We are already doing this for DMA-buf imports and also for amdgpu VM BOs for quite a while now.
If this doesn't run into any problems we are probably going to stop removing BOs from the LRU altogether.
v2: drop BUG_ON from ttm_bo_add_to_lru
Signed-off-by: Christian König christian.koenig@amd.com --- .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 9 ++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 4 ++-- drivers/gpu/drm/qxl/qxl_release.c | 2 +- drivers/gpu/drm/radeon/radeon_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 2 +- drivers/gpu/drm/ttm/ttm_bo.c | 23 ++++++++++--------- drivers/gpu/drm/ttm/ttm_execbuf_util.c | 20 +++++++++------- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 2 +- drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 3 ++- drivers/gpu/drm/vmwgfx/vmwgfx_validation.h | 2 +- include/drm/ttm/ttm_bo_driver.h | 5 +++- include/drm/ttm/ttm_execbuf_util.h | 3 ++- 14 files changed, 46 insertions(+), 35 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index e1cae4a37113..647e18f9e136 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -574,7 +574,7 @@ static int reserve_bo_and_vm(struct kgd_mem *mem, amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list, - false, &ctx->duplicates); + false, &ctx->duplicates, true); if (!ret) ctx->reserved = true; else { @@ -647,7 +647,7 @@ static int reserve_bo_and_cond_vms(struct kgd_mem *mem, }
ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list, - false, &ctx->duplicates); + false, &ctx->duplicates, true); if (!ret) ctx->reserved = true; else @@ -1800,7 +1800,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info) }
/* Reserve all BOs and page tables for validation */ - ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates); + ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates, + true); WARN(!list_empty(&duplicates), "Duplicates should be empty"); if (ret) goto out_free; @@ -2006,7 +2007,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef) }
ret = ttm_eu_reserve_buffers(&ctx.ticket, &ctx.list, - false, &duplicate_save); + false, &duplicate_save, true); if (ret) { pr_debug("Memory eviction: TTM Reserve Failed. Try again\n"); goto ttm_reserve_fail; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index d72cc583ebd1..fff558cf385b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -648,7 +648,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, }
r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, - &duplicates); + &duplicates, true); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) DRM_ERROR("ttm_eu_reserve_buffers failed.\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index 54dd02a898b9..06f83cac0d3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -79,7 +79,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, list_add(&csa_tv.head, &list); amdgpu_vm_get_pd_bo(vm, &list, &pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL); + r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true); if (r) { DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 7b840367004c..d513a5ad03dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -171,7 +171,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates); + r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true); if (r) { dev_err(adev->dev, "leaking bo va because " "we fail to reserve bo (%d)\n", r); @@ -608,7 +608,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates); + r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true); if (r) goto error_unref;
diff --git a/drivers/gpu/drm/qxl/qxl_release.c b/drivers/gpu/drm/qxl/qxl_release.c index 30f85f0130cb..49f9a9385393 100644 --- a/drivers/gpu/drm/qxl/qxl_release.c +++ b/drivers/gpu/drm/qxl/qxl_release.c @@ -256,7 +256,7 @@ int qxl_release_reserve_list(struct qxl_release *release, bool no_intr) return 0;
ret = ttm_eu_reserve_buffers(&release->ticket, &release->bos, - !no_intr, NULL); + !no_intr, NULL, true); if (ret) return ret;
diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index 44617dec8183..7411e69e2712 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -559,7 +559,7 @@ static void radeon_gem_va_update_vm(struct radeon_device *rdev, if (!vm_bos) return;
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL); + r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true); if (r) goto error_free;
diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 833e909706a9..36683de0300b 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -539,7 +539,7 @@ int radeon_bo_list_validate(struct radeon_device *rdev, u64 bytes_moved_threshold = radeon_bo_get_threshold_for_moves(rdev);
