Linaro-mm-sig

linaro-mm-sig@lists.linaro.org

15 participants
3025 discussions

[PATCH v3 11/20] drm/etnaviv: Use scheduler dependency handling

by Daniel Vetter

We need to pull the drm_sched_job_init much earlier, but that's very minor surgery. v2: Actually fix up cleanup paths by calling drm_sched_job_init, which I wanted to to in the previous round (and did, for all other drivers). Spotted by Lucas. Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: Lucas Stach <l.stach(a)pengutronix.de> Cc: Russell King <linux+etnaviv(a)armlinux.org.uk> Cc: Christian Gmeiner <christian.gmeiner(a)gmail.com> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: etnaviv(a)lists.freedesktop.org Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/etnaviv/etnaviv_gem.h | 5 +- drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c | 58 +++++++++--------- drivers/gpu/drm/etnaviv/etnaviv_sched.c | 63 +------------------- drivers/gpu/drm/etnaviv/etnaviv_sched.h | 3 +- 4 files changed, 35 insertions(+), 94 deletions(-) diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem.h b/drivers/gpu/drm/etnaviv/etnaviv_gem.h index 98e60df882b6..63688e6e4580 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem.h @@ -80,9 +80,6 @@ struct etnaviv_gem_submit_bo { u64 va; struct etnaviv_gem_object *obj; struct etnaviv_vram_mapping *mapping; - struct dma_fence *excl; - unsigned int nr_shared; - struct dma_fence **shared; }; /* Created per submit-ioctl, to track bo's and cmdstream bufs, etc, @@ -95,7 +92,7 @@ struct etnaviv_gem_submit { struct etnaviv_file_private *ctx; struct etnaviv_gpu *gpu; struct etnaviv_iommu_context *mmu_context, *prev_mmu_context; - struct dma_fence *out_fence, *in_fence; + struct dma_fence *out_fence; int out_fence_id; struct list_head node; /* GPU active submit list */ struct etnaviv_cmdbuf cmdbuf; diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c index 4dd7d9d541c0..5b97ce1299ad 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c @@ -188,16 +188,10 @@ static int submit_fence_sync(struct etnaviv_gem_submit *submit) if (submit->flags & ETNA_SUBMIT_NO_IMPLICIT) continue; - if (bo->flags & ETNA_SUBMIT_BO_WRITE) { - ret = dma_resv_get_fences(robj, &bo->excl, - &bo->nr_shared, - &bo->shared); - if (ret) - return ret; - } else { - bo->excl = dma_resv_get_excl_unlocked(robj); - } - + ret = drm_sched_job_await_implicit(&submit->sched_job, &bo->obj->base, + bo->flags & ETNA_SUBMIT_BO_WRITE); + if (ret) + return ret; } return ret; @@ -403,8 +397,6 @@ static void submit_cleanup(struct kref *kref) wake_up_all(&submit->gpu->fence_event); - if (submit->in_fence) - dma_fence_put(submit->in_fence); if (submit->out_fence) { /* first remove from IDR, so fence can not be found anymore */ mutex_lock(&submit->gpu->fence_lock); @@ -529,7 +521,7 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, ret = etnaviv_cmdbuf_init(priv->cmdbuf_suballoc, &submit->cmdbuf, ALIGN(args->stream_size, 8) + 8); if (ret) - goto err_submit_objects; + goto err_submit_put; submit->ctx = file->driver_priv; etnaviv_iommu_context_get(submit->ctx->mmu); @@ -537,51 +529,61 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, submit->exec_state = args->exec_state; submit->flags = args->flags; + ret = drm_sched_job_init(&submit->sched_job, + &ctx->sched_entity[args->pipe], + submit->ctx); + if (ret) + goto err_submit_put; + ret = submit_lookup_objects(submit, file, bos, args->nr_bos); if (ret) - goto err_submit_objects; + goto err_submit_job; if ((priv->mmu_global->version != ETNAVIV_IOMMU_V2) && !etnaviv_cmd_validate_one(gpu, stream, args->stream_size / 4, relocs, args->nr_relocs)) { ret = -EINVAL; - goto err_submit_objects; + goto err_submit_job; } if (args->flags & ETNA_SUBMIT_FENCE_FD_IN) { - submit->in_fence = sync_file_get_fence(args->fence_fd); - if (!submit->in_fence) { + struct dma_fence *in_fence = sync_file_get_fence(args->fence_fd); + if (!in_fence) { ret = -EINVAL; - goto err_submit_objects; + goto err_submit_job; } + + ret = drm_sched_job_await_fence(&submit->sched_job, in_fence); + if (ret) + goto err_submit_job; } ret = submit_pin_objects(submit); if (ret) - goto err_submit_objects; + goto err_submit_job; ret = submit_reloc(submit, stream, args->stream_size / 4, relocs, args->nr_relocs); if (ret) - goto err_submit_objects; + goto err_submit_job; ret = submit_perfmon_validate(submit, args->exec_state, pmrs); if (ret) - goto err_submit_objects; + goto err_submit_job; memcpy(submit->cmdbuf.vaddr, stream, args->stream_size); ret = submit_lock_objects(submit, &ticket); if (ret) - goto err_submit_objects; + goto err_submit_job; ret = submit_fence_sync(submit); if (ret) - goto err_submit_objects; + goto err_submit_job; - ret = etnaviv_sched_push_job(&ctx->sched_entity[args->pipe], submit); + ret = etnaviv_sched_push_job(submit); if (ret) - goto err_submit_objects; + goto err_submit_job; submit_attach_object_fences(submit); @@ -595,7 +597,7 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, sync_file = sync_file_create(submit->out_fence); if (!sync_file) { ret = -ENOMEM; - goto err_submit_objects; + goto err_submit_job; } fd_install(out_fence_fd, sync_file->file); } @@ -603,7 +605,9 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, args->fence_fd = out_fence_fd; args->fence = submit->out_fence_id; -err_submit_objects: +err_submit_job: + drm_sched_job_cleanup(&submit->sched_job); +err_submit_put: etnaviv_submit_put(submit); err_submit_ww_acquire: diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c index 180bb633d5c5..2bbbd6ccc95e 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c @@ -17,58 +17,6 @@ module_param_named(job_hang_limit, etnaviv_job_hang_limit, int , 0444); static int etnaviv_hw_jobs_limit = 4; module_param_named(hw_job_limit, etnaviv_hw_jobs_limit, int , 0444); -static struct dma_fence * -etnaviv_sched_dependency(struct drm_sched_job *sched_job, - struct drm_sched_entity *entity) -{ - struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job); - struct dma_fence *fence; - int i; - - if (unlikely(submit->in_fence)) { - fence = submit->in_fence; - submit->in_fence = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - - for (i = 0; i < submit->nr_bos; i++) { - struct etnaviv_gem_submit_bo *bo = &submit->bos[i]; - int j; - - if (bo->excl) { - fence = bo->excl; - bo->excl = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - - for (j = 0; j < bo->nr_shared; j++) { - if (!bo->shared[j]) - continue; - - fence = bo->shared[j]; - bo->shared[j] = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - kfree(bo->shared); - bo->nr_shared = 0; - bo->shared = NULL; - } - - return NULL; -} - static struct dma_fence *etnaviv_sched_run_job(struct drm_sched_job *sched_job) { struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job); @@ -140,29 +88,22 @@ static void etnaviv_sched_free_job(struct drm_sched_job *sched_job) } static const struct drm_sched_backend_ops etnaviv_sched_ops = { - .dependency = etnaviv_sched_dependency, .run_job = etnaviv_sched_run_job, .timedout_job = etnaviv_sched_timedout_job, .free_job = etnaviv_sched_free_job, }; -int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, - struct etnaviv_gem_submit *submit) +int etnaviv_sched_push_job(struct etnaviv_gem_submit *submit) { int ret = 0; /* * Hold the fence lock across the whole operation to avoid jobs being * pushed out of order with regard to their sched fence seqnos as - * allocated in drm_sched_job_init. + * allocated in drm_sched_job_arm. */ mutex_lock(&submit->gpu->fence_lock); - ret = drm_sched_job_init(&submit->sched_job, sched_entity, - submit->ctx); - if (ret) - goto out_unlock; - drm_sched_job_arm(&submit->sched_job); submit->out_fence = dma_fence_get(&submit->sched_job.s_fence->finished); diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.h b/drivers/gpu/drm/etnaviv/etnaviv_sched.h index c0a6796e22c9..baebfa069afc 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.h @@ -18,7 +18,6 @@ struct etnaviv_gem_submit *to_etnaviv_submit(struct drm_sched_job *sched_job) int etnaviv_sched_init(struct etnaviv_gpu *gpu); void etnaviv_sched_fini(struct etnaviv_gpu *gpu); -int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, - struct etnaviv_gem_submit *submit); +int etnaviv_sched_push_job(struct etnaviv_gem_submit *submit); #endif /* __ETNAVIV_SCHED_H__ */ -- 2.32.0

