linux/drivers/gpu/drm/amd/amdkfd, branch master

drm/amdkfd: Fix queue preemption/eviction failures by aligning control stack size to GPU page size

2026-03-30T19:16:39Z

The control stack size is calculated based on the number of CUs and waves, and is then aligned to PAGE_SIZE. When the resulting control stack size is aligned to 64 KB, GPU hangs and queue preemption failures are observed while running RCCL unit tests on systems with more than two GPUs. amdgpu 0048:0f:00.0: amdgpu: Queue preemption failed for queue with doorbell_id: 80030008 amdgpu 0048:0f:00.0: amdgpu: Failed to evict process queues amdgpu 0048:0f:00.0: amdgpu: GPU reset begin!. Source: 4 amdgpu 0048:0f:00.0: amdgpu: Queue preemption failed for queue with doorbell_id: 80030008 amdgpu 0048:0f:00.0: amdgpu: Failed to evict process queues amdgpu 0048:0f:00.0: amdgpu: Failed to restore process queues This issue is observed on both 4 KB and 64 KB system page-size configurations. This patch fixes the issue by aligning the control stack size to AMDGPU_GPU_PAGE_SIZE instead of PAGE_SIZE, so the control stack size will not be 64 KB on systems with a 64 KB page size and queue preemption works correctly. Additionally, In the current code, wg_data_size is aligned to PAGE_SIZE, which can waste memory if the system page size is large. In this patch, wg_data_size is aligned to AMDGPU_GPU_PAGE_SIZE. The cwsr_size, calculated from wg_data_size and the control stack size, is aligned to PAGE_SIZE. Reviewed-by: Felix Kuehling Signed-off-by: Donet Tom Signed-off-by: Alex Deucher

drm/amdgpu: use TTM_NUM_MOVE_FENCES when reserving fences

2026-03-30T19:16:33Z

Use TTM_NUM_MOVE_FENCES as an upperbound of how many fences ttm might need to deal with moves/evictions. Signed-off-by: Pierre-Eric Pelloux-Prayer Acked-by: Felix Kuehling Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amdgpu: allocate move entities dynamically

2026-03-30T19:16:15Z

No functional change for now, as we always allocate a single entity. Signed-off-by: Pierre-Eric Pelloux-Prayer Acked-by: Felix Kuehling Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amdkfd: fix kernel crash on releasing NULL sysfs entry

2026-03-30T19:14:51Z

there is an abnormal case that When a process re-opens kfd with different mm_struct(execve() called by user), the allocated p->kobj will be freed, but missed setting it to NULL, that will cause sysfs/kernel crash with NULL pointers in p->kobj on kfd_process_remove_sysfs() when releasing process, and the similar error on kfd_procfs_del_queue() as well. Signed-off-by: Eric Huang Reviewed-by: Kent Russell Signed-off-by: Alex Deucher

drm/amdkfd: Switch to dev_* printk stuff in kfd_int_process_v12_1.c

2026-03-30T19:13:14Z

dev_* printk stuff is multi-GPU friendly. Use dev_warn_ratelimited() for print_sq_intr_info_error() which is consistent with previous IPs. Use dev_dbg_ratelimited() for irrelevant node interrupt print to avoid too much noise. Signed-off-by: Lang Yu Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher

drm/amdgpu: Change AMDGPU_VA_RESERVED_TRAP_SIZE to 64KB

2026-03-30T18:38:11Z

Currently, AMDGPU_VA_RESERVED_TRAP_SIZE is hardcoded to 8KB, while KFD_CWSR_TBA_TMA_SIZE is defined as 2 * PAGE_SIZE. On systems with 4K pages, both values match (8KB), so allocation and reserved space are consistent. However, on 64K page-size systems, KFD_CWSR_TBA_TMA_SIZE becomes 128KB, while the reserved trap area remains 8KB. This mismatch causes the kernel to crash when running rocminfo or rccl unit tests. Kernel attempted to read user page (2) - exploit attempt? (uid: 1001) BUG: Kernel NULL pointer dereference on read at 0x00000002 Faulting instruction address: 0xc0000000002c8a64 Oops: Kernel access of bad area, sig: 11 [#1] LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=2048 NUMA pSeries CPU: 34 UID: 1001 PID: 9379 Comm: rocminfo Tainted: G E 6.19.0-rc4-amdgpu-00320-gf23176405700 #56 VOLUNTARY Tainted: [E]=UNSIGNED_MODULE Hardware name: IBM,9105-42A POWER10 (architected) 0x800200 0xf000006 of:IBM,FW1060.30 (ML1060_896) hv:phyp pSeries NIP: c0000000002c8a64 LR: c00000000125dbc8 CTR: c00000000125e730 REGS: c0000001e0957580 TRAP: 0300 Tainted: G E MSR: 8000000000009033 CR: 24008268 XER: 00000036 CFAR: c00000000125dbc4 DAR: 0000000000000002 DSISR: 40000000 IRQMASK: 1 GPR00: c00000000125d908 c0000001e0957820 c0000000016e8100 c00000013d814540 GPR04: 0000000000000002 c00000013d814550 0000000000000045 0000000000000000 GPR08: c00000013444d000 c00000013d814538 c00000013d814538 0000000084002268 GPR12: c00000000125e730 c000007e2ffd5f00 ffffffffffffffff 0000000000020000 GPR16: 0000000000000000 0000000000000002 c00000015f653000 0000000000000000 GPR20: c000000138662400 c00000013d814540 0000000000000000 c00000013d814500 GPR24: 0000000000000000 0000000000000002 c0000001e0957888 c0000001e0957878 GPR28: c00000013d814548 0000000000000000 c00000013d814540 c0000001e0957888 NIP [c0000000002c8a64] __mutex_add_waiter+0x24/0xc0 LR [c00000000125dbc8] __mutex_lock.constprop.0+0x318/0xd00 Call Trace: 0xc0000001e0957890 (unreliable) __mutex_lock.constprop.0+0x58/0xd00 amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu+0x6fc/0xb60 [amdgpu] kfd_process_alloc_gpuvm+0x54/0x1f0 [amdgpu] kfd_process_device_init_cwsr_dgpu+0xa4/0x1a0 [amdgpu] kfd_process_device_init_vm+0xd8/0x2e0 [amdgpu] kfd_ioctl_acquire_vm+0xd0/0x130 [amdgpu] kfd_ioctl+0x514/0x670 [amdgpu] sys_ioctl+0x134/0x180 system_call_exception+0x114/0x300 system_call_vectored_common+0x15c/0x2ec This patch changes AMDGPU_VA_RESERVED_TRAP_SIZE to 64 KB and KFD_CWSR_TBA_TMA_SIZE to the AMD GPU page size. This means we reserve 64 KB for the trap in the address space, but only allocate 8 KB within it. With this approach, the allocation size never exceeds the reserved area. Fixes: 34a1de0f7935 ("drm/amdkfd: Relocate TBA/TMA to opposite side of VM hole") Reviewed-by: Christian König Suggested-by: Felix Kuehling Suggested-by: Christian König Signed-off-by: Donet Tom Signed-off-by: Alex Deucher

