary. [How] Modify all prints to drop `amdgpu: `. Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello Signed-off-by: Wayne Lin -

[PATCH 06/24] drm/amd/display: Indirect buffer transport for FAMS2 commands

From: Oleh Kuzhylnyi [Why] The quantity and duration of FAMS2 commands are set to increase in future products. This necessitates the implementation of a new mechanism for chaining commands together, allowing all commands to be processed within a single transaction. [How] The indirect buffer acts

[PATCH 05/24] drm/amd/display: move RMCM programming

From: Yihan Zhu [WHY & HOW] Move only RMCM programming outside of dcn401. Extended HW definition in dc for memory layout to extend support. Reviewed-by: Jun Lei Signed-off-by: Yihan Zhu Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dc/dc.h | 31 ++ drivers/gpu/drm/amd/d

[PATCH 04/24] drm/amd/display: Support OLED SDR with AMD ABC

From: Camille Cho [Why] Nits programming for SDR panel is only supported by VESA ABC. [How] 1. Loose nits programming for OLED SDR panel with AMD ABC. 2. We support two ABC methods. Disable one before we program the other in case panel freaks out. 3. Update HDR judgement in setBR with a solid

[PATCH 03/24] drm/amd/display: DML21 Fixes

From: Austin Zheng - Store state related info inside mode_lib. - Fix bad DCFCLK deep sleep - Update FAMS structure in DMUB header Reviewed-by: Alvin Lee Signed-off-by: Austin Zheng Signed-off-by: Wayne Lin --- .../amd/display/dc/dml2/dml21/inc/dml_top.h | 1 - .../display/dc/dml2/dml21/in

[PATCH 00/24] DC Patches June 2nd, 2025

This DC patchset brings improvements in multiple areas. In summary, we highlight: - Fixes for DML21 - Support OLED SDR with AMD ABC - Indirect buffer transport for FAMS2 commands - Correct stream attributes setup timing - Correct non-OLED pre_T11_delay - Optime boot-up consuming time - Add suppor

[PATCH 01/24] drm/amd/display: [FW Promotion] Release 0.1.11.0

From: Taimur Hassan Refactoring some DMUB related structs and enum. Acked-by: Wayne Lin Signed-off-by: Taimur Hassan Signed-off-by: Tom Chung --- .../gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 34 +-- 1 file changed, 32 insertions(+), 2 deletions(-) diff --git a/drivers/gpu

[PATCH 02/24] drm/amd/display: Re-order FAMS2 sub commands

From: Alvin Lee [Why & How] New enums need to be added to the end to avoid back compat issues. Reviewed-by: Dillon Varone Signed-off-by: Alvin Lee Signed-off-by: Wayne Lin --- drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --

Re: [PATCH 2/5] drm/amdgpu: rework gfx9 queue reset

On Tue, May 27, 2025 at 12:39 PM Rodrigo Siqueira wrote: > > On 05/20, Alex Deucher wrote: > > On Mon, May 19, 2025 at 7:59 PM Rodrigo Siqueira > > wrote: > > > > > > On 05/02, Christian König wrote: > > > > Testing this feature turned out that it was a bit unstable. The > > > > CP_VMID_RESET re

RE: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the dependent read fences"

[Public] > From: amd-gfx On Behalf Of Christian > König > Sent: Tuesday, May 27, 2025 8:24 PM > To: Khatri, Sunil ; amd-gfx@lists.freedesktop.org; > Deucher, > Alexander > Subject: Re: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the > dependent read fences" > > On 5/27/25 13:27, Kh

[PATCH v2] drm/amdkfd: Map wptr BO to GART unconditionally

For simulation C models that don't run CP FW where adev->mes.sched_version is not populated correctly. This causes NULL dereference in amdgpu_amdkfd_free_gtt_mem(dev->adev, (void **)&pqn->q->wptr_bo_gart) and warning on unpinned BO in amdgpu_bo_gpu_offset(q->properties.wptr_bo). Compared with addi