4 years, 2 months

[PATCH v3 08/20] drm/lima: use scheduler dependency tracking

by Daniel Vetter

Nothing special going on here. Aside reviewing the code, it seems like drm_sched_job_arm() should be moved into lima_sched_context_queue_task and put under some mutex together with drm_sched_push_job(). See the kerneldoc for drm_sched_push_job(). Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: Qiang Yu <yuq825(a)gmail.com> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: lima(a)lists.freedesktop.org Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/lima/lima_gem.c | 4 ++-- drivers/gpu/drm/lima/lima_sched.c | 21 --------------------- drivers/gpu/drm/lima/lima_sched.h | 3 --- 3 files changed, 2 insertions(+), 26 deletions(-) diff --git a/drivers/gpu/drm/lima/lima_gem.c b/drivers/gpu/drm/lima/lima_gem.c index c528f40981bb..e54a88d5037a 100644 --- a/drivers/gpu/drm/lima/lima_gem.c +++ b/drivers/gpu/drm/lima/lima_gem.c @@ -267,7 +267,7 @@ static int lima_gem_sync_bo(struct lima_sched_task *task, struct lima_bo *bo, if (explicit) return 0; - return drm_gem_fence_array_add_implicit(&task->deps, &bo->base.base, write); + return drm_sched_job_await_implicit(&task->base, &bo->base.base, write); } static int lima_gem_add_deps(struct drm_file *file, struct lima_submit *submit) @@ -285,7 +285,7 @@ static int lima_gem_add_deps(struct drm_file *file, struct lima_submit *submit) if (err) return err; - err = drm_gem_fence_array_add(&submit->task->deps, fence); + err = drm_sched_job_await_fence(&submit->task->base, fence); if (err) { dma_fence_put(fence); return err; diff --git a/drivers/gpu/drm/lima/lima_sched.c b/drivers/gpu/drm/lima/lima_sched.c index e968b5a8f0b0..99d5f6f1a882 100644 --- a/drivers/gpu/drm/lima/lima_sched.c +++ b/drivers/gpu/drm/lima/lima_sched.c @@ -134,24 +134,15 @@ int lima_sched_task_init(struct lima_sched_task *task, task->num_bos = num_bos; task->vm = lima_vm_get(vm); - xa_init_flags(&task->deps, XA_FLAGS_ALLOC); - return 0; } void lima_sched_task_fini(struct lima_sched_task *task) { - struct dma_fence *fence; - unsigned long index; int i; drm_sched_job_cleanup(&task->base); - xa_for_each(&task->deps, index, fence) { - dma_fence_put(fence); - } - xa_destroy(&task->deps); - if (task->bos) { for (i = 0; i < task->num_bos; i++) drm_gem_object_put(&task->bos[i]->base.base); @@ -186,17 +177,6 @@ struct dma_fence *lima_sched_context_queue_task(struct lima_sched_task *task) return fence; } -static struct dma_fence *lima_sched_dependency(struct drm_sched_job *job, - struct drm_sched_entity *entity) -{ - struct lima_sched_task *task = to_lima_task(job); - - if (!xa_empty(&task->deps)) - return xa_erase(&task->deps, task->last_dep++); - - return NULL; -} - static int lima_pm_busy(struct lima_device *ldev) { int ret; @@ -472,7 +452,6 @@ static void lima_sched_free_job(struct drm_sched_job *job) } static const struct drm_sched_backend_ops lima_sched_ops = { - .dependency = lima_sched_dependency, .run_job = lima_sched_run_job, .timedout_job = lima_sched_timedout_job, .free_job = lima_sched_free_job, diff --git a/drivers/gpu/drm/lima/lima_sched.h b/drivers/gpu/drm/lima/lima_sched.h index ac70006b0e26..6a11764d87b3 100644 --- a/drivers/gpu/drm/lima/lima_sched.h +++ b/drivers/gpu/drm/lima/lima_sched.h @@ -23,9 +23,6 @@ struct lima_sched_task { struct lima_vm *vm; void *frame; - struct xarray deps; - unsigned long last_dep; - struct lima_bo **bos; int num_bos; -- 2.32.0

