kernel/drivers/gpu/drm/amd/amdgpu, branch linux-6.12.y

drm/amd: Fix hang on amdgpu unload by using pci_dev_is_disconnected()

2026-03-13T16:20:30Z

[ Upstream commit f7afda7fcd169a9168695247d07ad94cf7b9798f ] The commit 6a23e7b4332c ("drm/amd: Clean up kfd node on surprise disconnect") introduced early KFD cleanup when drm_dev_is_unplugged() returns true. However, this causes hangs during normal module unload (rmmod amdgpu). The issue occurs because drm_dev_unplug() is called in amdgpu_pci_remove() for all removal scenarios, not just surprise disconnects. This was done intentionally in commit 39934d3ed572 ("Revert "drm/amdgpu: TA unload messages are not actually sent to psp when amdgpu is uninstalled"") to fix IGT PCI software unplug test failures. As a result, drm_dev_is_unplugged() returns true even during normal module unload, triggering the early KFD cleanup inappropriately. The correct check should distinguish between: - Actual surprise disconnect (eGPU unplugged): pci_dev_is_disconnected() returns true - Normal module unload (rmmod): pci_dev_is_disconnected() returns false Replace drm_dev_is_unplugged() with pci_dev_is_disconnected() to ensure the early cleanup only happens during true hardware disconnect events. Cc: stable@vger.kernel.org Reported-by: Cal Peake Closes: https://lore.kernel.org/all/b0c22deb-c0fa-3343-33cf-fd9a77d7db99@absolutedigital.net/ Fixes: 6a23e7b4332c ("drm/amd: Clean up kfd node on surprise disconnect") Acked-by: Alex Deucher Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: Fix locking bugs in error paths

2026-03-13T16:20:19Z

[ Upstream commit 480ad5f6ead4a47b969aab6618573cd6822bb6a4 ] Do not unlock psp->ras_context.mutex if it has not been locked. This has been detected by the Clang thread-safety analyzer. Cc: Alex Deucher Cc: Christian König Cc: YiPeng Chai Cc: Hawking Zhang Cc: amd-gfx@lists.freedesktop.org Fixes: b3fb79cda568 ("drm/amdgpu: add mutex to protect ras shared memory") Acked-by: Christian König Signed-off-by: Bart Van Assche Signed-off-by: Alex Deucher (cherry picked from commit 6fa01b4335978051d2cd80841728fd63cc597970) Signed-off-by: Sasha Levin

drm/amdgpu: Replace kzalloc + copy_from_user with memdup_user

2026-03-13T16:20:19Z

[ Upstream commit 99eeb8358e6cdb7050bf2956370c15dcceda8c7e ] Replace kzalloc() followed by copy_from_user() with memdup_user() to improve and simplify ta_if_load_debugfs_write() and ta_if_invoke_debugfs_write(). No functional changes intended. Reviewed-by: Tvrtko Ursulin Signed-off-by: Thorsten Blum Signed-off-by: Alex Deucher Stable-dep-of: 480ad5f6ead4 ("drm/amdgpu: Fix locking bugs in error paths") Signed-off-by: Sasha Levin

drm/amdgpu: Unlock a mutex before destroying it

2026-03-13T16:20:19Z

[ Upstream commit 5e0bcc7b88bcd081aaae6f481b10d9ab294fcb69 ] Mutexes must be unlocked before these are destroyed. This has been detected by the Clang thread-safety analyzer. Cc: Alex Deucher Cc: Christian König Cc: Yang Wang Cc: Hawking Zhang Cc: amd-gfx@lists.freedesktop.org Fixes: f5e4cc8461c4 ("drm/amdgpu: implement RAS ACA driver framework") Reviewed-by: Yang Wang Acked-by: Christian König Signed-off-by: Bart Van Assche Signed-off-by: Alex Deucher (cherry picked from commit 270258ba320beb99648dceffb67e86ac76786e55) Signed-off-by: Sasha Levin

drm/amdgpu: keep vga memory on MacBooks with switchable graphics

2026-03-04T12:21:59Z

[ Upstream commit 096bb75e13cc508d3915b7604e356bcb12b17766 ] On Intel MacBookPros with switchable graphics, when the iGPU is enabled, the address of VRAM gets put at 0 in the dGPU's virtual address space. This is non-standard and seems to cause issues with the cursor if it ends up at 0. We have the framework to reserve memory at 0 in the address space, so enable it here if the vram start address is 0. Reviewed-and-tested-by: Mario Kleiner Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4302 Cc: stable@vger.kernel.org Cc: Mario Kleiner Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: fix sync handling in amdgpu_dma_buf_move_notify

2026-03-04T12:21:56Z

[ Upstream commit b18fc0ab837381c1a6ef28386602cd888f2d9edf ] Invalidating a dmabuf will impact other users of the shared BO. In the scenario where process A moves the BO, it needs to inform process B about the move and process B will need to update its page table. The commit fixes a synchronisation bug caused by the use of the ticket: it made amdgpu_vm_handle_moved behave as if updating the page table immediately was correct but in this case it's not. An example is the following scenario, with 2 GPUs and glxgears running on GPU0 and Xorg running on GPU1, on a system where P2P PCI isn't supported: glxgears: export linear buffer from GPU0 and import using GPU1 submit frame rendering to GPU0 submit tiled->linear blit Xorg: copy of linear buffer The sequence of jobs would be: drm_sched_job_run # GPU0, frame rendering drm_sched_job_queue # GPU0, blit drm_sched_job_done # GPU0, frame rendering drm_sched_job_run # GPU0, blit move linear buffer for GPU1 access # amdgpu_dma_buf_move_notify -> update pt # GPU0 It this point the blit job on GPU0 is still running and would likely produce a page fault. Cc: stable@vger.kernel.org Fixes: a448cb003edc ("drm/amdgpu: implement amdgpu_gem_prime_move_notify v2") Signed-off-by: Pierre-Eric Pelloux-Prayer Reviewed-by: Christian König Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: Adjust usleep_range in fence wait

2026-03-04T12:21:28Z

[ Upstream commit 3ee1c72606bd2842f0f377fd4b118362af0323ae ] Tune the sleep interval in the PSP fence wait loop from 10-100us to 60-100us.This adjustment results in an overall wait window of 1.2s (60us * 20000 iterations) to 2 seconds (100us * 20000 iterations), which guarantees that we can retrieve the correct fence value Signed-off-by: Ce Sun Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: Skip loading SDMA_RS64 in VF

2026-03-04T12:21:02Z

[ Upstream commit 39c21b81112321cbe1267b02c77ecd2161ce19aa ] VFs use the PF SDMA ucode and are unable to load SDMA_RS64. Signed-off-by: YuBiao Wang Signed-off-by: Victor Skvortsov Reviewed-by: Gavin Wan Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: avoid a warning in timedout job handler

2026-03-04T12:21:02Z

[ Upstream commit c8cf9ddc549fb93cb5a35f3fe23487b1e6707e74 ] Only set an error on the fence if the fence is not signalled. We can end up with a warning if the per queue reset path signals the fence and sets an error as part of the reset, but fails to recover. Reviewed-by: Timur Kristóf Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin

drm/amdgpu: add support for HDP IP version 6.1.1

2026-03-04T12:21:01Z

[ Upstream commit e2fd14f579b841f54a9b7162fef15234d8c0627a ] This initializes HDP IP version 6.1.1. Reviewed-by: Mario Limonciello Signed-off-by: Tim Huang Signed-off-by: Alex Deucher Signed-off-by: Sasha Levin