Re: [PATCH 01/10] Revert "drm/amd/amdgpu: add pipe1 hardware support"

On Tue, May 27, 2025 at 12:45 PM Rodrigo Siqueira wrote: > > On 05/22, Alex Deucher wrote: > > This reverts commit b7a1a0ef12b81957584fef7b61e2d5ec049c7209. > > > > A user reported stuttering under heavy gfx load with this commit. > > I suspect it's due to the fact that the gfx contexts are shared

Re: [PATCH] Documentation: add initial documenation for user queues

Hi Alex, Follow some comments and questions. On 05/02, Alex Deucher wrote: > Add an initial documentation page for user mode queues. > > Signed-off-by: Alex Deucher > --- > Documentation/gpu/amdgpu/index.rst | 1 + > Documentation/gpu/amdgpu/userq.rst | 196 + > 2

Re: [PATCH] Revert "drm/amd/display: no 3D and blnd LUT as DPP color caps for DCN401"

On 2025-05-27 13:06, Alex Hung wrote: This reverts commit 46e68dd5066c9831e9695c1756db017bb9c3762c since it breaks color enhancement in another OS, indicating these two fields are still used by color management in dcn401. Fixes: 46e68dd5066c ("drm/amd/display: no 3D and blnd LUT as DPP color

[PATCH] Revert "drm/amd/display: no 3D and blnd LUT as DPP color caps for DCN401"

This reverts commit 46e68dd5066c9831e9695c1756db017bb9c3762c since it breaks color enhancement in another OS, indicating these two fields are still used by color management in dcn401. Fixes: 46e68dd5066c ("drm/amd/display: no 3D and blnd LUT as DPP color caps for DCN401") Signed-off-by: Alex Hung

Re: [PATCH 02/10] drm/amdgpu: rework queue reset scheduler interaction

On 05/22, Alex Deucher wrote: > From: Christian König > > Stopping the scheduler for queue reset is generally a good idea because > it prevents any worker from touching the ring buffer. > > But using amdgpu_fence_driver_force_completion() before restarting it was > a really bad idea because it m

Re: [PATCH 01/10] Revert "drm/amd/amdgpu: add pipe1 hardware support"

On 05/22, Alex Deucher wrote: > This reverts commit b7a1a0ef12b81957584fef7b61e2d5ec049c7209. > > A user reported stuttering under heavy gfx load with this commit. > I suspect it's due to the fact that the gfx contexts are shared > between the pipes so if there is alot of load on one pipe, we coul

Re: [PATCH 2/5] drm/amdgpu: rework gfx9 queue reset

On 05/20, Alex Deucher wrote: > On Mon, May 19, 2025 at 7:59 PM Rodrigo Siqueira wrote: > > > > On 05/02, Christian König wrote: > > > Testing this feature turned out that it was a bit unstable. The > > > CP_VMID_RESET register takes the VMID which all submissions from should > > > be canceled. >

Re: [PATCH 1/2] drm/amdgpu/mes: add missing locking in helper functions

[Public] This series is Reviewed-by: Michael Chen From: amd-gfx on behalf of Alex Deucher Sent: Tuesday, May 27, 2025 10:46 AM To: Deucher, Alexander Cc: amd-gfx@lists.freedesktop.org Subject: Re: [PATCH 1/2] drm/amdgpu/mes: add missing locking in helper func

[PATCH] drm/amd: Export DMCUB version to sysfs

For supported ASICs DMCU version is exported, but ASICs that support DMCUB there is no information exported to sysfs. Add an attribute for DMCUB. Signed-off-by: Mario Limonciello --- drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c | 8 +--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --

[PATCH] drm/amdkfd: Fix kfd process ref leaking when userptr unmapping

kfd_lookup_process_by_pid increases process ref, the refcount is leaking. Fixes: 7a566d7f56f4 ("amd/amdkfd: Trigger segfault for early userptr unmmapping") Signed-off-by: Philip Yang --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 9 +++-- 1 file changed, 7 insertions(+), 2 deletions