4 years, 2 months

[PATCH v3 07/20] drm/panfrost: use scheduler dependency tracking

by Daniel Vetter

Just deletes some code that's now more shared. Note that thanks to the split into drm_sched_job_init/arm we can now easily pull the _init() part from under the submission lock way ahead where we're adding the sync file in-fences as dependencies. v2: Correctly clean up the partially set up job, now that job_init() and job_arm() are apart (Emma). Reviewed-by: Steven Price <steven.price(a)arm.com> (v1) Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: Rob Herring <robh(a)kernel.org> Cc: Tomeu Vizoso <tomeu.vizoso(a)collabora.com> Cc: Steven Price <steven.price(a)arm.com> Cc: Alyssa Rosenzweig <alyssa.rosenzweig(a)collabora.com> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/panfrost/panfrost_drv.c | 16 ++++++++--- drivers/gpu/drm/panfrost/panfrost_job.c | 37 +++---------------------- drivers/gpu/drm/panfrost/panfrost_job.h | 5 +--- 3 files changed, 17 insertions(+), 41 deletions(-) diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c index 1ffaef5ec5ff..9f53bea07d61 100644 --- a/drivers/gpu/drm/panfrost/panfrost_drv.c +++ b/drivers/gpu/drm/panfrost/panfrost_drv.c @@ -218,7 +218,7 @@ panfrost_copy_in_sync(struct drm_device *dev, if (ret) goto fail; - ret = drm_gem_fence_array_add(&job->deps, fence); + ret = drm_sched_job_await_fence(&job->base, fence); if (ret) goto fail; @@ -236,7 +236,7 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, struct drm_panfrost_submit *args = data; struct drm_syncobj *sync_out = NULL; struct panfrost_job *job; - int ret = 0; + int ret = 0, slot; if (!args->jc) return -EINVAL; @@ -258,14 +258,20 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, kref_init(&job->refcount); - xa_init_flags(&job->deps, XA_FLAGS_ALLOC); - job->pfdev = pfdev; job->jc = args->jc; job->requirements = args->requirements; job->flush_id = panfrost_gpu_get_latest_flush_id(pfdev); job->file_priv = file->driver_priv; + slot = panfrost_job_get_slot(job); + + ret = drm_sched_job_init(&job->base, + &job->file_priv->sched_entity[slot], + NULL); + if (ret) + goto fail_job_put; + ret = panfrost_copy_in_sync(dev, file, args, job); if (ret) goto fail_job; @@ -283,6 +289,8 @@ static int panfrost_ioctl_submit(struct drm_device *dev, void *data, drm_syncobj_replace_fence(sync_out, job->render_done_fence); fail_job: + drm_sched_job_cleanup(&job->base); +fail_job_put: panfrost_job_put(job); fail_out_sync: if (sync_out) diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c index 4bc962763e1f..86c843d8822e 100644 --- a/drivers/gpu/drm/panfrost/panfrost_job.c +++ b/drivers/gpu/drm/panfrost/panfrost_job.c @@ -102,7 +102,7 @@ static struct dma_fence *panfrost_fence_create(struct panfrost_device *pfdev, in return &fence->base; } -static int panfrost_job_get_slot(struct panfrost_job *job) +int panfrost_job_get_slot(struct panfrost_job *job) { /* JS0: fragment jobs. * JS1: vertex/tiler jobs @@ -242,13 +242,13 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js) static int panfrost_acquire_object_fences(struct drm_gem_object **bos, int bo_count, - struct xarray *deps) + struct drm_sched_job *job) { int i, ret; for (i = 0; i < bo_count; i++) { /* panfrost always uses write mode in its current uapi */ - ret = drm_gem_fence_array_add_implicit(deps, bos[i], true); + ret = drm_sched_job_await_implicit(job, bos[i], true); if (ret) return ret; } @@ -269,31 +269,21 @@ static void panfrost_attach_object_fences(struct drm_gem_object **bos, int panfrost_job_push(struct panfrost_job *job) { struct panfrost_device *pfdev = job->pfdev; - int slot = panfrost_job_get_slot(job); - struct drm_sched_entity *entity = &job->file_priv->sched_entity[slot]; struct ww_acquire_ctx acquire_ctx; int ret = 0; - ret = drm_gem_lock_reservations(job->bos, job->bo_count, &acquire_ctx); if (ret) return ret; mutex_lock(&pfdev->sched_lock); - - ret = drm_sched_job_init(&job->base, entity, NULL); - if (ret) { - mutex_unlock(&pfdev->sched_lock); - goto unlock; - } - drm_sched_job_arm(&job->base); job->render_done_fence = dma_fence_get(&job->base.s_fence->finished); ret = panfrost_acquire_object_fences(job->bos, job->bo_count, - &job->deps); + &job->base); if (ret) { mutex_unlock(&pfdev->sched_lock); goto unlock; @@ -318,15 +308,8 @@ static void panfrost_job_cleanup(struct kref *ref) { struct panfrost_job *job = container_of(ref, struct panfrost_job, refcount); - struct dma_fence *fence; - unsigned long index; unsigned int i; - xa_for_each(&job->deps, index, fence) { - dma_fence_put(fence); - } - xa_destroy(&job->deps); - dma_fence_put(job->done_fence); dma_fence_put(job->render_done_fence); @@ -365,17 +348,6 @@ static void panfrost_job_free(struct drm_sched_job *sched_job) panfrost_job_put(job); } -static struct dma_fence *panfrost_job_dependency(struct drm_sched_job *sched_job, - struct drm_sched_entity *s_entity) -{ - struct panfrost_job *job = to_panfrost_job(sched_job); - - if (!xa_empty(&job->deps)) - return xa_erase(&job->deps, job->last_dep++); - - return NULL; -} - static struct dma_fence *panfrost_job_run(struct drm_sched_job *sched_job) { struct panfrost_job *job = to_panfrost_job(sched_job); @@ -765,7 +737,6 @@ static void panfrost_reset_work(struct work_struct *work) } static const struct drm_sched_backend_ops panfrost_sched_ops = { - .dependency = panfrost_job_dependency, .run_job = panfrost_job_run, .timedout_job = panfrost_job_timedout, .free_job = panfrost_job_free diff --git a/drivers/gpu/drm/panfrost/panfrost_job.h b/drivers/gpu/drm/panfrost/panfrost_job.h index 82306a03b57e..77e6d0e6f612 100644 --- a/drivers/gpu/drm/panfrost/panfrost_job.h +++ b/drivers/gpu/drm/panfrost/panfrost_job.h @@ -19,10 +19,6 @@ struct panfrost_job { struct panfrost_device *pfdev; struct panfrost_file_priv *file_priv; - /* Contains both explicit and implicit fences */ - struct xarray deps; - unsigned long last_dep; - /* Fence to be signaled by IRQ handler when the job is complete. */ struct dma_fence *done_fence; @@ -42,6 +38,7 @@ int panfrost_job_init(struct panfrost_device *pfdev); void panfrost_job_fini(struct panfrost_device *pfdev); int panfrost_job_open(struct panfrost_file_priv *panfrost_priv); void panfrost_job_close(struct panfrost_file_priv *panfrost_priv); +int panfrost_job_get_slot(struct panfrost_job *job); int panfrost_job_push(struct panfrost_job *job); void panfrost_job_put(struct panfrost_job *job); void panfrost_job_enable_interrupts(struct panfrost_device *pfdev); -- 2.32.0

