diff options
Diffstat (limited to 'meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.14.71/2799-drm-amdgpu-Get-rid-of-dep_sync-as-a-seperate-object.patch')
| -rw-r--r-- | meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.14.71/2799-drm-amdgpu-Get-rid-of-dep_sync-as-a-seperate-object.patch | 331 |
1 files changed, 331 insertions, 0 deletions
diff --git a/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.14.71/2799-drm-amdgpu-Get-rid-of-dep_sync-as-a-seperate-object.patch b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.14.71/2799-drm-amdgpu-Get-rid-of-dep_sync-as-a-seperate-object.patch new file mode 100644 index 00000000..ab555ee7 --- /dev/null +++ b/meta-amd-bsp/recipes-kernel/linux/linux-yocto-4.14.71/2799-drm-amdgpu-Get-rid-of-dep_sync-as-a-seperate-object.patch @@ -0,0 +1,331 @@ +From 5fe0477277728251cd808cb3730b06ebb55fed78 Mon Sep 17 00:00:00 2001 +From: Andrey Grodzovsky <andrey.grodzovsky@amd.com> +Date: Mon, 13 Nov 2017 14:47:52 -0500 +Subject: [PATCH 2799/4131] drm/amdgpu: Get rid of dep_sync as a seperate + object. +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +Instead mark fence as explicit in it's amdgpu_sync_entry. + +v2: +Fix use after free bug and add new parameter description. +Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> +Reviewed-by: Christian König <christian.koenig@amd.com> +Signed-off-by: Alex Deucher <alexander.deucher@amd.com> + +Change-Id: I4e53b939b05856576b34d0c9aeadb808aac8b58e +--- + drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 - + drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 14 +++++++------- + drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 2 +- + drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 24 +++++++++++------------- + drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c | 14 ++++++++++---- + drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h | 4 ++-- + drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 ++++++------ + 7 files changed, 37 insertions(+), 34 deletions(-) + +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h +index d57c7ed..c32bfc4 100755 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h +@@ -1145,7 +1145,6 @@ struct amdgpu_job { + struct amdgpu_vm *vm; + struct amdgpu_ring *ring; + struct amdgpu_sync sync; +- struct amdgpu_sync dep_sync; + struct amdgpu_sync sched_sync; + struct amdgpu_ib *ibs; + struct dma_fence *fence; /* the hw fence */ +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +index 8b724b3..cbf9120 100644 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +@@ -827,7 +827,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p) + return r; + + r = amdgpu_sync_fence(adev, &p->job->sync, +- fpriv->prt_va->last_pt_update); ++ fpriv->prt_va->last_pt_update, false); + if (r) + return r; + +@@ -841,7 +841,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p) + return r; + + f = bo_va->last_pt_update; +- r = amdgpu_sync_fence(adev, &p->job->sync, f); ++ r = amdgpu_sync_fence(adev, &p->job->sync, f, false); + if (r) + return r; + } +@@ -864,7 +864,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p) + return r; + + f = bo_va->last_pt_update; +- r = amdgpu_sync_fence(adev, &p->job->sync, f); ++ r = amdgpu_sync_fence(adev, &p->job->sync, f, false); + if (r) + return r; + } +@@ -875,7 +875,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p) + if (r) + return r; + +- r = amdgpu_sync_fence(adev, &p->job->sync, vm->last_update); ++ r = amdgpu_sync_fence(adev, &p->job->sync, vm->last_update, false); + if (r) + return r; + +@@ -1079,8 +1079,8 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p, + amdgpu_ctx_put(ctx); + return r; + } else if (fence) { +- r = amdgpu_sync_fence(p->adev, &p->job->dep_sync, +- fence); ++ r = amdgpu_sync_fence(p->adev, &p->job->sync, fence, ++ true); + dma_fence_put(fence); + amdgpu_ctx_put(ctx); + if (r) +@@ -1100,7 +1100,7 @@ static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p, + if (r) + return r; + +- r = amdgpu_sync_fence(p->adev, &p->job->dep_sync, fence); ++ r = amdgpu_sync_fence(p->adev, &p->job->sync, fence, true); + dma_fence_put(fence); + + return r; +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +index b502338..1643897 100644 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +@@ -170,7 +170,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, + } + + if (ring->funcs->emit_pipeline_sync && job && +- ((tmp = amdgpu_sync_get_fence(&job->sched_sync)) || ++ ((tmp = amdgpu_sync_get_fence(&job->sched_sync, NULL)) || + amdgpu_vm_need_pipeline_sync(ring, job))) { + need_pipe_sync = true; + dma_fence_put(tmp); +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +index 4d09b25..a7ec57b 100644 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +@@ -60,7 +60,6 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs, + (*job)->num_ibs = num_ibs; + + amdgpu_sync_create(&(*job)->sync); +- amdgpu_sync_create(&(*job)->dep_sync); + amdgpu_sync_create(&(*job)->sched_sync); + (*job)->vram_lost_counter = atomic_read(&adev->vram_lost_counter); + +@@ -104,7 +103,6 @@ static void amdgpu_job_free_cb(struct amd_sched_job *s_job) + amdgpu_ring_priority_put(job->ring, s_job->s_priority); + dma_fence_put(job->fence); + amdgpu_sync_free(&job->sync); +- amdgpu_sync_free(&job->dep_sync); + amdgpu_sync_free(&job->sched_sync); + kfree(job); + } +@@ -115,7 +113,6 @@ void amdgpu_job_free(struct amdgpu_job *job) + + dma_fence_put(job->fence); + amdgpu_sync_free(&job->sync); +- amdgpu_sync_free(&job->dep_sync); + amdgpu_sync_free(&job->sched_sync); + kfree(job); + } +@@ -149,17 +146,18 @@ static struct dma_fence *amdgpu_job_dependency(struct amd_sched_job *sched_job, + { + struct amdgpu_job *job = to_amdgpu_job(sched_job); + struct amdgpu_vm *vm = job->vm; +- +- struct dma_fence *fence = amdgpu_sync_get_fence(&job->dep_sync); ++ bool explicit = false; + int r; +- +- if (amd_sched_dependency_optimized(fence, s_entity)) { +- r = amdgpu_sync_fence(job->adev, &job->sched_sync, fence); +- if (r) +- DRM_ERROR("Error adding fence to sync (%d)\n", r); ++ struct dma_fence *fence = amdgpu_sync_get_fence(&job->sync, &explicit); ++ ++ if (fence && explicit) { ++ if (amd_sched_dependency_optimized(fence, s_entity)) { ++ r = amdgpu_sync_fence(job->adev, &job->sched_sync, fence, false); ++ if (r) ++ DRM_ERROR("Error adding fence to sync (%d)\n", r); ++ } + } +- if (!fence) +- fence = amdgpu_sync_get_fence(&job->sync); ++ + while (fence == NULL && vm && !job->vm_id) { + struct amdgpu_ring *ring = job->ring; + +@@ -169,7 +167,7 @@ static struct dma_fence *amdgpu_job_dependency(struct amd_sched_job *sched_job, + if (r) + DRM_ERROR("Error getting VM ID (%d)\n", r); + +- fence = amdgpu_sync_get_fence(&job->sync); ++ fence = amdgpu_sync_get_fence(&job->sync, NULL); + } + + return fence; +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c +index a20d4ba..76fae9b 100755 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c +@@ -36,6 +36,7 @@ + struct amdgpu_sync_entry { + struct hlist_node node; + struct dma_fence *fence; ++ bool explicit; + }; + + static struct kmem_cache *amdgpu_sync_slab; +@@ -151,7 +152,7 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f) + * + */ + int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync, +- struct dma_fence *f) ++ struct dma_fence *f, bool explicit) + { + struct amdgpu_sync_entry *e; + +@@ -169,6 +170,8 @@ int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync, + if (!e) + return -ENOMEM; + ++ e->explicit = explicit; ++ + hash_add(sync->fences, &e->node, f->context); + e->fence = dma_fence_get(f); + return 0; +@@ -199,7 +202,7 @@ int amdgpu_sync_resv(struct amdgpu_device *adev, + + /* always sync to the exclusive fence */ + f = reservation_object_get_excl(resv); +- r = amdgpu_sync_fence(adev, sync, f); ++ r = amdgpu_sync_fence(adev, sync, f, false); + + if (explicit_sync) + return r; +@@ -234,7 +237,7 @@ int amdgpu_sync_resv(struct amdgpu_device *adev, + continue; + } + +- r = amdgpu_sync_fence(adev, sync, f); ++ r = amdgpu_sync_fence(adev, sync, f, false); + if (r) + break; + } +@@ -289,10 +292,11 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync, + * amdgpu_sync_get_fence - get the next fence from the sync object + * + * @sync: sync object to use ++ * @explicit: true if the next fence is explicit + * + * Get and removes the next fence from the sync object not signaled yet. + */ +-struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync) ++struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync, bool *explicit) + { + struct amdgpu_sync_entry *e; + struct hlist_node *tmp; +@@ -302,6 +306,8 @@ struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync) + hash_for_each_safe(sync->fences, i, tmp, e, node) { + + f = e->fence; ++ if (explicit) ++ *explicit = e->explicit; + + hash_del(&e->node); + kmem_cache_free(amdgpu_sync_slab, e); +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h +index b764864..6d428da 100755 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h +@@ -41,7 +41,7 @@ struct amdgpu_sync { + + void amdgpu_sync_create(struct amdgpu_sync *sync); + int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync, +- struct dma_fence *f); ++ struct dma_fence *f, bool explicit); + int amdgpu_sync_resv(struct amdgpu_device *adev, + struct amdgpu_sync *sync, + struct reservation_object *resv, +@@ -49,7 +49,7 @@ int amdgpu_sync_resv(struct amdgpu_device *adev, + bool explicit_sync); + struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync, + struct amdgpu_ring *ring); +-struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync); ++struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync, bool *explicit); + int amdgpu_sync_clone(struct amdgpu_device *adev, struct amdgpu_sync *source, + struct amdgpu_sync *clone); + int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr); +diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +index 335d37d..e384ee3c 100644 +--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c ++++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +@@ -497,7 +497,7 @@ static int amdgpu_vm_grab_reserved_vmid_locked(struct amdgpu_vm *vm, + id->pd_gpu_addr = 0; + tmp = amdgpu_sync_peek_fence(&id->active, ring); + if (tmp) { +- r = amdgpu_sync_fence(adev, sync, tmp); ++ r = amdgpu_sync_fence(adev, sync, tmp, false); + return r; + } + } +@@ -505,7 +505,7 @@ static int amdgpu_vm_grab_reserved_vmid_locked(struct amdgpu_vm *vm, + /* Good we can use this VMID. Remember this submission as + * user of the VMID. + */ +- r = amdgpu_sync_fence(ring->adev, &id->active, fence); ++ r = amdgpu_sync_fence(ring->adev, &id->active, fence, false); + if (r) + goto out; + +@@ -592,7 +592,7 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, + } + + +- r = amdgpu_sync_fence(ring->adev, sync, &array->base); ++ r = amdgpu_sync_fence(ring->adev, sync, &array->base, false); + dma_fence_put(&array->base); + if (r) + goto error; +@@ -635,7 +635,7 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, + /* Good we can use this VMID. Remember this submission as + * user of the VMID. + */ +- r = amdgpu_sync_fence(ring->adev, &id->active, fence); ++ r = amdgpu_sync_fence(ring->adev, &id->active, fence, false); + if (r) + goto error; + +@@ -655,7 +655,7 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, + id = idle; + + /* Remember this submission as user of the VMID */ +- r = amdgpu_sync_fence(ring->adev, &id->active, fence); ++ r = amdgpu_sync_fence(ring->adev, &id->active, fence, false); + if (r) + goto error; + +@@ -1666,7 +1666,7 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, + addr = 0; + } + +- r = amdgpu_sync_fence(adev, &job->sync, exclusive); ++ r = amdgpu_sync_fence(adev, &job->sync, exclusive, false); + if (r) + goto error_free; + +-- +2.7.4 + |