Re: [PATCH 1/2] drm/amdgpu/mes: add missing locking in helper functions

Ping on this series? Alex On Mon, May 19, 2025 at 4:04 PM Alex Deucher wrote: > > We need to take the MES lock. > > Signed-off-by: Alex Deucher > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 16 > 1 file changed, 16 insertions(+) > > diff --git a/drivers/gpu/drm/amd/amdgpu/

Re: [PATCH] Documentation: add initial documenation for user queues

Ping? Alex On Fri, May 2, 2025 at 5:00 PM Alex Deucher wrote: > > Add an initial documentation page for user mode queues. > > Signed-off-by: Alex Deucher > --- > Documentation/gpu/amdgpu/index.rst | 1 + > Documentation/gpu/amdgpu/userq.rst | 196 + > 2 files chan

Re: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the dependent read fences"

On 5/27/25 13:27, Khatri, Sunil wrote: > [AMD Official Use Only - AMD Internal Distribution Only] > > -Original Message- > From: Koenig, Christian > Sent: Tuesday, May 27, 2025 2:32 PM > To: Khatri, Sunil ; amd-gfx@lists.freedesktop.org; > Deucher, Alexander > Subject: Re: [PATCH] Rever

[PATCH v2] drm/amdkfd: Move the process suspend and resume out of full access

For the suspend and resume process, exclusive access is not required. Therefore, it can be moved out of the full access section to reduce the duration of exclusive access. Signed-off-by: Emily Deng --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 16 + drivers/gpu/drm/amd/amdgpu/amdgpu_a

RE: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the dependent read fences"

[AMD Official Use Only - AMD Internal Distribution Only] -Original Message- From: Koenig, Christian Sent: Tuesday, May 27, 2025 2:32 PM To: Khatri, Sunil ; amd-gfx@lists.freedesktop.org; Deucher, Alexander Subject: Re: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the depend

Re: [PATCH] Revert "drm/amdgpu: promote the implicit sync to the dependent read fences"

On 5/27/25 10:58, Sunil Khatri wrote: > This reverts commit 714bbbf20a7266e48632fab466563e695af9acb5. > bisected to this change which is causing the flikering issue in the UI > for various apps like glxgears and unigen heaven. Is that flickering also there when using kernel queues? If not then wi

[PATCH] Revert "drm/amdgpu: promote the implicit sync to the dependent read fences"

This reverts commit 714bbbf20a7266e48632fab466563e695af9acb5. bisected to this change which is causing the flikering issue in the UI for various apps like glxgears and unigen heaven. Also when we set user_queue=1, which enable both user and kernel submissions the userqueue tests in IGT also fail.

Re: [PATCH 1/4] drm/sched: optimize drm_sched_job_add_dependency

On Mon, 2025-05-26 at 11:25 +0200, Christian König wrote: > On 5/23/25 16:16, Danilo Krummrich wrote: > > On Fri, May 23, 2025 at 04:11:39PM +0200, Danilo Krummrich wrote: > > > On Fri, May 23, 2025 at 02:56:40PM +0200, Christian König wrote: > > > > It turned out that we can actually massively opt

Re: [PATCH 1/4] drm/sched: optimize drm_sched_job_add_dependency

On Mon, 2025-05-26 at 13:16 +0200, Christian König wrote: > On 5/26/25 11:34, Philipp Stanner wrote: > > On Mon, 2025-05-26 at 11:25 +0200, Christian König wrote: > > > On 5/23/25 16:16, Danilo Krummrich wrote: > > > > On Fri, May 23, 2025 at 04:11:39PM +0200, Danilo Krummrich > > > > wrote: > > >

Re: [PATCH 1/4] drm/sched: optimize drm_sched_job_add_dependency a bit

+Cc Matthew, again :) On Thu, 2025-05-22 at 18:19 +0200, Christian König wrote: > On 5/22/25 16:27, Tvrtko Ursulin wrote: > > > > On 22/05/2025 14:41, Christian König wrote: > > > Since we already iterated over the xarray we know at which index > > > the new > > > entry should be stored. So inste