drm/amd: Fix MQD and control stack alignment for non-4K

2026-03-30T18:36:04Z

For gfxV9, due to a hardware bug ("based on the comments in the code here [1]"), the control stack of a user-mode compute queue must be allocated immediately after the page boundary of its regular MQD buffer. To handle this, we allocate an enlarged MQD buffer where the first page is used as the MQD and the remaining pages store the control stack. Although these regions share the same BO, they require different memory types: the MQD must be UC (uncached), while the control stack must be NC (non-coherent), matching the behavior when the control stack is allocated in user space. This logic works correctly on systems where the CPU page size matches the GPU page size (4K). However, the current implementation aligns both the MQD and the control stack to the CPU PAGE_SIZE. On systems with a larger CPU page size, the entire first CPU page is marked UC—even though that page may contain multiple GPU pages. The GPU treats the second 4K GPU page inside that CPU page as part of the control stack, but it is incorrectly mapped as UC. This patch fixes the issue by aligning both the MQD and control stack sizes to the GPU page size (4K). The first 4K page is correctly marked as UC for the MQD, and the remaining GPU pages are marked NC for the control stack. This ensures proper memory type assignment on systems with larger CPU page sizes. [1]: https://elixir.bootlin.com/linux/v6.18/source/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c#L118 Acked-by: Felix Kuehling Signed-off-by: Donet Tom Signed-off-by: Alex Deucher

drm/amdkfd: Align expected_queue_size to PAGE_SIZE

2026-03-30T18:35:57Z

The AQL queue size can be 4K, but the minimum buffer object (BO) allocation size is PAGE_SIZE. On systems with a page size larger than 4K, the expected queue size does not match the allocated BO size, causing queue creation to fail. Align the expected queue size to PAGE_SIZE so that it matches the allocated BO size and allows queue creation to succeed. Reviewed-by: Felix Kuehling Signed-off-by: Donet Tom Signed-off-by: Alex Deucher

drm/amdkfd: Fix NULL pointer check order in kfd_ioctl_create_process

2026-03-24T17:29:40Z

In kfd_ioctl_create_process(), the pointer 'p' is used before checking if it is NULL. The code accesses p->context_id before validating 'p'. This can lead to a possible NULL pointer dereference. Move the NULL check before using 'p' so that the pointer is validated before access. Fixes the below: drivers/gpu/drm/amd/amdgpu/../amdkfd/kfd_chardev.c:3177 kfd_ioctl_create_process() warn: variable dereferenced before check 'p' (see line 3174) Fixes: cc6b66d661fd ("amdkfd: introduce new ioctl AMDKFD_IOC_CREATE_PROCESS") Cc: Zhu Lingshan Cc: Felix Kuehling Cc: Christian König Cc: Alex Deucher Cc: Dan Carpenter Signed-off-by: Srinivasan Shanmugam Reviewed-by: Christian König Signed-off-by: Alex Deucher

drm/amd/amdgpu: Fix build errors due to declarations after labels

2026-03-17T14:42:47Z

In C90 (which the kernel uses with -std=gnu89), declarations must appear at the beginning of a block and cannot follow a label. The switch cases in amdgpu_discovery.c and gmc_v12_1.c contained variable declarations immediately after case labels, causing the compiler to error: drivers/gpu/drm/amd/amdgpu/gmc_v12_1.c:533:3: error: a label can only be part of a statement and a declaration is not a statement Reviewed-by: Alex Deucher Signed-off-by: Jesse Zhang Signed-off-by: Alex Deucher