INIT_LIST_HEAD(&duplicates); - r = ttm_eu_reserve_buffers(ticket, head, true, &duplicates); + r = ttm_eu_reserve_buffers(ticket, head, true, &duplicates, true); if (unlikely(r != 0)) { return r; } diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 2845fceb2fbd..06bbcd2679b2 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -173,19 +173,20 @@ void ttm_bo_add_to_lru(struct ttm_buffer_object *bo)
reservation_object_assert_held(bo->resv);
- if (!(bo->mem.placement & TTM_PL_FLAG_NO_EVICT)) { - BUG_ON(!list_empty(&bo->lru)); + if (!list_empty(&bo->lru)) + return;
- man = &bdev->man[bo->mem.mem_type]; - list_add_tail(&bo->lru, &man->lru[bo->priority]); - kref_get(&bo->list_kref); + if (bo->mem.placement & TTM_PL_FLAG_NO_EVICT) + return;
- if (bo->ttm && !(bo->ttm->page_flags & - (TTM_PAGE_FLAG_SG | TTM_PAGE_FLAG_SWAPPED))) { - list_add_tail(&bo->swap, - &bdev->glob->swap_lru[bo->priority]); - kref_get(&bo->list_kref); - } + man = &bdev->man[bo->mem.mem_type]; + list_add_tail(&bo->lru, &man->lru[bo->priority]); + kref_get(&bo->list_kref); + + if (bo->ttm && !(bo->ttm->page_flags & + (TTM_PAGE_FLAG_SG | TTM_PAGE_FLAG_SWAPPED))) { + list_add_tail(&bo->swap, &bdev->glob->swap_lru[bo->priority]); + kref_get(&bo->list_kref); } } EXPORT_SYMBOL(ttm_bo_add_to_lru); diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c index 0075eb9a0b52..957ec375a4ba 100644 --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c @@ -69,7 +69,8 @@ void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket, list_for_each_entry(entry, list, head) { struct ttm_buffer_object *bo = entry->bo;
- ttm_bo_add_to_lru(bo); + if (list_empty(&bo->lru)) + ttm_bo_add_to_lru(bo); reservation_object_unlock(bo->resv); } spin_unlock(&glob->lru_lock); @@ -93,7 +94,7 @@ EXPORT_SYMBOL(ttm_eu_backoff_reservation);
int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket, struct list_head *list, bool intr, - struct list_head *dups) + struct list_head *dups, bool del_lru) { struct ttm_bo_global *glob; struct ttm_validate_buffer *entry; @@ -172,11 +173,11 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket, list_add(&entry->head, list); }
- if (ticket) - ww_acquire_done(ticket); - spin_lock(&glob->lru_lock); - ttm_eu_del_from_lru_locked(list); - spin_unlock(&glob->lru_lock); + if (del_lru) { + spin_lock(&glob->lru_lock); + ttm_eu_del_from_lru_locked(list); + spin_unlock(&glob->lru_lock); + } return 0; } EXPORT_SYMBOL(ttm_eu_reserve_buffers); @@ -203,7 +204,10 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket, reservation_object_add_shared_fence(bo->resv, fence); else reservation_object_add_excl_fence(bo->resv, fence); - ttm_bo_add_to_lru(bo); + if (list_empty(&bo->lru)) + ttm_bo_add_to_lru(bo); + else + ttm_bo_move_to_lru_tail(bo, NULL); reservation_object_unlock(bo->resv); } spin_unlock(&glob->lru_lock); diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 161b80fee492..5cffaa24259f 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -63,7 +63,7 @@ static int virtio_gpu_object_list_validate(struct ww_acquire_ctx *ticket, struct virtio_gpu_object *qobj; int ret;
- ret = ttm_eu_reserve_buffers(ticket, head, true, NULL); + ret = ttm_eu_reserve_buffers(ticket, head, true, NULL, true); if (ret != 0) return ret;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index a7c30e567f09..d28cbedba0b5 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -465,7 +465,8 @@ vmw_resource_check_buffer(struct ww_acquire_ctx *ticket, val_buf->bo = &res->backup->base; val_buf->num_shared = 0; list_add_tail(&val_buf->head, &val_list); - ret = ttm_eu_reserve_buffers(ticket, &val_list, interruptible, NULL); + ret = ttm_eu_reserve_buffers(ticket, &val_list, interruptible, NULL, + true); if (unlikely(ret != 0)) goto out_no_reserve;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_validation.h b/drivers/gpu/drm/vmwgfx/vmwgfx_validation.h index 3b396fea40d7..ac435b51f4eb 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_validation.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_validation.h @@ -165,7 +165,7 @@ vmw_validation_bo_reserve(struct vmw_validation_context *ctx, bool intr) { return ttm_eu_reserve_buffers(&ctx->ticket, &ctx->bo_list, intr, - NULL); + NULL, true); }
/** diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index c008346c2401..fc0d995ac90d 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -769,7 +769,10 @@ static inline void ttm_bo_unreserve(struct ttm_buffer_object *bo) { if (!(bo->mem.placement & TTM_PL_FLAG_NO_EVICT)) { spin_lock(&bo->bdev->glob->lru_lock); - ttm_bo_add_to_lru(bo); + if (list_empty(&bo->lru)) + ttm_bo_add_to_lru(bo); + else + ttm_bo_move_to_lru_tail(bo, NULL); spin_unlock(&bo->bdev->glob->lru_lock); } reservation_object_unlock(bo->resv); diff --git a/include/drm/ttm/ttm_execbuf_util.h b/include/drm/ttm/ttm_execbuf_util.h index 621615fa7728..7e46cc678e7e 100644 --- a/include/drm/ttm/ttm_execbuf_util.h +++ b/include/drm/ttm/ttm_execbuf_util.h @@ -70,6 +70,7 @@ extern void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket, * @list: thread private list of ttm_validate_buffer structs. * @intr: should the wait be interruptible * @dups: [out] optional list of duplicates. + * @del_lru: true if BOs should be removed from the LRU. * * Tries to reserve bos pointed to by the list entries for validation. * If the function returns 0, all buffers are marked as "unfenced", @@ -98,7 +99,7 @@ extern void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
extern int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket, struct list_head *list, bool intr, - struct list_head *dups); + struct list_head *dups, bool del_lru);
/** * function ttm_eu_fence_buffer_objects.
When a signal arrives we should return immediately for handling it and not try other placements first.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 06bbcd2679b2..7b59e5ecde7f 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -979,7 +979,6 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, uint32_t cur_flags = 0; bool type_found = false; bool type_ok = false; - bool has_erestartsys = false; int i, ret;
ret = reservation_object_reserve_shared(bo->resv, 1); @@ -1070,8 +1069,8 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, mem->placement = cur_flags; return 0; } - if (ret == -ERESTARTSYS) - has_erestartsys = true; + if (ret && ret != -EBUSY) + return ret; }
if (!type_found) { @@ -1079,7 +1078,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, return -EINVAL; }
- return (has_erestartsys) ? -ERESTARTSYS : -ENOMEM; + return -ENOMEM; } EXPORT_SYMBOL(ttm_bo_mem_space);
If drivers don't prefer a system memory placement they should not but it into the placement list first.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 7b59e5ecde7f..702cd89adbf9 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -1012,8 +1012,12 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, ttm_flag_masked(&cur_flags, place->flags, ~TTM_PL_MASK_MEMTYPE);
- if (mem_type == TTM_PL_SYSTEM) - break; + if (mem_type == TTM_PL_SYSTEM) { + mem->mem_type = mem_type; + mem->placement = cur_flags; + mem->mm_node = NULL; + return 0; + }
ret = (*man->func->get_node)(man, bo, place, mem); if (unlikely(ret)) @@ -1025,16 +1029,12 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, (*man->func->put_node)(man, mem); return ret; } - break; + mem->mem_type = mem_type; + mem->placement = cur_flags; + return 0; } }
- if ((type_ok && (mem_type == TTM_PL_SYSTEM)) || mem->mm_node) { - mem->mem_type = mem_type; - mem->placement = cur_flags; - return 0; - } - for (i = 0; i < placement->num_busy_placement; ++i) { const struct ttm_place *place = &placement->busy_placement[i];
We tried this once before, but that turned out to be more complicated than thought. With all the right prerequisites it looks like we can do this now.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 127 ++++++++++++++++++----------------- 1 file changed, 66 insertions(+), 61 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 702cd89adbf9..0dbb23b0dedd 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -893,13 +893,12 @@ static int ttm_bo_add_move_fence(struct ttm_buffer_object *bo, * space, or we've evicted everything and there isn't enough space. */ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, - uint32_t mem_type, - const struct ttm_place *place, - struct ttm_mem_reg *mem, - struct ttm_operation_ctx *ctx) + const struct ttm_place *place, + struct ttm_mem_reg *mem, + struct ttm_operation_ctx *ctx) { struct ttm_bo_device *bdev = bo->bdev; - struct ttm_mem_type_manager *man = &bdev->man[mem_type]; + struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type]; int ret;
do { @@ -908,11 +907,11 @@ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, return ret; if (mem->mm_node) break; - ret = ttm_mem_evict_first(bdev, mem_type, place, ctx); + ret = ttm_mem_evict_first(bdev, mem->mem_type, place, ctx); if (unlikely(ret != 0)) return ret; } while (1); - mem->mem_type = mem_type; + return ttm_bo_add_move_fence(bo, man, mem); }
