From: Donet Tom Date: Thu, 25 Jun 2026 07:52:06 +0000 (+0530) Subject: drm/amdgpu: Fix AMDGPU_GTT_MAX_TRANSFER_SIZE for non-4K systems X-Git-Tag: v7.2-rc2~10^2~2^2~5 X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=8b7033c0c5dcc3b3bd8403453d2793ec4514ae62;p=thirdparty%2Fkernel%2Flinux.git drm/amdgpu: Fix AMDGPU_GTT_MAX_TRANSFER_SIZE for non-4K systems Running RCCL unit tests on a system with a 64K PAGE_SIZE triggers the following warning and causes the test to terminate on latest upstream kernel: WARNING: drivers/gpu/drm/amd/amdgpu/amdgpu_object.c:1335 at amdgpu_bo_release_notify+0x1bc/0x280 [amdgpu], CPU#18: rccl-UnitTests/33151 Call trace: amdgpu_bo_release_notify ttm_bo_release amdgpu_gem_object_free drm_gem_object_free amdgpu_bo_unref amdgpu_bo_create amdgpu_bo_create_user amdgpu_gem_object_create amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu kfd_ioctl_alloc_memory_of_gpu kfd_ioctl sys_ioctl The warning is triggered because amdgpu_ttm_next_clear_entity() returns NULL when a clear buffer operation is requested. This happens because the GART window allocation for the default_entity, clear_entity and move_entity fails during initialization. Commit [1] introduced separate GART windows for the default_entity, clear_entity and move_entity of each SDMA instance. Their sizes are derived from AMDGPU_GTT_MAX_TRANSFER_SIZE, which is currently defined as 1024 pages. This implicitly assumes a 4K PAGE_SIZE, where 1024 pages correspond to a 4MB transfer. On a 64K PAGE_SIZE system, however, the same value expands to 64MB. The default_entity and clear_entity each allocate one AMDGPU_GTT_MAX_TRANSFER_SIZE GART window, while the move_entity allocates two such windows. This results in 16MB of GART space per SDMA instance on a 4K PAGE_SIZE system, but 256MB per SDMA instance on a 64K PAGE_SIZE system. On an MI210 system with five SDMA instances and a 512MB GART aperture, the total GART space required becomes 1.25GB, exceeding the available GART aperture. Consequently, GART window allocation fails, amdgpu_ttm_next_clear_entity() returns NULL, and the above warning is triggered. Redefine AMDGPU_GTT_MAX_TRANSFER_SIZE in bytes instead of page units. Where a page count is required, convert it using PAGE_SHIFT. This preserves the existing 4MB transfer size across all PAGE_SIZE configurations while keeping GART window allocations within the available GART aperture. [1] https://lore.kernel.org/all/20260408100327.1372-3-pierre-eric.pelloux-prayer@amd.com/#t Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5435 Fixes: 897ee11ec020 ("drm/amdgpu: create multiple clear/move ttm entities") Signed-off-by: Donet Tom Signed-off-by: Alex Deucher (cherry picked from commit 27213b776a666d3030de5acc3cd75278197b0494) Cc: stable@vger.kernel.org --- diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 00b5317f77f8..025625e7e800 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -208,9 +208,10 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity, void *cpu_addr; uint64_t flags; int r; + const u64 GTT_MAX_PAGES = (AMDGPU_GTT_MAX_TRANSFER_SIZE >> PAGE_SHIFT); BUG_ON(adev->mman.buffer_funcs->copy_max_bytes < - AMDGPU_GTT_MAX_TRANSFER_SIZE * 8); + GTT_MAX_PAGES * AMDGPU_GPU_PAGES_IN_CPU_PAGE * 8); if (WARN_ON(mem->mem_type == AMDGPU_PL_PREEMPT)) return -EINVAL; @@ -230,7 +231,7 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity, offset = mm_cur->start & ~PAGE_MASK; num_pages = PFN_UP(*size + offset); - num_pages = min_t(uint32_t, num_pages, AMDGPU_GTT_MAX_TRANSFER_SIZE); + num_pages = min_t(uint32_t, num_pages, GTT_MAX_PAGES); *size = min(*size, (uint64_t)num_pages * PAGE_SIZE - offset); @@ -2033,6 +2034,7 @@ static int amdgpu_ttm_buffer_entity_init(struct amdgpu_gtt_mgr *mgr, u32 num_gart_windows) { int i, r, num_pages; + const u64 GTT_MAX_PAGES = (AMDGPU_GTT_MAX_TRANSFER_SIZE >> PAGE_SHIFT); r = drm_sched_entity_init(&entity->base, prio, scheds, num_schedulers, NULL); if (r) @@ -2045,7 +2047,7 @@ static int amdgpu_ttm_buffer_entity_init(struct amdgpu_gtt_mgr *mgr, if (num_gart_windows == 0) return 0; - num_pages = num_gart_windows * AMDGPU_GTT_MAX_TRANSFER_SIZE; + num_pages = num_gart_windows * GTT_MAX_PAGES; r = amdgpu_gtt_mgr_alloc_entries(mgr, &entity->gart_node, num_pages, DRM_MM_INSERT_BEST); if (r) { @@ -2056,7 +2058,7 @@ static int amdgpu_ttm_buffer_entity_init(struct amdgpu_gtt_mgr *mgr, for (i = 0; i < num_gart_windows; i++) { entity->gart_window_offs[i] = amdgpu_gtt_node_to_byte_offset(&entity->gart_node) + - i * AMDGPU_GTT_MAX_TRANSFER_SIZE * PAGE_SIZE; + i * GTT_MAX_PAGES * PAGE_SIZE; } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 2d72fa217274..b5d938b31383 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -39,7 +39,7 @@ #define AMDGPU_PL_MMIO_REMAP (TTM_PL_PRIV + 5) #define __AMDGPU_PL_NUM (TTM_PL_PRIV + 6) -#define AMDGPU_GTT_MAX_TRANSFER_SIZE 1024 +#define AMDGPU_GTT_MAX_TRANSFER_SIZE (1ULL << 22) extern const struct attribute_group amdgpu_vram_mgr_attr_group; extern const struct attribute_group amdgpu_gtt_mgr_attr_group; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c index 226e76ae0be7..7cd236c1ff75 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c @@ -128,7 +128,7 @@ svm_migrate_copy_memory_gart(struct amdgpu_device *adev, dma_addr_t *sys, enum MIGRATION_COPY_DIR direction, struct dma_fence **mfence) { - const u64 GTT_MAX_PAGES = AMDGPU_GTT_MAX_TRANSFER_SIZE; + const u64 GTT_MAX_PAGES = (AMDGPU_GTT_MAX_TRANSFER_SIZE >> PAGE_SHIFT); struct amdgpu_ring *ring; struct amdgpu_ttm_buffer_entity *entity; u64 gart_s, gart_d;