4 years, 2 months

[PATCH v3 04/20] drm/sched: Add dependency tracking

by Daniel Vetter

Instead of just a callback we can just glue in the gem helpers that panfrost, v3d and lima currently use. There's really not that many ways to skin this cat. On the naming bikeshed: The idea for using _await_ to denote adding dependencies to a job comes from i915, where that's used quite extensively all over the place, in lots of datastructures. v2/3: Rebased. Reviewed-by: Steven Price <steven.price(a)arm.com> (v1) Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: David Airlie <airlied(a)linux.ie> Cc: Daniel Vetter <daniel(a)ffwll.ch> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: Andrey Grodzovsky <andrey.grodzovsky(a)amd.com> Cc: Lee Jones <lee.jones(a)linaro.org> Cc: Nirmoy Das <nirmoy.aiemd(a)gmail.com> Cc: Boris Brezillon <boris.brezillon(a)collabora.com> Cc: Luben Tuikov <luben.tuikov(a)amd.com> Cc: Alex Deucher <alexander.deucher(a)amd.com> Cc: Jack Zhang <Jack.Zhang1(a)amd.com> Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/scheduler/sched_entity.c | 18 +++- drivers/gpu/drm/scheduler/sched_main.c | 103 +++++++++++++++++++++++ include/drm/gpu_scheduler.h | 31 ++++++- 3 files changed, 146 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c index 4e1124ed80e0..c7e6d29c9a33 100644 --- a/drivers/gpu/drm/scheduler/sched_entity.c +++ b/drivers/gpu/drm/scheduler/sched_entity.c @@ -211,6 +211,19 @@ static void drm_sched_entity_kill_jobs_cb(struct dma_fence *f, job->sched->ops->free_job(job); } +static struct dma_fence * +drm_sched_job_dependency(struct drm_sched_job *job, + struct drm_sched_entity *entity) +{ + if (!xa_empty(&job->dependencies)) + return xa_erase(&job->dependencies, job->last_dependency++); + + if (job->sched->ops->dependency) + return job->sched->ops->dependency(job, entity); + + return NULL; +} + /** * drm_sched_entity_kill_jobs - Make sure all remaining jobs are killed * @@ -229,7 +242,7 @@ static void drm_sched_entity_kill_jobs(struct drm_sched_entity *entity) struct drm_sched_fence *s_fence = job->s_fence; /* Wait for all dependencies to avoid data corruptions */ - while ((f = job->sched->ops->dependency(job, entity))) + while ((f = drm_sched_job_dependency(job, entity))) dma_fence_wait(f, false); drm_sched_fence_scheduled(s_fence); @@ -419,7 +432,6 @@ static bool drm_sched_entity_add_dependency_cb(struct drm_sched_entity *entity) */ struct drm_sched_job *drm_sched_entity_pop_job(struct drm_sched_entity *entity) { - struct drm_gpu_scheduler *sched = entity->rq->sched; struct drm_sched_job *sched_job; sched_job = to_drm_sched_job(spsc_queue_peek(&entity->job_queue)); @@ -427,7 +439,7 @@ struct drm_sched_job *drm_sched_entity_pop_job(struct drm_sched_entity *entity) return NULL; while ((entity->dependency = - sched->ops->dependency(sched_job, entity))) { + drm_sched_job_dependency(sched_job, entity))) { trace_drm_sched_job_wait_dep(sched_job, entity->dependency); if (drm_sched_entity_add_dependency_cb(entity)) diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 7e94754eb34c..ad62f1d2991c 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -594,6 +594,8 @@ int drm_sched_job_init(struct drm_sched_job *job, INIT_LIST_HEAD(&job->list); + xa_init_flags(&job->dependencies, XA_FLAGS_ALLOC); + return 0; } EXPORT_SYMBOL(drm_sched_job_init); @@ -631,6 +633,98 @@ void drm_sched_job_arm(struct drm_sched_job *job) } EXPORT_SYMBOL(drm_sched_job_arm); +/** + * drm_sched_job_await_fence - adds the fence as a job dependency + * @job: scheduler job to add the dependencies to + * @fence: the dma_fence to add to the list of dependencies. + * + * Note that @fence is consumed in both the success and error cases. + * + * Returns: + * 0 on success, or an error on failing to expand the array. + */ +int drm_sched_job_await_fence(struct drm_sched_job *job, + struct dma_fence *fence) +{ + struct dma_fence *entry; + unsigned long index; + u32 id = 0; + int ret; + + if (!fence) + return 0; + + /* Deduplicate if we already depend on a fence from the same context. + * This lets the size of the array of deps scale with the number of + * engines involved, rather than the number of BOs. + */ + xa_for_each(&job->dependencies, index, entry) { + if (entry->context != fence->context) + continue; + + if (dma_fence_is_later(fence, entry)) { + dma_fence_put(entry); + xa_store(&job->dependencies, index, fence, GFP_KERNEL); + } else { + dma_fence_put(fence); + } + return 0; + } + + ret = xa_alloc(&job->dependencies, &id, fence, xa_limit_32b, GFP_KERNEL); + if (ret != 0) + dma_fence_put(fence); + + return ret; +} +EXPORT_SYMBOL(drm_sched_job_await_fence); + +/** + * drm_sched_job_await_implicit - adds implicit dependencies as job dependencies + * @job: scheduler job to add the dependencies to + * @obj: the gem object to add new dependencies from. + * @write: whether the job might write the object (so we need to depend on + * shared fences in the reservation object). + * + * This should be called after drm_gem_lock_reservations() on your array of + * GEM objects used in the job but before updating the reservations with your + * own fences. + * + * Returns: + * 0 on success, or an error on failing to expand the array. + */ +int drm_sched_job_await_implicit(struct drm_sched_job *job, + struct drm_gem_object *obj, + bool write) +{ + int ret; + struct dma_fence **fences; + unsigned int i, fence_count; + + if (!write) { + struct dma_fence *fence = dma_resv_get_excl_unlocked(obj->resv); + + return drm_sched_job_await_fence(job, fence); + } + + ret = dma_resv_get_fences(obj->resv, NULL, &fence_count, &fences); + if (ret || !fence_count) + return ret; + + for (i = 0; i < fence_count; i++) { + ret = drm_sched_job_await_fence(job, fences[i]); + if (ret) + break; + } + + for (; i < fence_count; i++) + dma_fence_put(fences[i]); + kfree(fences); + return ret; +} +EXPORT_SYMBOL(drm_sched_job_await_implicit); + + /** * drm_sched_job_cleanup - clean up scheduler job resources * @job: scheduler job to clean up @@ -646,6 +740,9 @@ EXPORT_SYMBOL(drm_sched_job_arm); */ void drm_sched_job_cleanup(struct drm_sched_job *job) { + struct dma_fence *fence; + unsigned long index; + if (kref_read(&job->s_fence->finished.refcount)) { /* drm_sched_job_arm() has been called */ dma_fence_put(&job->s_fence->finished); @@ -655,6 +752,12 @@ void drm_sched_job_cleanup(struct drm_sched_job *job) } job->s_fence = NULL; + + xa_for_each(&job->dependencies, index, fence) { + dma_fence_put(fence); + } + xa_destroy(&job->dependencies); + } EXPORT_SYMBOL(drm_sched_job_cleanup); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 83afc3aa8e2f..74fb321dbc44 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -27,9 +27,12 @@ #include <drm/spsc_queue.h> #include <linux/dma-fence.h> #include <linux/completion.h> +#include <linux/xarray.h> #define MAX_WAIT_SCHED_ENTITY_Q_EMPTY msecs_to_jiffies(1000) +struct drm_gem_object; + struct drm_gpu_scheduler; struct drm_sched_rq; @@ -198,6 +201,16 @@ struct drm_sched_job { enum drm_sched_priority s_priority; struct drm_sched_entity *entity; struct dma_fence_cb cb; + /** + * @dependencies: + * + * Contains the dependencies as struct dma_fence for this job, see + * drm_sched_job_await_fence() and drm_sched_job_await_implicit(). + */ + struct xarray dependencies; + + /** @last_dependency: tracks @dependencies as they signal */ + unsigned long last_dependency; }; static inline bool drm_sched_invalidate_job(struct drm_sched_job *s_job, @@ -220,9 +233,14 @@ enum drm_gpu_sched_stat { */ struct drm_sched_backend_ops { /** - * @dependency: Called when the scheduler is considering scheduling - * this job next, to get another struct dma_fence for this job to - * block on. Once it returns NULL, run_job() may be called. + * @dependency: + * + * Called when the scheduler is considering scheduling this job next, to + * get another struct dma_fence for this job to block on. Once it + * returns NULL, run_job() may be called. + * + * If a driver exclusively uses drm_sched_job_await_fence() and + * drm_sched_job_await_implicit() this can be ommitted and left as NULL. */ struct dma_fence *(*dependency)(struct drm_sched_job *sched_job, struct drm_sched_entity *s_entity); @@ -349,6 +367,13 @@ int drm_sched_job_init(struct drm_sched_job *job, struct drm_sched_entity *entity, void *owner); void drm_sched_job_arm(struct drm_sched_job *job); +int drm_sched_job_await_fence(struct drm_sched_job *job, + struct dma_fence *fence); +int drm_sched_job_await_implicit(struct drm_sched_job *job, + struct drm_gem_object *obj, + bool write); + + void drm_sched_entity_modify_sched(struct drm_sched_entity *entity, struct drm_gpu_scheduler **sched_list, unsigned int num_sched_list); -- 2.32.0

4 years, 2 months

Re: [Linaro-mm-sig] [PATCH v4 0/2] Add p2p via dmabuf to habanalabs

by Daniel Vetter

On Mon, Jul 05, 2021 at 04:03:12PM +0300, Oded Gabbay wrote: > Hi, > I'm sending v4 of this patch-set following the long email thread. > I want to thank Jason for reviewing v3 and pointing out the errors, saving > us time later to debug it :) > > I consulted with Christian on how to fix patch 2 (the implementation) and > at the end of the day I shamelessly copied the relevant content from > amdgpu_vram_mgr_alloc_sgt() and amdgpu_dma_buf_attach(), regarding the > usage of dma_map_resource() and pci_p2pdma_distance_many(), respectively. > > I also made a few improvements after looking at the relevant code in amdgpu. > The details are in the changelog of patch 2. > > I took the time to write an import code into the driver, allowing me to > check real P2P with two Gaudi devices, one as exporter and the other as > importer. I'm not going to include the import code in the product, it was > just for testing purposes (although I can share it if anyone wants). > > I run it on a bare-metal environment with IOMMU enabled, on a sky-lake CPU > with a white-listed PCIe bridge (to make the pci_p2pdma_distance_many happy). > > Greg, I hope this will be good enough for you to merge this code. So we're officially going to use dri-devel for technical details review and then Greg for merging so we don't have to deal with other merge criteria dri-devel folks have? I don't expect anything less by now, but it does make the original claim that drivers/misc will not step all over accelerators folks a complete farce under the totally-not-a-gpu banner. This essentially means that for any other accelerator stack that doesn't fit the dri-devel merge criteria, even if it's acting like a gpu and uses other gpu driver stuff, you can just send it to Greg and it's good to go. There's quite a lot of these floating around actually (and many do have semi-open runtimes, like habanalabs have now too, just not open enough to be actually useful). It's going to be absolutely lovely having to explain to these companies in background chats why habanalabs gets away with their stack and they don't. Or maybe we should just merge them all and give up on the idea of having open cross-vendor driver stacks for these accelerators. Thanks, Daniel > > Thanks, > Oded > > Oded Gabbay (1): > habanalabs: define uAPI to export FD for DMA-BUF > > Tomer Tayar (1): > habanalabs: add support for dma-buf exporter > > drivers/misc/habanalabs/Kconfig | 1 + > drivers/misc/habanalabs/common/habanalabs.h | 26 ++ > drivers/misc/habanalabs/common/memory.c | 480 +++++++++++++++++++- > drivers/misc/habanalabs/gaudi/gaudi.c | 1 + > drivers/misc/habanalabs/goya/goya.c | 1 + > include/uapi/misc/habanalabs.h | 28 +- > 6 files changed, 532 insertions(+), 5 deletions(-) > > -- > 2.25.1 > -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch

4 years, 2 months

[PATCH v2 08/11] drm/etnaviv: Use scheduler dependency handling

by Daniel Vetter

We need to pull the drm_sched_job_init much earlier, but that's very minor surgery. Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: Lucas Stach <l.stach(a)pengutronix.de> Cc: Russell King <linux+etnaviv(a)armlinux.org.uk> Cc: Christian Gmeiner <christian.gmeiner(a)gmail.com> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: etnaviv(a)lists.freedesktop.org Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/etnaviv/etnaviv_gem.h | 5 +- drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c | 32 +++++----- drivers/gpu/drm/etnaviv/etnaviv_sched.c | 61 +------------------- drivers/gpu/drm/etnaviv/etnaviv_sched.h | 3 +- 4 files changed, 20 insertions(+), 81 deletions(-) diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem.h b/drivers/gpu/drm/etnaviv/etnaviv_gem.h index 98e60df882b6..63688e6e4580 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem.h @@ -80,9 +80,6 @@ struct etnaviv_gem_submit_bo { u64 va; struct etnaviv_gem_object *obj; struct etnaviv_vram_mapping *mapping; - struct dma_fence *excl; - unsigned int nr_shared; - struct dma_fence **shared; }; /* Created per submit-ioctl, to track bo's and cmdstream bufs, etc, @@ -95,7 +92,7 @@ struct etnaviv_gem_submit { struct etnaviv_file_private *ctx; struct etnaviv_gpu *gpu; struct etnaviv_iommu_context *mmu_context, *prev_mmu_context; - struct dma_fence *out_fence, *in_fence; + struct dma_fence *out_fence; int out_fence_id; struct list_head node; /* GPU active submit list */ struct etnaviv_cmdbuf cmdbuf; diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c index 4dd7d9d541c0..92478a50a580 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c @@ -188,16 +188,10 @@ static int submit_fence_sync(struct etnaviv_gem_submit *submit) if (submit->flags & ETNA_SUBMIT_NO_IMPLICIT) continue; - if (bo->flags & ETNA_SUBMIT_BO_WRITE) { - ret = dma_resv_get_fences(robj, &bo->excl, - &bo->nr_shared, - &bo->shared); - if (ret) - return ret; - } else { - bo->excl = dma_resv_get_excl_unlocked(robj); - } - + ret = drm_sched_job_await_implicit(&submit->sched_job, &bo->obj->base, + bo->flags & ETNA_SUBMIT_BO_WRITE); + if (ret) + return ret; } return ret; @@ -403,8 +397,6 @@ static void submit_cleanup(struct kref *kref) wake_up_all(&submit->gpu->fence_event); - if (submit->in_fence) - dma_fence_put(submit->in_fence); if (submit->out_fence) { /* first remove from IDR, so fence can not be found anymore */ mutex_lock(&submit->gpu->fence_lock); @@ -537,6 +529,12 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, submit->exec_state = args->exec_state; submit->flags = args->flags; + ret = drm_sched_job_init(&submit->sched_job, + &ctx->sched_entity[args->pipe], + submit->ctx); + if (ret) + goto err_submit_objects; + ret = submit_lookup_objects(submit, file, bos, args->nr_bos); if (ret) goto err_submit_objects; @@ -549,11 +547,15 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, } if (args->flags & ETNA_SUBMIT_FENCE_FD_IN) { - submit->in_fence = sync_file_get_fence(args->fence_fd); - if (!submit->in_fence) { + struct dma_fence *in_fence = sync_file_get_fence(args->fence_fd); + if (!in_fence) { ret = -EINVAL; goto err_submit_objects; } + + ret = drm_sched_job_await_fence(&submit->sched_job, in_fence); + if (ret) + goto err_submit_objects; } ret = submit_pin_objects(submit); @@ -579,7 +581,7 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, if (ret) goto err_submit_objects; - ret = etnaviv_sched_push_job(&ctx->sched_entity[args->pipe], submit); + ret = etnaviv_sched_push_job(submit); if (ret) goto err_submit_objects; diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c index 180bb633d5c5..c98d67320be3 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c @@ -17,58 +17,6 @@ module_param_named(job_hang_limit, etnaviv_job_hang_limit, int , 0444); static int etnaviv_hw_jobs_limit = 4; module_param_named(hw_job_limit, etnaviv_hw_jobs_limit, int , 0444); -static struct dma_fence * -etnaviv_sched_dependency(struct drm_sched_job *sched_job, - struct drm_sched_entity *entity) -{ - struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job); - struct dma_fence *fence; - int i; - - if (unlikely(submit->in_fence)) { - fence = submit->in_fence; - submit->in_fence = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - - for (i = 0; i < submit->nr_bos; i++) { - struct etnaviv_gem_submit_bo *bo = &submit->bos[i]; - int j; - - if (bo->excl) { - fence = bo->excl; - bo->excl = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - - for (j = 0; j < bo->nr_shared; j++) { - if (!bo->shared[j]) - continue; - - fence = bo->shared[j]; - bo->shared[j] = NULL; - - if (!dma_fence_is_signaled(fence)) - return fence; - - dma_fence_put(fence); - } - kfree(bo->shared); - bo->nr_shared = 0; - bo->shared = NULL; - } - - return NULL; -} - static struct dma_fence *etnaviv_sched_run_job(struct drm_sched_job *sched_job) { struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job); @@ -140,14 +88,12 @@ static void etnaviv_sched_free_job(struct drm_sched_job *sched_job) } static const struct drm_sched_backend_ops etnaviv_sched_ops = { - .dependency = etnaviv_sched_dependency, .run_job = etnaviv_sched_run_job, .timedout_job = etnaviv_sched_timedout_job, .free_job = etnaviv_sched_free_job, }; -int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, - struct etnaviv_gem_submit *submit) +int etnaviv_sched_push_job(struct etnaviv_gem_submit *submit) { int ret = 0; @@ -158,11 +104,6 @@ int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, */ mutex_lock(&submit->gpu->fence_lock); - ret = drm_sched_job_init(&submit->sched_job, sched_entity, - submit->ctx); - if (ret) - goto out_unlock; - drm_sched_job_arm(&submit->sched_job); submit->out_fence = dma_fence_get(&submit->sched_job.s_fence->finished); diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.h b/drivers/gpu/drm/etnaviv/etnaviv_sched.h index c0a6796e22c9..baebfa069afc 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.h @@ -18,7 +18,6 @@ struct etnaviv_gem_submit *to_etnaviv_submit(struct drm_sched_job *sched_job) int etnaviv_sched_init(struct etnaviv_gpu *gpu); void etnaviv_sched_fini(struct etnaviv_gpu *gpu); -int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, - struct etnaviv_gem_submit *submit); +int etnaviv_sched_push_job(struct etnaviv_gem_submit *submit); #endif /* __ETNAVIV_SCHED_H__ */ -- 2.32.0.rc2

