- 13 Aug, 2024 40 commits
-
-
David (Ming Qiang) Wu authored
Because gpu_addr is updated in the calling routine (amdgpu_cs_patch_ibs()),it is removed in the callback. Use .patch_cs_in_place instead of .parse_cs for amdgpu_vce_ring_parse_cs_vm() as there is no need for keeping a temporary IB, therefore ib->sa_bo is NULL and amdgpu_ib_free() is removed. Reviewed-by:
Alex Deucher <alexander.deucher@amd.com> Signed-off-by:
David (Ming Qiang) Wu <David.Wu3@amd.com> Signed-off-by:
-
David (Ming Qiang) Wu authored
Add JPEG IB command parser to ensure registers in the command are within the JPEG IP block. Reviewed-by:
-
Jack Xiao authored
Use mes pipe to unmap kcq and kgq. Signed-off-by:
Jack Xiao <Jack.Xiao@amd.com> Acked-by:
Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by:
-
Jack Xiao authored
Free memory for two pipes and unmap pipe0 via pipe1. Signed-off-by:
-
Jack Xiao authored
Configure two pipes with different hardware resources. Signed-off-by:
-
Jack Xiao authored
Adjust mes12 sw/hw initiailization for both pipe0 and pipe1 enablement. The two pipes are almost identical pipe. Pipe0 behaves like schq and pipe1 like kiq, pipe0 was mapped by pipe1. Signed-off-by:
-
Jack Xiao authored
Add mes pipe switch to let caller choose pipe to submit packet. Signed-off-by:
-
Victor Skvortsov authored
Register access from userspace should be blocked until reset is complete. Signed-off-by:
Victor Skvortsov <victor.skvortsov@amd.com> Reviewed-by:
-
Jonathan Kim authored
If queue reset fails, tell the CP to reset the pipe. Since queues multiplex context per pipe and we've issued a device wide preemption prior to the hang, we can assume the hung pipe only has one queue to reset on pipe reset. Signed-off-by:
Jonathan Kim <jonathan.kim@amd.com> Reviewed-by:
-
Lijo Lazar authored
On EFI BIOSes, PCI ROM may be exported through EFI_PCI_IO_PROTOCOL and expansion ROM BARs may not be enabled. Choose to read from EFI exported ROM data before reading PCI Expansion ROM BAR. Signed-off-by:
Lijo Lazar <lijo.lazar@amd.com> Reviewed-by:
-
Jack Xiao authored
Enable unified mes firmware to load on pipe0 and pipe1. Signed-off-by:
-
Victor Skvortsov authored
VFs do not perform HW fini/suspend in FLR, so the dpm_enabled is incorrectly kept enabled. Add interface to disable it in virt_pre_reset call. v2: Made implementation generic for all asics v3: Re-order conditionals so PP_MP1_STATE_FLR is only evaluated on VF Signed-off-by:
-
Victor Skvortsov authored
KIQ timeouts no longer seen. This reverts commit 3a19a8af. Signed-off-by:
Zhigang Luo <zhigang.luo@amd.com> Signed-off-by:
-
Yinjie Yao authored
kmd_fw_shared changed in VCN5 Signed-off-by:
Yinjie Yao <yinjie.yao@amd.com> Reviewed-by:
Ruijing Dong <ruijing.dong@amd.com> Acked-by:
-
Lijo Lazar authored
If there are multiple nodes per kfd device, add nodeid to location_id to differentiate. Signed-off-by:
-
Kenneth Feng authored
add HDP_SD support on gc 12.0.0/1 Signed-off-by:
Kenneth Feng <kenneth.feng@amd.com> Reviewed-by:
Yang Wang <kevinyang.wang@amd.com> Signed-off-by:
-
Victor Zhao authored
on MI300/MI308 UBB products, when doing mode1 reset, since 1 gpu need to wait all 8 gpus finish mode1 reset and then do re-init. As observed, sometimes the gpu which triggered the reset need to wait 15s for all gpus to finish. If poll msg timeout, guest driver will send the reset message again, and may mess up the following reinit sequence on other gpus. So extend the time to cover the maximum time needed to recover. Signed-off-by:
Victor Zhao <Victor.Zhao@amd.com> Acked-by:
-
Martin Leung authored
This version brings along following fixes: - Fix some cursor issue - Fix print format specifiers in DC_LOG_IPS - Fix minor coding errors in dml21 phase 5 - Fix MST BW calculation Regression - Improve FAM control for DCN401 - Add null pointer checks for some code - Refactor 3DLUT for non-DMA - Optimize vstartup position for AS-SDP - Update to using new dccg callbacks - Enable otg synchronization logic for DCN321 - Disable DCN401 UCLK P-State support on full updates Acked-by:
Wayne Lin <wayne.lin@amd.com> Signed-off-by:
