diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdkfd/kfd_process.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdkfd/kfd_process.c | 48 |
1 files changed, 30 insertions, 18 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index e3d64ec8c353..6c83a519b3a1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -43,6 +43,7 @@ struct mm_struct; #include "kfd_device_queue_manager.h" #include "kfd_iommu.h" #include "kfd_svm.h" +#include "kfd_smi_events.h" /* * List of struct kfd_process (field kfd_process). @@ -693,7 +694,7 @@ static void kfd_process_free_gpuvm(struct kgd_mem *mem, struct kfd_dev *dev = pdd->dev; if (kptr) { - amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(dev->adev, mem); + amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem); kptr = NULL; } @@ -733,7 +734,7 @@ static int kfd_process_alloc_gpuvm(struct kfd_process_device *pdd, } if (kptr) { - err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(kdev->adev, + err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel( (struct kgd_mem *)*mem, kptr, NULL); if (err) { pr_debug("Map GTT BO to kernel failed\n"); @@ -999,7 +1000,7 @@ static void kfd_process_kunmap_signal_bo(struct kfd_process *p) if (!mem) goto out; - amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(kdev->adev, mem); + amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem); out: mutex_unlock(&p->mutex); @@ -1114,6 +1115,15 @@ static void kfd_process_wq_release(struct work_struct *work) struct kfd_process *p = container_of(work, struct kfd_process, release_work); + kfd_process_dequeue_from_all_devices(p); + pqm_uninit(&p->pqm); + + /* Signal the eviction fence after user mode queues are + * destroyed. This allows any BOs to be freed without + * triggering pointless evictions or waiting for fences. + */ + dma_fence_signal(p->ef); + kfd_process_remove_sysfs(p); kfd_iommu_unbind_process(p); @@ -1178,20 +1188,8 @@ static void kfd_process_notifier_release(struct mmu_notifier *mn, cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); - mutex_lock(&p->mutex); - - kfd_process_dequeue_from_all_devices(p); - pqm_uninit(&p->pqm); - /* Indicate to other users that MM is no longer valid */ p->mm = NULL; - /* Signal the eviction fence after user mode queues are - * destroyed. This allows any BOs to be freed without - * triggering pointless evictions or waiting for fences. - */ - dma_fence_signal(p->ef); - - mutex_unlock(&p->mutex); mmu_notifier_put(&p->mmu_notifier); } @@ -1404,6 +1402,11 @@ static struct kfd_process *create_process(const struct task_struct *thread) hash_add_rcu(kfd_processes_table, &process->kfd_processes, (uintptr_t)process->mm); + /* Avoid free_notifier to start kfd_process_wq_release if + * mmu_notifier_get failed because of pending signal. + */ + kref_get(&process->ref); + /* MMU notifier registration must be the last call that can fail * because after this point we cannot unwind the process creation. * After this point, mmu_notifier_put will trigger the cleanup by @@ -1416,6 +1419,7 @@ static struct kfd_process *create_process(const struct task_struct *thread) } BUG_ON(mn != &process->mmu_notifier); + kfd_unref_process(process); get_task_struct(process->lead_thread); return process; @@ -1736,7 +1740,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm) * Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues(struct kfd_process *p) +int kfd_process_evict_queues(struct kfd_process *p, uint32_t trigger) { int r = 0; int i; @@ -1745,6 +1749,9 @@ int kfd_process_evict_queues(struct kfd_process *p) for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; + kfd_smi_event_queue_eviction(pdd->dev, p->lead_thread->pid, + trigger); + r = pdd->dev->dqm->ops.evict_process_queues(pdd->dev->dqm, &pdd->qpd); /* evict return -EIO if HWS is hang or asic is resetting, in this case @@ -1769,6 +1776,9 @@ fail: if (n_evicted == 0) break; + + kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid); + if (pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm, &pdd->qpd)) pr_err("Failed to restore queues\n"); @@ -1788,6 +1798,8 @@ int kfd_process_restore_queues(struct kfd_process *p) for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; + kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid); + r = pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm, &pdd->qpd); if (r) { @@ -1849,7 +1861,7 @@ static void evict_process_worker(struct work_struct *work) flush_delayed_work(&p->restore_work); pr_debug("Started evicting pasid 0x%x\n", p->pasid); - ret = kfd_process_evict_queues(p); + ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_TTM); if (!ret) { dma_fence_signal(p->ef); dma_fence_put(p->ef); @@ -1916,7 +1928,7 @@ void kfd_suspend_all_processes(void) cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); - if (kfd_process_evict_queues(p)) + if (kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_SUSPEND)) pr_err("Failed to suspend process 0x%x\n", p->pasid); dma_fence_signal(p->ef); dma_fence_put(p->ef); |