From 386021394394eccef248dc5eb9c9370240821a8c Mon Sep 17 00:00:00 2001 From: Matthew Brost Date: Tue, 12 Mar 2024 11:39:07 -0700 Subject: drm/xe: Invalidate userptr VMA on page pin fault MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Rather than return an error to the user or ban the VM when userptr VMA page pin fails with -EFAULT, invalidate VMA mappings. This supports the UMD use case of freeing userptr while still having bindings. Now that non-faulting VMs can invalidate VMAs, drop the usm prefix for the tile_invalidated member. v2: - Fix build error (CI) v3: - Don't invalidate VMA if in fault mode, rather kill VM (Thomas) - Update commit message with tile_invalidated name chagne (Thomas) - Wait VM bookkeep slots with VM resv lock (Thomas) v4: - Move list_del_init(&userptr.repin_link) after error check (Thomas) - Assert not in fault mode (Matthew) Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Signed-off-by: Matthew Brost Reviewed-by: Thomas Hellström Link: https://patchwork.freedesktop.org/patch/msgid/20240312183907.933835-1-matthew.brost@intel.com (cherry picked from commit 521db22a1d70dbc596a07544a738416025b1b63c) Signed-off-by: Lucas De Marchi --- drivers/gpu/drm/xe/xe_gt_pagefault.c | 4 ++-- drivers/gpu/drm/xe/xe_trace.h | 2 +- drivers/gpu/drm/xe/xe_vm.c | 32 ++++++++++++++++++++++++-------- drivers/gpu/drm/xe/xe_vm_types.h | 7 ++----- 4 files changed, 29 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c index 73c535193a98..241c294270d9 100644 --- a/drivers/gpu/drm/xe/xe_gt_pagefault.c +++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c @@ -69,7 +69,7 @@ static bool access_is_atomic(enum access_type access_type) static bool vma_is_valid(struct xe_tile *tile, struct xe_vma *vma) { return BIT(tile->id) & vma->tile_present && - !(BIT(tile->id) & vma->usm.tile_invalidated); + !(BIT(tile->id) & vma->tile_invalidated); } static bool vma_matches(struct xe_vma *vma, u64 page_addr) @@ -226,7 +226,7 @@ retry_userptr: if (xe_vma_is_userptr(vma)) ret = xe_vma_userptr_check_repin(to_userptr_vma(vma)); - vma->usm.tile_invalidated &= ~BIT(tile->id); + vma->tile_invalidated &= ~BIT(tile->id); unlock_dma_resv: drm_exec_fini(&exec); diff --git a/drivers/gpu/drm/xe/xe_trace.h b/drivers/gpu/drm/xe/xe_trace.h index 3b97633d81d8..d82c138f1ece 100644 --- a/drivers/gpu/drm/xe/xe_trace.h +++ b/drivers/gpu/drm/xe/xe_trace.h @@ -464,7 +464,7 @@ DEFINE_EVENT(xe_vma, xe_vma_userptr_invalidate, TP_ARGS(vma) ); -DEFINE_EVENT(xe_vma, xe_vma_usm_invalidate, +DEFINE_EVENT(xe_vma, xe_vma_invalidate, TP_PROTO(struct xe_vma *vma), TP_ARGS(vma) ); diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index e3bde897f6e8..3dd8bb64bb00 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -708,6 +708,7 @@ int xe_vm_userptr_pin(struct xe_vm *vm) int err = 0; LIST_HEAD(tmp_evict); + xe_assert(vm->xe, !xe_vm_in_fault_mode(vm)); lockdep_assert_held_write(&vm->lock); /* Collect invalidated userptrs */ @@ -724,11 +725,27 @@ int xe_vm_userptr_pin(struct xe_vm *vm) list_for_each_entry_safe(uvma, next, &vm->userptr.repin_list, userptr.repin_link) { err = xe_vma_userptr_pin_pages(uvma); - if (err < 0) - return err; + if (err == -EFAULT) { + list_del_init(&uvma->userptr.repin_link); - list_del_init(&uvma->userptr.repin_link); - list_move_tail(&uvma->vma.combined_links.rebind, &vm->rebind_list); + /* Wait for pending binds */ + xe_vm_lock(vm, false); + dma_resv_wait_timeout(xe_vm_resv(vm), + DMA_RESV_USAGE_BOOKKEEP, + false, MAX_SCHEDULE_TIMEOUT); + + err = xe_vm_invalidate_vma(&uvma->vma); + xe_vm_unlock(vm); + if (err) + return err; + } else { + if (err < 0) + return err; + + list_del_init(&uvma->userptr.repin_link); + list_move_tail(&uvma->vma.combined_links.rebind, + &vm->rebind_list); + } } return 0; @@ -1987,7 +2004,7 @@ static int xe_vm_prefetch(struct xe_vm *vm, struct xe_vma *vma, return err; } - if (vma->tile_mask != (vma->tile_present & ~vma->usm.tile_invalidated)) { + if (vma->tile_mask != (vma->tile_present & ~vma->tile_invalidated)) { return xe_vm_bind(vm, vma, q, xe_vma_bo(vma), syncs, num_syncs, true, first_op, last_op); } else { @@ -3185,9 +3202,8 @@ int xe_vm_invalidate_vma(struct xe_vma *vma) u8 id; int ret; - xe_assert(xe, xe_vm_in_fault_mode(xe_vma_vm(vma))); xe_assert(xe, !xe_vma_is_null(vma)); - trace_xe_vma_usm_invalidate(vma); + trace_xe_vma_invalidate(vma); /* Check that we don't race with page-table updates */ if (IS_ENABLED(CONFIG_PROVE_LOCKING)) { @@ -3225,7 +3241,7 @@ int xe_vm_invalidate_vma(struct xe_vma *vma) } } - vma->usm.tile_invalidated = vma->tile_mask; + vma->tile_invalidated = vma->tile_mask; return 0; } diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h index 292f8cadb40f..713996f7dc59 100644 --- a/drivers/gpu/drm/xe/xe_vm_types.h +++ b/drivers/gpu/drm/xe/xe_vm_types.h @@ -83,11 +83,8 @@ struct xe_vma { struct work_struct destroy_work; }; - /** @usm: unified shared memory state */ - struct { - /** @tile_invalidated: VMA has been invalidated */ - u8 tile_invalidated; - } usm; + /** @tile_invalidated: VMA has been invalidated */ + u8 tile_invalidated; /** @tile_mask: Tile mask of where to create binding for this VMA */ u8 tile_mask; -- cgit v1.2.3 From d58b4ef63b5024993906e74f04fda8220ad4c162 Mon Sep 17 00:00:00 2001 From: Himal Prasad Ghimiray Date: Wed, 13 Mar 2024 20:35:44 +0530 Subject: drm/xe: Return if kobj creation is failed Return after warning regarding kobj creation failure. Fixes: 4ae3aeab32d7 ("drm/xe: Add vram frequency sysfs attributes") Cc: Sujaritha Sundaresan Cc: Tejas Upadhyay Cc: Bommu Krishnaiah Reviewed-by: Tejas Upadhyay Signed-off-by: Himal Prasad Ghimiray Link: https://patchwork.freedesktop.org/patch/msgid/20240313150545.2830408-2-himal.prasad.ghimiray@intel.com Signed-off-by: Rodrigo Vivi (cherry picked from commit 989d07ac6bb7d269e975f85e8f683f496faa0380) Signed-off-by: Lucas De Marchi --- drivers/gpu/drm/xe/xe_vram_freq.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/xe/xe_vram_freq.c b/drivers/gpu/drm/xe/xe_vram_freq.c index 079cc283a186..c5f6b5a5d117 100644 --- a/drivers/gpu/drm/xe/xe_vram_freq.c +++ b/drivers/gpu/drm/xe/xe_vram_freq.c @@ -111,8 +111,10 @@ void xe_vram_freq_sysfs_init(struct xe_tile *tile) return; kobj = kobject_create_and_add("memory", tile->sysfs); - if (!kobj) + if (!kobj) { drm_warn(&xe->drm, "failed to add memory directory, err: %d\n", -ENOMEM); + return; + } err = sysfs_create_group(kobj, &freq_group_attrs); if (err) { -- cgit v1.2.3 From dd8a07f06dfd946e0eea1a3323d52e7c28a6ed80 Mon Sep 17 00:00:00 2001 From: José Roberto de Souza Date: Wed, 13 Mar 2024 10:13:18 -0700 Subject: drm/xe: Skip VMAs pin when requesting signal to the last XE_EXEC MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Doing a XE_EXEC with num_batch_buffer == 0 makes signals passed as argument to be signaled when the last real XE_EXEC is completed. But to do that it was first pinning all VMAs in drm_gpuvm_exec_lock(), this patch remove this pinning as it is not required. This change also help Mesa implementing memory over-commiting recovery as it needs to unbind not needed VMAs when the whole VM can't fit in GPU memory but it can only do the unbiding when the last XE_EXEC is completed. So with this change Mesa can get the signal it want without getting out-of-memory errors. Fixes: eb9702ad2986 ("drm/xe: Allow num_batch_buffer / num_binds == 0 in IOCTLs") Cc: Thomas Hellstrom Co-developed-by: Matthew Brost Signed-off-by: José Roberto de Souza Reviewed-by: Matthew Brost Signed-off-by: Matthew Brost Link: https://patchwork.freedesktop.org/patch/msgid/20240313171318.121066-1-jose.souza@intel.com (cherry picked from commit 58480c1c912ff8146d067301a0d04cca318b4a66) Signed-off-by: Lucas De Marchi --- drivers/gpu/drm/xe/xe_exec.c | 41 +++++++++++++++++++++++------------------ 1 file changed, 23 insertions(+), 18 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c index 952496c6260d..826c8b389672 100644 --- a/drivers/gpu/drm/xe/xe_exec.c +++ b/drivers/gpu/drm/xe/xe_exec.c @@ -235,6 +235,29 @@ retry: goto err_unlock_list; } + if (!args->num_batch_buffer) { + err = xe_vm_lock(vm, true); + if (err) + goto err_unlock_list; + + if (!xe_vm_in_lr_mode(vm)) { + struct dma_fence *fence; + + fence = xe_sync_in_fence_get(syncs, num_syncs, q, vm); + if (IS_ERR(fence)) { + err = PTR_ERR(fence); + goto err_unlock_list; + } + for (i = 0; i < num_syncs; i++) + xe_sync_entry_signal(&syncs[i], NULL, fence); + xe_exec_queue_last_fence_set(q, vm, fence); + dma_fence_put(fence); + } + + xe_vm_unlock(vm); + goto err_unlock_list; + } + vm_exec.vm = &vm->gpuvm; vm_exec.flags = DRM_EXEC_INTERRUPTIBLE_WAIT; if (xe_vm_in_lr_mode(vm)) { @@ -254,24 +277,6 @@ retry: goto err_exec; } - if (!args->num_batch_buffer) { - if (!xe_vm_in_lr_mode(vm)) { - struct dma_fence *fence; - - fence = xe_sync_in_fence_get(syncs, num_syncs, q, vm); - if (IS_ERR(fence)) { - err = PTR_ERR(fence); - goto err_exec; - } - for (i = 0; i < num_syncs; i++) - xe_sync_entry_signal(&syncs[i], NULL, fence); - xe_exec_queue_last_fence_set(q, vm, fence); - dma_fence_put(fence); - } - - goto err_exec; - } - if (xe_exec_queue_is_lr(q) && xe_exec_queue_ring_full(q)) { err = -EWOULDBLOCK; /* Aliased to -EAGAIN */ skip_retry = true; -- cgit v1.2.3