4 years, 2 months

[PATCH v2 02/11] drm/sched: Add dependency tracking

by Daniel Vetter

Instead of just a callback we can just glue in the gem helpers that panfrost, v3d and lima currently use. There's really not that many ways to skin this cat. On the naming bikeshed: The idea for using _await_ to denote adding dependencies to a job comes from i915, where that's used quite extensively all over the place, in lots of datastructures. v2: Rebased. Reviewed-by: Steven Price <steven.price(a)arm.com> (v1) Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: David Airlie <airlied(a)linux.ie> Cc: Daniel Vetter <daniel(a)ffwll.ch> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: Andrey Grodzovsky <andrey.grodzovsky(a)amd.com> Cc: Lee Jones <lee.jones(a)linaro.org> Cc: Nirmoy Das <nirmoy.aiemd(a)gmail.com> Cc: Boris Brezillon <boris.brezillon(a)collabora.com> Cc: Luben Tuikov <luben.tuikov(a)amd.com> Cc: Alex Deucher <alexander.deucher(a)amd.com> Cc: Jack Zhang <Jack.Zhang1(a)amd.com> Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/gpu/drm/scheduler/sched_entity.c | 18 +++- drivers/gpu/drm/scheduler/sched_main.c | 103 +++++++++++++++++++++++ include/drm/gpu_scheduler.h | 31 ++++++- 3 files changed, 146 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c index f7347c284886..b6f72fafd504 100644 --- a/drivers/gpu/drm/scheduler/sched_entity.c +++ b/drivers/gpu/drm/scheduler/sched_entity.c @@ -211,6 +211,19 @@ static void drm_sched_entity_kill_jobs_cb(struct dma_fence *f, job->sched->ops->free_job(job); } +static struct dma_fence * +drm_sched_job_dependency(struct drm_sched_job *job, + struct drm_sched_entity *entity) +{ + if (!xa_empty(&job->dependencies)) + return xa_erase(&job->dependencies, job->last_dependency++); + + if (job->sched->ops->dependency) + return job->sched->ops->dependency(job, entity); + + return NULL; +} + /** * drm_sched_entity_kill_jobs - Make sure all remaining jobs are killed * @@ -229,7 +242,7 @@ static void drm_sched_entity_kill_jobs(struct drm_sched_entity *entity) struct drm_sched_fence *s_fence = job->s_fence; /* Wait for all dependencies to avoid data corruptions */ - while ((f = job->sched->ops->dependency(job, entity))) + while ((f = drm_sched_job_dependency(job, entity))) dma_fence_wait(f, false); drm_sched_fence_scheduled(s_fence); @@ -419,7 +432,6 @@ static bool drm_sched_entity_add_dependency_cb(struct drm_sched_entity *entity) */ struct drm_sched_job *drm_sched_entity_pop_job(struct drm_sched_entity *entity) { - struct drm_gpu_scheduler *sched = entity->rq->sched; struct drm_sched_job *sched_job; sched_job = to_drm_sched_job(spsc_queue_peek(&entity->job_queue)); @@ -427,7 +439,7 @@ struct drm_sched_job *drm_sched_entity_pop_job(struct drm_sched_entity *entity) return NULL; while ((entity->dependency = - sched->ops->dependency(sched_job, entity))) { + drm_sched_job_dependency(sched_job, entity))) { trace_drm_sched_job_wait_dep(sched_job, entity->dependency); if (drm_sched_entity_add_dependency_cb(entity)) diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 5e84e1500c32..12d533486518 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -605,6 +605,8 @@ int drm_sched_job_init(struct drm_sched_job *job, INIT_LIST_HEAD(&job->list); + xa_init_flags(&job->dependencies, XA_FLAGS_ALLOC); + return 0; } EXPORT_SYMBOL(drm_sched_job_init); @@ -628,6 +630,98 @@ void drm_sched_job_arm(struct drm_sched_job *job) } EXPORT_SYMBOL(drm_sched_job_arm); +/** + * drm_sched_job_await_fence - adds the fence as a job dependency + * @job: scheduler job to add the dependencies to + * @fence: the dma_fence to add to the list of dependencies. + * + * Note that @fence is consumed in both the success and error cases. + * + * Returns: + * 0 on success, or an error on failing to expand the array. + */ +int drm_sched_job_await_fence(struct drm_sched_job *job, + struct dma_fence *fence) +{ + struct dma_fence *entry; + unsigned long index; + u32 id = 0; + int ret; + + if (!fence) + return 0; + + /* Deduplicate if we already depend on a fence from the same context. + * This lets the size of the array of deps scale with the number of + * engines involved, rather than the number of BOs. + */ + xa_for_each(&job->dependencies, index, entry) { + if (entry->context != fence->context) + continue; + + if (dma_fence_is_later(fence, entry)) { + dma_fence_put(entry); + xa_store(&job->dependencies, index, fence, GFP_KERNEL); + } else { + dma_fence_put(fence); + } + return 0; + } + + ret = xa_alloc(&job->dependencies, &id, fence, xa_limit_32b, GFP_KERNEL); + if (ret != 0) + dma_fence_put(fence); + + return ret; +} +EXPORT_SYMBOL(drm_sched_job_await_fence); + +/** + * drm_sched_job_await_implicit - adds implicit dependencies as job dependencies + * @job: scheduler job to add the dependencies to + * @obj: the gem object to add new dependencies from. + * @write: whether the job might write the object (so we need to depend on + * shared fences in the reservation object). + * + * This should be called after drm_gem_lock_reservations() on your array of + * GEM objects used in the job but before updating the reservations with your + * own fences. + * + * Returns: + * 0 on success, or an error on failing to expand the array. + */ +int drm_sched_job_await_implicit(struct drm_sched_job *job, + struct drm_gem_object *obj, + bool write) +{ + int ret; + struct dma_fence **fences; + unsigned int i, fence_count; + + if (!write) { + struct dma_fence *fence = dma_resv_get_excl_unlocked(obj->resv); + + return drm_sched_job_await_fence(job, fence); + } + + ret = dma_resv_get_fences(obj->resv, NULL, &fence_count, &fences); + if (ret || !fence_count) + return ret; + + for (i = 0; i < fence_count; i++) { + ret = drm_sched_job_await_fence(job, fences[i]); + if (ret) + break; + } + + for (; i < fence_count; i++) + dma_fence_put(fences[i]); + kfree(fences); + return ret; +} +EXPORT_SYMBOL(drm_sched_job_await_implicit); + + /** * drm_sched_job_cleanup - clean up scheduler job resources * @job: scheduler job to clean up @@ -643,6 +737,9 @@ EXPORT_SYMBOL(drm_sched_job_arm); */ void drm_sched_job_cleanup(struct drm_sched_job *job) { + struct dma_fence *fence; + unsigned long index; + if (!kref_read(&job->s_fence->finished.refcount)) { /* drm_sched_job_arm() has been called */ dma_fence_put(&job->s_fence->finished); @@ -652,6 +749,12 @@ void drm_sched_job_cleanup(struct drm_sched_job *job) } job->s_fence = NULL; + + xa_for_each(&job->dependencies, index, fence) { + dma_fence_put(fence); + } + xa_destroy(&job->dependencies); + } EXPORT_SYMBOL(drm_sched_job_cleanup); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 83afc3aa8e2f..74fb321dbc44 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -27,9 +27,12 @@ #include <drm/spsc_queue.h> #include <linux/dma-fence.h> #include <linux/completion.h> +#include <linux/xarray.h> #define MAX_WAIT_SCHED_ENTITY_Q_EMPTY msecs_to_jiffies(1000) +struct drm_gem_object; + struct drm_gpu_scheduler; struct drm_sched_rq; @@ -198,6 +201,16 @@ struct drm_sched_job { enum drm_sched_priority s_priority; struct drm_sched_entity *entity; struct dma_fence_cb cb; + /** + * @dependencies: + * + * Contains the dependencies as struct dma_fence for this job, see + * drm_sched_job_await_fence() and drm_sched_job_await_implicit(). + */ + struct xarray dependencies; + + /** @last_dependency: tracks @dependencies as they signal */ + unsigned long last_dependency; }; static inline bool drm_sched_invalidate_job(struct drm_sched_job *s_job, @@ -220,9 +233,14 @@ enum drm_gpu_sched_stat { */ struct drm_sched_backend_ops { /** - * @dependency: Called when the scheduler is considering scheduling - * this job next, to get another struct dma_fence for this job to - * block on. Once it returns NULL, run_job() may be called. + * @dependency: + * + * Called when the scheduler is considering scheduling this job next, to + * get another struct dma_fence for this job to block on. Once it + * returns NULL, run_job() may be called. + * + * If a driver exclusively uses drm_sched_job_await_fence() and + * drm_sched_job_await_implicit() this can be ommitted and left as NULL. */ struct dma_fence *(*dependency)(struct drm_sched_job *sched_job, struct drm_sched_entity *s_entity); @@ -349,6 +367,13 @@ int drm_sched_job_init(struct drm_sched_job *job, struct drm_sched_entity *entity, void *owner); void drm_sched_job_arm(struct drm_sched_job *job); +int drm_sched_job_await_fence(struct drm_sched_job *job, + struct dma_fence *fence); +int drm_sched_job_await_implicit(struct drm_sched_job *job, + struct drm_gem_object *obj, + bool write); + + void drm_sched_entity_modify_sched(struct drm_sched_entity *entity, struct drm_gpu_scheduler **sched_list, unsigned int num_sched_list); -- 2.32.0.rc2