@@ -960,6 +959,51 @@ static bool ttm_bo_mt_compatible(struct ttm_mem_type_manager *man, return true; }
+/** + * ttm_bo_mem_placement - check if placement is compatible + * @bo: BO to find memory for + * @place: where to search + * @mem: the memory object to fill in + * @ctx: operation context + * + * Check if placement is compatible and fill in mem structure. + * Returns -EBUSY if placement won't work or negative error code. + * 0 when placement can be used. + */ +static int ttm_bo_mem_placement(struct ttm_buffer_object *bo, + const struct ttm_place *place, + struct ttm_mem_reg *mem, + struct ttm_operation_ctx *ctx) +{ + struct ttm_bo_device *bdev = bo->bdev; + uint32_t mem_type = TTM_PL_SYSTEM; + struct ttm_mem_type_manager *man; + uint32_t cur_flags = 0; + int ret; + + ret = ttm_mem_type_from_place(place, &mem_type); + if (ret) + return ret; + + man = &bdev->man[mem_type]; + if (!man->has_type || !man->use_type) + return -EBUSY; + + if (!ttm_bo_mt_compatible(man, mem_type, place, &cur_flags)) + return -EBUSY; + + cur_flags = ttm_bo_select_caching(man, bo->mem.placement, cur_flags); + /* + * Use the access and other non-mapping-related flag bits from + * the memory placement flags to the current flags + */ + ttm_flag_masked(&cur_flags, place->flags, ~TTM_PL_MASK_MEMTYPE); + + mem->mem_type = mem_type; + mem->placement = cur_flags; + return 0; +} + /** * Creates space for memory region @mem according to its type. * @@ -974,11 +1018,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx) { struct ttm_bo_device *bdev = bo->bdev; - struct ttm_mem_type_manager *man; - uint32_t mem_type = TTM_PL_SYSTEM; - uint32_t cur_flags = 0; bool type_found = false; - bool type_ok = false; int i, ret;
ret = reservation_object_reserve_shared(bo->resv, 1); @@ -988,37 +1028,20 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, mem->mm_node = NULL; for (i = 0; i < placement->num_placement; ++i) { const struct ttm_place *place = &placement->placement[i]; + struct ttm_mem_type_manager *man;
- ret = ttm_mem_type_from_place(place, &mem_type); + ret = ttm_bo_mem_placement(bo, place, mem, ctx); + if (ret == -EBUSY) + continue; if (ret) return ret; - man = &bdev->man[mem_type]; - if (!man->has_type || !man->use_type) - continue; - - type_ok = ttm_bo_mt_compatible(man, mem_type, place, - &cur_flags); - - if (!type_ok) - continue;
type_found = true; - cur_flags = ttm_bo_select_caching(man, bo->mem.placement, - cur_flags); - /* - * Use the access and other non-mapping-related flag bits from - * the memory placement flags to the current flags - */ - ttm_flag_masked(&cur_flags, place->flags, - ~TTM_PL_MASK_MEMTYPE); - - if (mem_type == TTM_PL_SYSTEM) { - mem->mem_type = mem_type; - mem->placement = cur_flags; - mem->mm_node = NULL; + mem->mm_node = NULL; + if (mem->mem_type == TTM_PL_SYSTEM) return 0; - }
+ man = &bdev->man[mem->mem_type]; ret = (*man->func->get_node)(man, bo, place, mem); if (unlikely(ret)) return ret; @@ -1029,8 +1052,6 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, (*man->func->put_node)(man, mem); return ret; } - mem->mem_type = mem_type; - mem->placement = cur_flags; return 0; } } @@ -1038,37 +1059,21 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, for (i = 0; i < placement->num_busy_placement; ++i) { const struct ttm_place *place = &placement->busy_placement[i];
- ret = ttm_mem_type_from_place(place, &mem_type); + ret = ttm_bo_mem_placement(bo, place, mem, ctx); + if (ret == -EBUSY) + continue; if (ret) return ret; - man = &bdev->man[mem_type]; - if (!man->has_type || !man->use_type) - continue; - if (!ttm_bo_mt_compatible(man, mem_type, place, &cur_flags)) - continue;
type_found = true; - cur_flags = ttm_bo_select_caching(man, bo->mem.placement, - cur_flags); - /* - * Use the access and other non-mapping-related flag bits from - * the memory placement flags to the current flags - */ - ttm_flag_masked(&cur_flags, place->flags, - ~TTM_PL_MASK_MEMTYPE); - - if (mem_type == TTM_PL_SYSTEM) { - mem->mem_type = mem_type; - mem->placement = cur_flags; - mem->mm_node = NULL; + mem->mm_node = NULL; + if (mem->mem_type == TTM_PL_SYSTEM) return 0; - }
- ret = ttm_bo_mem_force_space(bo, mem_type, place, mem, ctx); - if (ret == 0 && mem->mm_node) { - mem->placement = cur_flags; + ret = ttm_bo_mem_force_space(bo, place, mem, ctx); + if (ret == 0 && mem->mm_node) return 0; - } + if (ret && ret != -EBUSY) return ret; }
Move BOs which are currently in a lower domain to the new LRU before allocating backing space while validating.