Martin Leung <Martin.Leung@amd.com> Signed-off-by:
Tom Chung <chiahsuan.chung@amd.com> Tested-by:
Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by:
-
Rodrigo Siqueira authored
[why & how] Remove unnecessary call to REG_SEQ_SUBMIT and REG_SEQ_WAIT_DONE, since those macros are not necessary anymore at the dpp1 set degamma. Those are part of an old implementation. Acked-by:
Rodrigo Siqueira <Rodrigo.Siqueira@amd.com> Signed-off-by:
-
Rodrigo Siqueira authored
[why & how] When the commit 9d84c7ef ("drm/amd/display: Correct cursor position on horizontal mirror") was introduced, it used the wrong calculation for the position copy for X. This commit uses the correct calculation for that based on the original patch. Fixes: 9d84c7ef ("drm/amd/display: Correct cursor position on horizontal mirror") Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Acked-by:
-
Melissa Wen authored
[why & how] Cursor gets clipped off in the middle of the screen with hw rotation 180. Fix a miscalculation of cursor offset when it's placed near the edges in the pipe split case. Cursor bugs with hw rotation were reported on AMD issue tracker: https://gitlab.freedesktop.org/drm/amd/-/issues/2247 The issues on rotation 270 was fixed by: https://lore.kernel.org/amd-gfx/20221118125935.4013669-22-Brian.Chang@amd.com/ that partially addressed the rotation 180 too. So, this patch is the final bits for rotation 180. Reported-by:
Xaver Hugl <xaver.hugl@gmail.com> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2247Reviewed-by:
Harry Wentland <harry.wentland@amd.com> Fixes: 9d84c7ef ("drm/amd/display: Correct cursor position on horizontal mirror") Signed-off-by:
Melissa Wen <mwen@igalia.com> Signed-off-by:
Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by:
-
Rodrigo Siqueira authored
[why & how] When the commit 5324e2b2 ("drm/amd/display: Add driver support for future FAMS versions") was introduced, it missed some of the FAM2 code. This commit introduces the code that control the FAM enable and disable. Fixes: 5324e2b2 ("drm/amd/display: Add driver support for future FAMS versions") Acked-by:
-
Rodrigo Siqueira authored
[why & how] Remove force_backlight_start_level since it is never used. Acked-by:
-
Fangzhi Zuo authored
[Why & How] Revert commit 8b2cb32c ("drm/amd/display: FEC overhead should be checked once for mst slot nums") Because causes bw calculation regression Cc: mario.limonciello@amd.com Cc: alexander.deucher@amd.com Reported-by: jirislaby@kernel.org Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3495 Closes: https://bugzilla.suse.com/show_bug.cgi?id=1228093Reviewed-by:
Fangzhi Zuo <Jerry.Zuo@amd.com> Signed-off-by:
-
Loan Chen authored
[Why] Tiled display cannot synchronize properly after S3. The fix for commit 5f0c7491 ("drm/amd/display: Fix for otg synchronization logic") is not enable in DCN321, which causes the otg is excluded from synchronization. [How] Enable otg synchronization logic in dcn321. Fixes: 5f0c7491 ("drm/amd/display: Fix for otg synchronization logic") Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Reviewed-by:
Alvin Lee <alvin.lee2@amd.com> Signed-off-by:
Loan Chen <lo-an.chen@amd.com> Signed-off-by:
-
Charlene Liu authored
[why & how] this is to remove redundant msg to pmfw at boot/resume since bios already power up dcn. Reviewed-by:
Chris Park <chris.park@amd.com> Signed-off-by:
Charlene Liu <Charlene.Liu@amd.com> Signed-off-by:
-
Dillon Varone authored
[WHY&HOW] Set max VTotal cap for dcn401 because VTotal register is only 16 bits wide on dcn401. Reviewed-by:
Dillon Varone <dillon.varone@amd.com> Signed-off-by:
-
Dillon Varone authored
[WHY] In certain scenarios DC can internally trigger back to back full updates which will miss some required programming that is normally deferred until post update via optimize_bandwidth. [HOW] In back to back update scenarios, wait for pending updates to complete and perform any strictly required outstanding programming. Reviewed-by:
-
Dillon Varone authored
[WHY&HOW] It is not guaranteed even for HW exclusive P-State methods (like VActive) that P-state will be supported properly until optimize bandwidth is called, so unconditionally disable it on full updates. Reviewed-by:
-
Dillon Varone authored
[WHY] Stream ID's associated with phantom pipes can change often as they are reconstructed on full updates, however they can remain identical depending on the required update. [HOW] In the case phantom streams and pipe topologies remain the same between updates, mark the transition as seamless. Reviewed-by:
-
Srinivasan Shanmugam authored
This commit adds a null check for the 'afb' variable in the amdgpu_dm_plane_handle_cursor_update function. Previously, 'afb' was assumed to be null, but was used later in the code without a null check. This could potentially lead to a null pointer dereference. Changes since v1: - Moved the null check for 'afb' to the line where 'afb' is used. (Alex) Fixes the below: drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_plane.c:1298 amdgpu_dm_plane_handle_cursor_update() error: we previously assumed 'afb' could be null (see line 1252) Cc: Tom Chung <chiahsuan.chung@amd.com> Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com> Cc: Roman Li <roman.li@amd.com> Cc: Alex Hung <alex.hung@amd.com> Cc: Aurabindo Pillai <aurabindo.pillai@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Co-developed-by:
Alex Hung <alex.hung@amd.com> Signed-off-by:
Srinivasan Shanmugam <srinivasan.shanmugam@amd.com> Reviewed-by:
-
Robin Chen authored
[Why] In current design, the vstartup position is adjusted to vblank start position when AS-SDP is enabled. However when the vblank length is too big, it may over vstartup boundary. [How] To adjust vstartup position to 1 line before vsync position. Reviewed-by:
Anthony Koo <anthony.koo@amd.com> Signed-off-by:
Robin Chen <robin.chen@amd.com> Signed-off-by:
-
Roman Li authored
[Why] %d specifier is used for printing unsigned values. It can result in negative values in logs for unsigned variables. [How] Replace %d with %u for unsigned. Reviewed-by:
Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by:
Roman Li <Roman.Li@amd.com> Signed-off-by:
-
Philip Yang authored
Add helper function kfd_queue_unreference_buffers to reduce queue buffer refcount, separate it from release queue buffers. Because it is circular locking to hold dqm_lock to take vm lock, kfd_ioctl_destroy_queue should take vm lock, unreference queue buffers first, but not release queue buffers, to handle error in case failed to hold vm lock. Then hold dqm_lock to remove queue from queue list and then release queue buffers. Restore process worker restore queue hold dqm_lock, will always find the queue with valid queue buffers. v2 (Felix): - renamed kfd_queue_unreference_buffer(s) to kfd_queue_unref_bo_va(s) - added two FIXME comments for follow up Signed-off-by:
Philip Yang <Philip.Yang@amd.com> Signed-off-by:
Felix Kuehling <felix.kuehling@amd.com> Reviewed-by:
-
Jonathan Kim authored
When users dynamically set the partition mode through sysfs writes, this can lead to a double lock situation where the KFD is trying to take the partition lock when updating the recommended SDMA engines. Have the KFD reference its saved socket device number count instead. Also ensure we have enough SDMA xGMI engines to report the recommended engines in the first place. Fixes: e06b71b2 ("drm/amdkfd: allow users to target recommended SDMA engines") Signed-off-by:
-
Sunil Khatri authored
Change condition, if (ptr == NULL) to if (!ptr) for a better format and fix the warning. Reviewed-by:
Sunil Khatri <sunil.khatri@amd.com> Signed-off-by:
-
Sunil Khatri authored
Change condition, if (ptr == NULL) to if (!ptr) for a better format and fix the warning. Reviewed-by:
-
Sunil Khatri authored
Change condition, if (ptr == NULL) to if (!ptr) for a better format and fix the warning. Reviewed-by:
-
Sunil Khatri authored
Change if (ptr == NULL) to if (!ptr) for a better format and fix the warning. Reviewed-by:
-
Wayne Lin authored
[why & how] Make sure plane_state is not null before calling a function that dereferences it. Besides, remove redundant codes. Reviewed-by:
-