4 years, 2 months

[PATCH 7/7] dma-resv: Give the docs a do-over

by Daniel Vetter

Specifically document the new/clarified rules around how the shared fences do not have any ordering requirements against the exclusive fence. But also document all the things a bit better, given how central struct dma_resv to dynamic buffer management the docs have been very inadequat. - Lots more links to other pieces of the puzzle. Unfortunately ttm_buffer_object has no docs, so no links :-( - Explain/complain a bit about dma_resv_locking_ctx(). I still don't like that one, but fixing the ttm call chains is going to be horrible. Plus we want to plug in real slowpath locking when we do that anyway. - Main part of the patch is some actual docs for struct dma_resv. Overall I think we still have a lot of bad naming in this area (e.g. dma_resv.fence is singular, but contains the multiple shared fences), but I think that's more indicative of how the semantics and rules are just not great. Another thing that's real awkard is how chaining exclusive fences right now means direct dma_resv.exclusive_fence pointer access with an rcu_assign_pointer. Not so great either. Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com> Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/dma-buf/dma-resv.c | 22 ++++++-- include/linux/dma-resv.h | 104 +++++++++++++++++++++++++++++++++++-- 2 files changed, 116 insertions(+), 10 deletions(-) diff --git a/drivers/dma-buf/dma-resv.c b/drivers/dma-buf/dma-resv.c index f26c71747d43..898f8d894bbd 100644 --- a/drivers/dma-buf/dma-resv.c +++ b/drivers/dma-buf/dma-resv.c @@ -48,6 +48,8 @@ * write operations) or N shared fences (read operations). The RCU * mechanism is used to protect read access to fences from locked * write-side updates. + * + * See struct dma_resv for more details. */ DEFINE_WD_CLASS(reservation_ww_class); @@ -137,7 +139,11 @@ EXPORT_SYMBOL(dma_resv_fini); * @num_fences: number of fences we want to add * * Should be called before dma_resv_add_shared_fence(). Must - * be called with obj->lock held. + * be called with @obj locked through dma_resv_lock(). + * + * Note that the preallocated slots need to be re-reserved if @obj is unlocked + * at any time before callind dma_resv_add_shared_fence(). This is validate when + * CONFIG_DEBUG_MUTEXES is enabled. * * RETURNS * Zero for success, or -errno @@ -234,8 +240,10 @@ EXPORT_SYMBOL(dma_resv_reset_shared_max); * @obj: the reservation object * @fence: the shared fence to add * - * Add a fence to a shared slot, obj->lock must be held, and + * Add a fence to a shared slot, @obj must be locked with dma_resv_lock(), and * dma_resv_reserve_shared() has been called. + * + * See also &dma_resv.fence for a discussion of the semantics. */ void dma_resv_add_shared_fence(struct dma_resv *obj, struct dma_fence *fence) { @@ -280,7 +288,9 @@ EXPORT_SYMBOL(dma_resv_add_shared_fence); * @obj: the reservation object * @fence: the shared fence to add * - * Add a fence to the exclusive slot. The obj->lock must be held. + * Add a fence to the exclusive slot. @obj must be locked with dma_resv_lock(). + * Note that this function replaces all fences attached to @obj, see also + * &dma_resv.fence_excl for a discussion of the semantics. */ void dma_resv_add_excl_fence(struct dma_resv *obj, struct dma_fence *fence) { @@ -609,9 +619,11 @@ static inline int dma_resv_test_signaled_single(struct dma_fence *passed_fence) * fence * * Callers are not required to hold specific locks, but maybe hold - * dma_resv_lock() already + * dma_resv_lock() already. + * * RETURNS - * true if all fences signaled, else false + * + * True if all fences signaled, else false. */ bool dma_resv_test_signaled(struct dma_resv *obj, bool test_all) { diff --git a/include/linux/dma-resv.h b/include/linux/dma-resv.h index e1ca2080a1ff..c77fd54d033f 100644 --- a/include/linux/dma-resv.h +++ b/include/linux/dma-resv.h @@ -62,16 +62,90 @@ struct dma_resv_list { /** * struct dma_resv - a reservation object manages fences for a buffer - * @lock: update side lock - * @seq: sequence count for managing RCU read-side synchronization - * @fence_excl: the exclusive fence, if there is one currently - * @fence: list of current shared fences + * + * There are multiple uses for this, with sometimes slightly different rules in + * how the fence slots are used. + * + * One use is to synchronize cross-driver access to a struct dma_buf, either for + * dynamic buffer management or just to handle implicit synchronization between + * different users of the buffer in userspace. See &dma_buf.resv for a more + * in-depth discussion. + * + * The other major use is to manage access and locking within a driver in a + * buffer based memory manager. struct ttm_buffer_object is the canonical + * example here, since this is were reservation objects originated from. But use + * in drivers is spreading and some drivers also manage struct + * drm_gem_object with the same scheme. */ struct dma_resv { + /** + * @lock: + * + * Update side lock. Don't use directly, instead use the wrapper + * functions like dma_resv_lock() and dma_resv_unlock(). + * + * Drivers which use the reservation object to manage memory dynamically + * also use this lock to protect buffer object state like placement, + * allocation policies or throughout command submission. + */ struct ww_mutex lock; + + /** + * @seq: + * + * Sequence count for managing RCU read-side synchronization, allows + * read-only access to @fence_excl and @fence while ensuring we take a + * consistent snapshot. + */ seqcount_ww_mutex_t seq; + /** + * @fence_excl: + * + * The exclusive fence, if there is one currently. + * + * There are two was to update this fence: + * + * - First by calling dma_resv_add_excl_fence(), which replaces all + * fences attached to the reservation object. To guarantee that no + * fences are lost this new fence must signal only after all previous + * fences, both shared and exclusive, have signalled. In some cases it + * is convenient to achieve that by attaching a struct dma_fence_array + * with all the new and old fences. + * + * - Alternatively the fence can be set directly, which leaves the + * shared fences unchanged. To guarantee that no fences are lost this + * new fence must signale only after the previous exclusive fence has + * singalled. Since the shared fences are staying intact, it is not + * necessary to maintain any ordering against those. If semantically + * only a new access is added without actually treating the previous + * one as a dependency the exclusive fences can be strung together + * using struct dma_fence_chain. + * + * Note that actual semantics of what an exclusive or shared fence mean + * is defined by the user, for reservation objects shared across drivers + * see &dma_buf.resv. + */ struct dma_fence __rcu *fence_excl; + + /** + * @fence: + * + * List of current shared fences. + * + * There are no ordering constraints of shared fences against the + * exclusive fence slot. If a waiter needs to wait for all access, it + * has to wait for both set of fences to signal. + * + * A new fence is added by calling dma_resv_add_shared_fence(). Since + * this often needs to be done past the point of no return in command + * submission it cannot fail, and therefor sufficient slots need to be + * reserved by calling dma_resv_reserve_shared(). + * + * Note that actual semantics of what an exclusive or shared fence mean + * is defined by the user, for reservation objects shared across drivers + * see &dma_buf.resv. + */ struct dma_resv_list __rcu *fence; }; @@ -98,6 +172,13 @@ static inline void dma_resv_reset_shared_max(struct dma_resv *obj) {} * undefined order, a #ww_acquire_ctx is passed to unwind if a cycle * is detected. See ww_mutex_lock() and ww_acquire_init(). A reservation * object may be locked by itself by passing NULL as @ctx. + * + * When a die situation is indicated by returning -EDEADLK all locks held by + * @ctx must be unlocked and then dma_resv_lock_slow() called on @obj. + * + * Unlocked by calling dma_resv_lock(). + * + * See also dma_resv_lock_interruptible() for the interruptible variant. */ static inline int dma_resv_lock(struct dma_resv *obj, struct ww_acquire_ctx *ctx) @@ -119,6 +200,12 @@ static inline int dma_resv_lock(struct dma_resv *obj, * undefined order, a #ww_acquire_ctx is passed to unwind if a cycle * is detected. See ww_mutex_lock() and ww_acquire_init(). A reservation * object may be locked by itself by passing NULL as @ctx. + * + * When a die situation is indicated by returning -EDEADLK all locks held by + * @ctx must be unlocked and then dma_resv_lock_slow_interruptible() called on + * @obj. + * + * Unlocked by calling dma_resv_lock(). */ static inline int dma_resv_lock_interruptible(struct dma_resv *obj, struct ww_acquire_ctx *ctx) @@ -134,6 +221,8 @@ static inline int dma_resv_lock_interruptible(struct dma_resv *obj, * Acquires the reservation object after a die case. This function * will sleep until the lock becomes available. See dma_resv_lock() as * well. + * + * See also dma_resv_lock_slow_interruptible() for the interruptible variant. */ static inline void dma_resv_lock_slow(struct dma_resv *obj, struct ww_acquire_ctx *ctx) @@ -167,7 +256,7 @@ static inline int dma_resv_lock_slow_interruptible(struct dma_resv *obj, * if they overlap with a writer. * * Also note that since no context is provided, no deadlock protection is - * possible. + * possible, which is also not needed for a trylock. * * Returns true if the lock was acquired, false otherwise. */ @@ -193,6 +282,11 @@ static inline bool dma_resv_is_locked(struct dma_resv *obj) * * Returns the context used to lock a reservation object or NULL if no context * was used or the object is not locked at all. + * + * WARNING: This interface is pretty horrible, but TTM needs it because it + * doesn't pass the struct ww_acquire_ctx around in some very long callchains. + * Everyone else just uses it to check whether they're holding a reservation or + * not. */ static inline struct ww_acquire_ctx *dma_resv_locking_ctx(struct dma_resv *obj) { -- 2.32.0