This makes sure that we always have enough entries on the LRU to allow for other processes to wait for an operation to complete.
v2: generalize the test v3: fix rebase error
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 42 +++++++++++++++++++++++++++--------- 1 file changed, 32 insertions(+), 10 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 0dbb23b0dedd..8a8859cf60e8 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -166,7 +166,8 @@ static void ttm_bo_release_list(struct kref *list_kref) ttm_mem_global_free(bdev->glob->mem_glob, acc_size); }
-void ttm_bo_add_to_lru(struct ttm_buffer_object *bo) +static void ttm_bo_add_mem_to_lru(struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem) { struct ttm_bo_device *bdev = bo->bdev; struct ttm_mem_type_manager *man; @@ -176,10 +177,10 @@ void ttm_bo_add_to_lru(struct ttm_buffer_object *bo) if (!list_empty(&bo->lru)) return;
- if (bo->mem.placement & TTM_PL_FLAG_NO_EVICT) + if (mem->placement & TTM_PL_FLAG_NO_EVICT) return;
- man = &bdev->man[bo->mem.mem_type]; + man = &bdev->man[mem->mem_type]; list_add_tail(&bo->lru, &man->lru[bo->priority]); kref_get(&bo->list_kref);
@@ -189,6 +190,11 @@ void ttm_bo_add_to_lru(struct ttm_buffer_object *bo) kref_get(&bo->list_kref); } } + +void ttm_bo_add_to_lru(struct ttm_buffer_object *bo) +{ + ttm_bo_add_mem_to_lru(bo, &bo->mem); +} EXPORT_SYMBOL(ttm_bo_add_to_lru);
static void ttm_bo_ref_bug(struct kref *list_kref) @@ -1001,6 +1007,14 @@ static int ttm_bo_mem_placement(struct ttm_buffer_object *bo,
mem->mem_type = mem_type; mem->placement = cur_flags; + + if (bo->mem.mem_type < mem_type && !list_empty(&bo->lru)) { + spin_lock(&bo->bdev->glob->lru_lock); + ttm_bo_del_from_lru(bo); + ttm_bo_add_mem_to_lru(bo, mem); + spin_unlock(&bo->bdev->glob->lru_lock); + } + return 0; }
@@ -1034,7 +1048,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, if (ret == -EBUSY) continue; if (ret) - return ret; + goto error;
type_found = true; mem->mm_node = NULL; @@ -1044,13 +1058,13 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, man = &bdev->man[mem->mem_type]; ret = (*man->func->get_node)(man, bo, place, mem); if (unlikely(ret)) - return ret; + goto error;
if (mem->mm_node) { ret = ttm_bo_add_move_fence(bo, man, mem); if (unlikely(ret)) { (*man->func->put_node)(man, mem); - return ret; + goto error; } return 0; } @@ -1063,7 +1077,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, if (ret == -EBUSY) continue; if (ret) - return ret; + goto error;
type_found = true; mem->mm_node = NULL; @@ -1075,15 +1089,23 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, return 0;
if (ret && ret != -EBUSY) - return ret; + goto error; }
+ ret = -ENOMEM; if (!type_found) { pr_err(TTM_PFX "No compatible memory type found\n"); - return -EINVAL; + ret = -EINVAL; }
- return -ENOMEM; +error: + if (bo->mem.mem_type == TTM_PL_SYSTEM && !list_empty(&bo->lru)) { + spin_lock(&bo->bdev->glob->lru_lock); + ttm_bo_move_to_lru_tail(bo, NULL); + spin_unlock(&bo->bdev->glob->lru_lock); + } + + return ret; } EXPORT_SYMBOL(ttm_bo_mem_space);
BOs on the LRU might be blocked during command submission and cause OOM situations.
Avoid this by blocking for the first busy BO not locked by the same ticket as the BO we are searching space for.
v10: completely start over with the patch since we didn't handled a whole bunch of corner cases.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 77 ++++++++++++++++++++++++++++++------ 1 file changed, 66 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 8a8859cf60e8..c7de667d482a 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -773,32 +773,72 @@ EXPORT_SYMBOL(ttm_bo_eviction_valuable); * b. Otherwise, trylock it. */ static bool ttm_bo_evict_swapout_allowable(struct ttm_buffer_object *bo, - struct ttm_operation_ctx *ctx, bool *locked) + struct ttm_operation_ctx *ctx, bool *locked, bool *busy) { bool ret = false;
- *locked = false; if (bo->resv == ctx->resv) { reservation_object_assert_held(bo->resv); if (ctx->flags & TTM_OPT_FLAG_ALLOW_RES_EVICT || !list_empty(&bo->ddestroy)) ret = true; + *locked = false; + if (busy) + *busy = false; } else { - *locked = reservation_object_trylock(bo->resv); - ret = *locked; + ret = reservation_object_trylock(bo->resv); + *locked = ret; + if (busy) + *busy = !ret; }
return ret; }
+/** + * ttm_mem_evict_wait_busy - wait for a busy BO to become available + * + * @busy_bo: BO which couldn't be locked with trylock + * @ctx: operation context + * @ticket: acquire ticket + * + * Try to lock a busy buffer object to avoid failing eviction. + */ +static int ttm_mem_evict_wait_busy(struct ttm_buffer_object *busy_bo, + struct ttm_operation_ctx *ctx, + struct ww_acquire_ctx *ticket) +{ + int r; + + if (!busy_bo || !ticket) + return -EBUSY; + + if (ctx->interruptible) + r = reservation_object_lock_interruptible(busy_bo->resv, + ticket); + else + r = reservation_object_lock(busy_bo->resv, ticket); + + /* + * TODO: It would be better to keep the BO locked until allocation is at + * least tried one more time, but that would mean a much larger rework + * of TTM. + */ + if (!r) + reservation_object_unlock(busy_bo->resv); + + return r == -EDEADLK ? -EAGAIN : r; +} + static int ttm_mem_evict_first(struct ttm_bo_device *bdev, uint32_t mem_type, const struct ttm_place *place, - struct ttm_operation_ctx *ctx) + struct ttm_operation_ctx *ctx, + struct ww_acquire_ctx *ticket) { + struct ttm_buffer_object *bo = NULL, *busy_bo = NULL; struct ttm_bo_global *glob = bdev->glob; struct ttm_mem_type_manager *man = &bdev->man[mem_type]; - struct ttm_buffer_object *bo = NULL; bool locked = false; unsigned i; int ret; @@ -806,8 +846,15 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev, spin_lock(&glob->lru_lock); for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { list_for_each_entry(bo, &man->lru[i], lru) { - if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked)) + bool busy; + + if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked, + &busy)) { + if (busy && !busy_bo && + bo->resv->lock.ctx != ticket) + busy_bo = bo; continue; + }
if (place && !bdev->driver->eviction_valuable(bo, place)) { @@ -826,8 +873,13 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev, }
if (!bo) { + if (busy_bo) + ttm_bo_get(busy_bo); spin_unlock(&glob->lru_lock); - return -EBUSY; + ret = ttm_mem_evict_wait_busy(busy_bo, ctx, ticket); + if (busy_bo) + ttm_bo_put(busy_bo); + return ret; }
kref_get(&bo->list_kref); @@ -913,7 +965,8 @@ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, return ret; if (mem->mm_node) break; - ret = ttm_mem_evict_first(bdev, mem->mem_type, place, ctx); + ret = ttm_mem_evict_first(bdev, mem->mem_type, place, ctx, + bo->resv->lock.ctx); if (unlikely(ret != 0)) return ret; } while (1); @@ -1428,7 +1481,8 @@ static int ttm_bo_force_list_clean(struct ttm_bo_device *bdev, for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { while (!list_empty(&man->lru[i])) { spin_unlock(&glob->lru_lock); - ret = ttm_mem_evict_first(bdev, mem_type, NULL, &ctx); + ret = ttm_mem_evict_first(bdev, mem_type, NULL, &ctx, + NULL); if (ret) return ret; spin_lock(&glob->lru_lock); @@ -1799,7 +1853,8 @@ int ttm_bo_swapout(struct ttm_bo_global *glob, struct ttm_operation_ctx *ctx) spin_lock(&glob->lru_lock); for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { list_for_each_entry(bo, &glob->swap_lru[i], swap) { - if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked)) { + if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked, + NULL)) { ret = 0; break; }
From: Chunming Zhou david1.zhou@amd.com
add ticket for display bo, so that it can preempt busy bo.
v2: fix stupid rebase error
Change-Id: I9f031cdcc8267de00e819ae303baa0a52df8ebb9 Signed-off-by: Chunming Zhou david1.zhou@amd.com Reviewed-by: Christian König christian.koenig@amd.com --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 21 ++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 4a1755bce96c..56f320f3fd72 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -4182,6 +4182,9 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane, struct amdgpu_device *adev; struct amdgpu_bo *rbo; struct dm_plane_state *dm_plane_state_new, *dm_plane_state_old; + struct list_head list; + struct ttm_validate_buffer tv; + struct ww_acquire_ctx ticket; uint64_t tiling_flags; uint32_t domain; int r; @@ -4198,9 +4201,17 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane, obj = new_state->fb->obj[0]; rbo = gem_to_amdgpu_bo(obj); adev = amdgpu_ttm_adev(rbo->tbo.bdev); - r = amdgpu_bo_reserve(rbo, false); - if (unlikely(r != 0)) + INIT_LIST_HEAD(&list); + + tv.bo = &rbo->tbo; + tv.num_shared = 1; + list_add(&tv.head, &list); + + r = ttm_eu_reserve_buffers(&ticket, &list, false, NULL, true); + if (r) { + dev_err(adev->dev, "fail to reserve bo (%d)\n", r); return r; + }
if (plane->type != DRM_PLANE_TYPE_CURSOR) domain = amdgpu_display_supported_domains(adev); @@ -4211,21 +4222,21 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane, if (unlikely(r != 0)) { if (r != -ERESTARTSYS) DRM_ERROR("Failed to pin framebuffer with error %d\n", r); - amdgpu_bo_unreserve(rbo); + ttm_eu_backoff_reservation(&ticket, &list); return r; }
r = amdgpu_ttm_alloc_gart(&rbo->tbo); if (unlikely(r != 0)) { amdgpu_bo_unpin(rbo); - amdgpu_bo_unreserve(rbo); + ttm_eu_backoff_reservation(&ticket, &list); DRM_ERROR("%p bind failed\n", rbo); return r; }
amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
- amdgpu_bo_unreserve(rbo); + ttm_eu_backoff_reservation(&ticket, &list);
afb->address = amdgpu_bo_gpu_offset(rbo);
The messages about amdgpu_cs_list_validate are duplicated because the caller will complain into the logs as well and we can also get interrupted by a signal here.
Also fix the the caller to not report -EAGAIN from validation.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index fff558cf385b..20f2955d2a55 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -671,16 +671,12 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, }
r = amdgpu_cs_list_validate(p, &duplicates); - if (r) { - DRM_ERROR("amdgpu_cs_list_validate(duplicates) failed.\n"); + if (r) goto error_validate; - }
r = amdgpu_cs_list_validate(p, &p->validated); - if (r) { - DRM_ERROR("amdgpu_cs_list_validate(validated) failed.\n"); + if (r) goto error_validate; - }
amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved, p->bytes_moved_vis); @@ -1383,7 +1379,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) if (r) { if (r == -ENOMEM) DRM_ERROR("Not enough memory for command submission!\n"); - else if (r != -ERESTARTSYS) + else if (r != -ERESTARTSYS && r != -EAGAIN) DRM_ERROR("Failed to process the buffer list %d!\n", r); goto out; }
And only move them in on validation. This allows for better control when multiple processes are fighting over those resources.
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 93b2c5a48a71..30493429851e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -495,7 +495,11 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, #endif
bo->tbo.bdev = &adev->mman.bdev; - amdgpu_bo_placement_from_domain(bo, bp->domain); + if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA | + AMDGPU_GEM_DOMAIN_GDS)) + amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU); + else + amdgpu_bo_placement_from_domain(bo, bp->domain); if (bp->type == ttm_bo_type_kernel) bo->tbo.priority = 1;
This avoids OOM situations when we have lots of threads submitting at the same time.
v3: apply this to the whole driver, not just CS
Signed-off-by: Christian König christian.koenig@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 20f2955d2a55..3e2da24cd17a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -648,7 +648,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, }
r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, - &duplicates, true); + &duplicates, false); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) DRM_ERROR("ttm_eu_reserve_buffers failed.\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index 06f83cac0d3a..f660628e6af9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -79,7 +79,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, list_add(&csa_tv.head, &list); amdgpu_vm_get_pd_bo(vm, &list, &pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true); + r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, false); if (r) { DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index d513a5ad03dd..ed25a4e14404 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -171,7 +171,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true); + r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, false); if (r) { dev_err(adev->dev, "leaking bo va because " "we fail to reserve bo (%d)\n", r); @@ -608,7 +608,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true); + r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, false); if (r) goto error_unref;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index c430e8259038..d60593cc436e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -155,7 +155,7 @@ static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr) struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); int r;
- r = ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL); + r = __ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) dev_err(adev->dev, "%p reserve failed\n", bo);
Patch #1,#5,#6,#8,#9,#10 are Reviewed-by: Chunming Zhou david1.zhou@amd.com Patch #2,#3,#4 are Acked-by: Chunming Zhou david1.zhou@amd.com
-David
-----Original Message----- From: dri-devel dri-devel-bounces@lists.freedesktop.org On Behalf Of Christian K?nig Sent: Wednesday, May 29, 2019 8:27 PM To: dri-devel@lists.freedesktop.org; amd-gfx@lists.freedesktop.org Subject: [PATCH 10/10] drm/amdgpu: stop removing BOs from the LRU v3
This avoids OOM situations when we have lots of threads submitting at the same time.
v3: apply this to the whole driver, not just CS
Signed-off-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 20f2955d2a55..3e2da24cd17a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -648,7 +648,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, }
r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
&duplicates, true);
&duplicates, false);diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index 06f83cac0d3a..f660628e6af9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -79,7 +79,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, list_add(&csa_tv.head, &list); amdgpu_vm_get_pd_bo(vm, &list, &pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, false); if (r) { DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index d513a5ad03dd..ed25a4e14404 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -171,7 +171,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, false); if (r) { dev_err(adev->dev, "leaking bo va because " "we fail to reserve bo (%d)\n", r);
@@ -608,7 +608,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, false); if (r) goto error_unref;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index c430e8259038..d60593cc436e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -155,7 +155,7 @@ static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr) struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); int r;
- r = ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL);
- r = __ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) dev_err(adev->dev, "%p reserve failed\n", bo);
-- 2.17.1
dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi Christian,
The series is:
Tested-by: Pierre-Eric Pelloux-Prayer pierre-eric.pelloux-prayer@amd.com
Pierre-Eric
On 29/05/2019 14:27, Christian König wrote:
This avoids OOM situations when we have lots of threads submitting at the same time.
v3: apply this to the whole driver, not just CS
Signed-off-by: Christian König christian.koenig@amd.com
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 20f2955d2a55..3e2da24cd17a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -648,7 +648,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, }
r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
&duplicates, true);
&duplicates, false);diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index 06f83cac0d3a..f660628e6af9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -79,7 +79,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, list_add(&csa_tv.head, &list); amdgpu_vm_get_pd_bo(vm, &list, &pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, false); if (r) { DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index d513a5ad03dd..ed25a4e14404 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -171,7 +171,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, false); if (r) { dev_err(adev->dev, "leaking bo va because " "we fail to reserve bo (%d)\n", r);
@@ -608,7 +608,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true);
- r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, false); if (r) goto error_unref;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index c430e8259038..d60593cc436e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -155,7 +155,7 @@ static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr) struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); int r;
- r = ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL);
- r = __ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) dev_err(adev->dev, "%p reserve failed\n", bo);
dri-devel@lists.freedesktop.org
-
Christian König -
Pelloux-prayer, Pierre-eric -
Zhou, David(ChunMing)