4 years, 2 months

Re: [Linaro-mm-sig] [PATCH v4 0/2] Add p2p via dmabuf to habanalabs

by Daniel Vetter

On Tue, Jul 6, 2021 at 12:03 PM Oded Gabbay <oded.gabbay(a)gmail.com> wrote: > > On Tue, Jul 6, 2021 at 11:40 AM Daniel Vetter <daniel(a)ffwll.ch> wrote: > > > > On Mon, Jul 05, 2021 at 04:03:12PM +0300, Oded Gabbay wrote: > > > Hi, > > > I'm sending v4 of this patch-set following the long email thread. > > > I want to thank Jason for reviewing v3 and pointing out the errors, saving > > > us time later to debug it :) > > > > > > I consulted with Christian on how to fix patch 2 (the implementation) and > > > at the end of the day I shamelessly copied the relevant content from > > > amdgpu_vram_mgr_alloc_sgt() and amdgpu_dma_buf_attach(), regarding the > > > usage of dma_map_resource() and pci_p2pdma_distance_many(), respectively. > > > > > > I also made a few improvements after looking at the relevant code in amdgpu. > > > The details are in the changelog of patch 2. > > > > > > I took the time to write an import code into the driver, allowing me to > > > check real P2P with two Gaudi devices, one as exporter and the other as > > > importer. I'm not going to include the import code in the product, it was > > > just for testing purposes (although I can share it if anyone wants). > > > > > > I run it on a bare-metal environment with IOMMU enabled, on a sky-lake CPU > > > with a white-listed PCIe bridge (to make the pci_p2pdma_distance_many happy). > > > > > > Greg, I hope this will be good enough for you to merge this code. > > > > So we're officially going to use dri-devel for technical details review > > and then Greg for merging so we don't have to deal with other merge > > criteria dri-devel folks have? > I'm glad to receive any help or review, regardless of the subsystem > the person giving that help belongs to. > > > > > I don't expect anything less by now, but it does make the original claim > > that drivers/misc will not step all over accelerators folks a complete > > farce under the totally-not-a-gpu banner. > > > > This essentially means that for any other accelerator stack that doesn't > > fit the dri-devel merge criteria, even if it's acting like a gpu and uses > > other gpu driver stuff, you can just send it to Greg and it's good to go. > > What's wrong with Greg ??? ;) > > On a more serious note, yes, I do think the dri-devel merge criteria > is very extreme, and effectively drives-out many AI accelerator > companies that want to contribute to the kernel but can't/won't open > their software IP and patents. > > I think the expectation from AI startups (who are 90% of the deep > learning field) to cooperate outside of company boundaries is not > realistic, especially on the user-side, where the real IP of the > company resides. > > Personally I don't think there is a real justification for that at > this point of time, but if it will make you (and other people here) > happy I really don't mind creating a non-gpu accelerator subsystem > that will contain all the totally-not-a-gpu accelerators, and will > have a more relaxed criteria for upstreaming. Something along an > "rdma-core" style library looks like the correct amount of user-level > open source that should be enough. > > The question is, what will happen later ? Will it be sufficient to > "allow" us to use dmabuf and maybe other gpu stuff in the future (e.g. > hmm) ? > > If the community and dri-devel maintainers (and you among them) will > assure me it is good enough, then I'll happily contribute my work and > personal time to organize this effort and implement it. I think dri-devel stance is pretty clear and well known: We want the userspace to be open, because that's where most of the driver stack is. Without an open driver stack there's no way to ever have anything cross-vendor. And that includes the compiler and anything else you need to drive the hardware. Afaik linux cpu arch ports are also not accepted if there's no open gcc or llvm port around, because without that the overall stack just becomes useless. If that means AI companies don't want to open our their hw specs enough to allow that, so be it - all you get in that case is offloading the kernel side of the stack for convenience, with zero long term prospects to ever make this into a cross vendor subsystem stack that does something useful. If the business case says you can't open up your hw enough for that, I really don't see the point in merging such a driver, it'll be an unmaintainable stack by anyone else who's not having access to those NDA covered specs and patents and everything. If the stack is actually cross vendor to begin with that's just bonus, but generally that doesn't happen voluntarily and needs a few years to decades to get there. So that's not really something we require. tldr; just a runtime isn't enough for dri-devel. Now Greg seems to be happy to merge kernel drivers that aren't useful with the open bits provided, so *shrug*. Cheers, Daniel PS: If requiring an actually useful open driver stack is somehow *extreme* I have no idea why we even bother with merging device drivers to upstream. Just make a stable driver api and done, vendors can then do whatever they feel like and protect their "valuable IP and patents" or whatever it is. > Thanks, > oded > > > > > There's quite a lot of these floating around actually (and many do have > > semi-open runtimes, like habanalabs have now too, just not open enough to > > be actually useful). It's going to be absolutely lovely having to explain > > to these companies in background chats why habanalabs gets away with their > > stack and they don't. > > > > Or maybe we should just merge them all and give up on the idea of having > > open cross-vendor driver stacks for these accelerators. > > > > Thanks, Daniel > > > > > > > > Thanks, > > > Oded > > > > > > Oded Gabbay (1): > > > habanalabs: define uAPI to export FD for DMA-BUF > > > > > > Tomer Tayar (1): > > > habanalabs: add support for dma-buf exporter > > > > > > drivers/misc/habanalabs/Kconfig | 1 + > > > drivers/misc/habanalabs/common/habanalabs.h | 26 ++ > > > drivers/misc/habanalabs/common/memory.c | 480 +++++++++++++++++++- > > > drivers/misc/habanalabs/gaudi/gaudi.c | 1 + > > > drivers/misc/habanalabs/goya/goya.c | 1 + > > > include/uapi/misc/habanalabs.h | 28 +- > > > 6 files changed, 532 insertions(+), 5 deletions(-) > > > > > > -- > > > 2.25.1 > > > > > > > -- > > Daniel Vetter > > Software Engineer, Intel Corporation > > http://blog.ffwll.ch -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch

4 years, 2 months

Re: [Linaro-mm-sig] [PATCH v4 2/2] habanalabs: add support for dma-buf exporter

by Jason Gunthorpe

On Tue, Jul 06, 2021 at 12:44:49PM +0300, Oded Gabbay wrote: > > > + /* In case we got a large memory area to export, we need to divide it > > > + * to smaller areas because each entry in the dmabuf sgt can only > > > + * describe unsigned int. > > > + */ > > > > Huh? This is forming a SGL, it should follow the SGL rules which means > > you have to fragment based on the dma_get_max_seg_size() of the > > importer device. > > > hmm > I don't see anyone in drm checking this value (and using it) when > creating the SGL when exporting dmabuf. (e.g. > amdgpu_vram_mgr_alloc_sgt) For dmabuf the only importer is RDMA and it doesn't care, but you certainly should not introduce a hardwired constant instead of using the correct function. Jason

4 years, 2 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig