// SPDX-License-Identifier: GPL-2.0 OR MIT /* * Copyright 2014-2022 Advanced Micro Devices, Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "amdgpu_amdkfd.h" #include "amdgpu.h" struct mm_struct; #include "kfd_priv.h" #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). * Unique/indexed by mm_struct* */ DEFINE_HASHTABLE(kfd_processes_table, KFD_PROCESS_TABLE_SIZE); static DEFINE_MUTEX(kfd_processes_mutex); DEFINE_SRCU(kfd_processes_srcu); /* For process termination handling */ static struct workqueue_struct *kfd_process_wq; /* Ordered, single-threaded workqueue for restoring evicted * processes. Restoring multiple processes concurrently under memory * pressure can lead to processes blocking each other from validating * their BOs and result in a live-lock situation where processes * remain evicted indefinitely. */ static struct workqueue_struct *kfd_restore_wq; static struct kfd_process *find_process(const struct task_struct *thread, bool ref); static void kfd_process_ref_release(struct kref *ref); static struct kfd_process *create_process(const struct task_struct *thread); static int kfd_process_init_cwsr_apu(struct kfd_process *p, struct file *filep); static void evict_process_worker(struct work_struct *work); static void restore_process_worker(struct work_struct *work); static void kfd_process_device_destroy_cwsr_dgpu(struct kfd_process_device *pdd); struct kfd_procfs_tree { struct kobject *kobj; }; static struct kfd_procfs_tree procfs; /* * Structure for SDMA activity tracking */ struct kfd_sdma_activity_handler_workarea { struct work_struct sdma_activity_work; struct kfd_process_device *pdd; uint64_t sdma_activity_counter; }; struct temp_sdma_queue_list { uint64_t __user *rptr; uint64_t sdma_val; unsigned int queue_id; struct list_head list; }; static void kfd_sdma_activity_worker(struct work_struct *work) { struct kfd_sdma_activity_handler_workarea *workarea; struct kfd_process_device *pdd; uint64_t val; struct mm_struct *mm; struct queue *q; struct qcm_process_device *qpd; struct device_queue_manager *dqm; int ret = 0; struct temp_sdma_queue_list sdma_q_list; struct temp_sdma_queue_list *sdma_q, *next; workarea = container_of(work, struct kfd_sdma_activity_handler_workarea, sdma_activity_work); pdd = workarea->pdd; if (!pdd) return; dqm = pdd->dev->dqm; qpd = &pdd->qpd; if (!dqm || !qpd) return; /* * Total SDMA activity is current SDMA activity + past SDMA activity * Past SDMA count is stored in pdd. * To get the current activity counters for all active SDMA queues, * we loop over all SDMA queues and get their counts from user-space. * * We cannot call get_user() with dqm_lock held as it can cause * a circular lock dependency situation. To read the SDMA stats, * we need to do the following: * * 1. Create a temporary list of SDMA queue nodes from the qpd->queues_list, * with dqm_lock/dqm_unlock(). * 2. Call get_user() for each node in temporary list without dqm_lock. * Save the SDMA count for each node and also add the count to the total * SDMA count counter. * Its possible, during this step, a few SDMA queue nodes got deleted * from the qpd->queues_list. * 3. Do a second pass over qpd->queues_list to check if any nodes got deleted. * If any node got deleted, its SDMA count would be captured in the sdma * past activity counter. So subtract the SDMA counter stored in step 2 * for this node from the total SDMA count. */ INIT_LIST_HEAD(&sdma_q_list.list); /* * Create the temp list of all SDMA queues */ dqm_lock(dqm); list_for_each_entry(q, &qpd->queues_list, list) { if ((q->properties.type != KFD_QUEUE_TYPE_SDMA) && (q->properties.type != KFD_QUEUE_TYPE_SDMA_XGMI)) continue; sdma_q = kzalloc(sizeof(struct temp_sdma_queue_list), GFP_KERNEL); if (!sdma_q) { dqm_unlock(dqm); goto cleanup; } INIT_LIST_HEAD(&sdma_q->list); sdma_q->rptr = (uint64_t __user *)q->properties.read_ptr; sdma_q->queue_id = q->properties.queue_id; list_add_tail(&sdma_q->list, &sdma_q_list.list); } /* * If the temp list is empty, then no SDMA queues nodes were found in * qpd->queues_list. Return the past activity count as the total sdma * count */ if (list_empty(&sdma_q_list.list)) { workarea->sdma_activity_counter = pdd->sdma_past_activity_counter; dqm_unlock(dqm); return; } dqm_unlock(dqm); /* * Get the usage count for each SDMA queue in temp_list. */ mm = get_task_mm(pdd->process->lead_thread); if (!mm) goto cleanup; kthread_use_mm(mm); list_for_each_entry(sdma_q, &sdma_q_list.list, list) { val = 0; ret = read_sdma_queue_counter(sdma_q->rptr, &val); if (ret) { pr_debug("Failed to read SDMA queue active counter for queue id: %d", sdma_q->queue_id); } else { sdma_q->sdma_val = val; workarea->sdma_activity_counter += val; } } kthread_unuse_mm(mm); mmput(mm); /* * Do a second iteration over qpd_queues_list to check if any SDMA * nodes got deleted while fetching SDMA counter. */ dqm_lock(dqm); workarea->sdma_activity_counter += pdd->sdma_past_activity_counter; list_for_each_entry(q, &qpd->queues_list, list) { if (list_empty(&sdma_q_list.list)) break; if ((q->properties.type != KFD_QUEUE_TYPE_SDMA) && (q->properties.type != KFD_QUEUE_TYPE_SDMA_XGMI)) continue; list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) { if (((uint64_t __user *)q->properties.read_ptr == sdma_q->rptr) && (sdma_q->queue_id == q->properties.queue_id)) { list_del(&sdma_q->list); kfree(sdma_q); break; } } } dqm_unlock(dqm); /* * If temp list is not empty, it implies some queues got deleted * from qpd->queues_list during SDMA usage read. Subtract the SDMA * count for each node from the total SDMA count. */ list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) { workarea->sdma_activity_counter -= sdma_q->sdma_val; list_del(&sdma_q->list); kfree(sdma_q); } return; cleanup: list_for_each_entry_safe(sdma_q, next, &sdma_q_list.list, list) { list_del(&sdma_q->list); kfree(sdma_q); } } /** * kfd_get_cu_occupancy - Collect number of waves in-flight on this device * by current process. Translates acquired wave count into number of compute units * that are occupied. * * @attr: Handle of attribute that allows reporting of wave count. The attribute * handle encapsulates GPU device it is associated with, thereby allowing collection * of waves in flight, etc * @buffer: Handle of user provided buffer updated with wave count * * Return: Number of bytes written to user buffer or an error value */ static int kfd_get_cu_occupancy(struct attribute *attr, char *buffer) { int cu_cnt; int wave_cnt; int max_waves_per_cu; struct kfd_dev *dev = NULL; struct kfd_process *proc = NULL; struct kfd_process_device *pdd = NULL; pdd = container_of(attr, struct kfd_process_device, attr_cu_occupancy); dev = pdd->dev; if (dev->kfd2kgd->get_cu_occupancy == NULL) return -EINVAL; cu_cnt = 0; proc = pdd->process; if (pdd->qpd.queue_count == 0) { pr_debug("Gpu-Id: %d has no active queues for process %d\n", dev->id, proc->pasid); return snprintf(buffer, PAGE_SIZE, "%d\n", cu_cnt); } /* Collect wave count from device if it supports */ wave_cnt = 0; max_waves_per_cu = 0; dev->kfd2kgd->get_cu_occupancy(dev->adev, proc->pasid, &wave_cnt, &max_waves_per_cu); /* Translate wave count to number of compute units */ cu_cnt = (wave_cnt + (max_waves_per_cu - 1)) / max_waves_per_cu; return snprintf(buffer, PAGE_SIZE, "%d\n", cu_cnt); } static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr, char *buffer) { if (strcmp(attr->name, "pasid") == 0) { struct kfd_process *p = container_of(attr, struct kfd_process, attr_pasid); return snprintf(buffer, PAGE_SIZE, "%d\n", p->pasid); } else if (strncmp(attr->name, "vram_", 5) == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_vram); return snprintf(buffer, PAGE_SIZE, "%llu\n", READ_ONCE(pdd->vram_usage)); } else if (strncmp(attr->name, "sdma_", 5) == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_sdma); struct kfd_sdma_activity_handler_workarea sdma_activity_work_handler; INIT_WORK(&sdma_activity_work_handler.sdma_activity_work, kfd_sdma_activity_worker); sdma_activity_work_handler.pdd = pdd; sdma_activity_work_handler.sdma_activity_counter = 0; schedule_work(&sdma_activity_work_handler.sdma_activity_work); flush_work(&sdma_activity_work_handler.sdma_activity_work); return snprintf(buffer, PAGE_SIZE, "%llu\n", (sdma_activity_work_handler.sdma_activity_counter)/ SDMA_ACTIVITY_DIVISOR); } else { pr_err("Invalid attribute"); return -EINVAL; } return 0; } static void kfd_procfs_kobj_release(struct kobject *kobj) { kfree(kobj); } static const struct sysfs_ops kfd_procfs_ops = { .show = kfd_procfs_show, }; static struct kobj_type procfs_type = { .release = kfd_procfs_kobj_release, .sysfs_ops = &kfd_procfs_ops, }; void kfd_procfs_init(void) { int ret = 0; procfs.kobj = kfd_alloc_struct(procfs.kobj); if (!procfs.kobj) return; ret = kobject_init_and_add(procfs.kobj, &procfs_type, &kfd_device->kobj, "proc"); if (ret) { pr_warn("Could not create procfs proc folder"); /* If we fail to create the procfs, clean up */ kfd_procfs_shutdown(); } } void kfd_procfs_shutdown(void) { if (procfs.kobj) { kobject_del(procfs.kobj); kobject_put(procfs.kobj); procfs.kobj = NULL; } } static ssize_t kfd_procfs_queue_show(struct kobject *kobj, struct attribute *attr, char *buffer) { struct queue *q = container_of(kobj, struct queue, kobj); if (!strcmp(attr->name, "size")) return snprintf(buffer, PAGE_SIZE, "%llu", q->properties.queue_size); else if (!strcmp(attr->name, "type")) return snprintf(buffer, PAGE_SIZE, "%d", q->properties.type); else if (!strcmp(attr->name, "gpuid")) return snprintf(buffer, PAGE_SIZE, "%u", q->device->id); else pr_err("Invalid attribute"); return 0; } static ssize_t kfd_procfs_stats_show(struct kobject *kobj, struct attribute *attr, char *buffer) { if (strcmp(attr->name, "evicted_ms") == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_evict); uint64_t evict_jiffies; evict_jiffies = atomic64_read(&pdd->evict_duration_counter); return snprintf(buffer, PAGE_SIZE, "%llu\n", jiffies64_to_msecs(evict_jiffies)); /* Sysfs handle that gets CU occupancy is per device */ } else if (strcmp(attr->name, "cu_occupancy") == 0) { return kfd_get_cu_occupancy(attr, buffer); } else { pr_err("Invalid attribute"); } return 0; } static ssize_t kfd_sysfs_counters_show(struct kobject *kobj, struct attribute *attr, char *buf) { struct kfd_process_device *pdd; if (!strcmp(attr->name, "faults")) { pdd = container_of(attr, struct kfd_process_device, attr_faults); return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->faults)); } if (!strcmp(attr->name, "page_in")) { pdd = container_of(attr, struct kfd_process_device, attr_page_in); return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->page_in)); } if (!strcmp(attr->name, "page_out")) { pdd = container_of(attr, struct kfd_process_device, attr_page_out); return sysfs_emit(buf, "%llu\n", READ_ONCE(pdd->page_out)); } return 0; } static struct attribute attr_queue_size = { .name = "size", .mode = KFD_SYSFS_FILE_MODE }; static struct attribute attr_queue_type = { .name = "type", .mode = KFD_SYSFS_FILE_MODE }; static struct attribute attr_queue_gpuid = { .name = "gpuid", .mode = KFD_SYSFS_FILE_MODE }; static struct attribute *procfs_queue_attrs[] = { &attr_queue_size, &attr_queue_type, &attr_queue_gpuid, NULL }; ATTRIBUTE_GROUPS(procfs_queue); static const struct sysfs_ops procfs_queue_ops = { .show = kfd_procfs_queue_show, }; static struct kobj_type procfs_queue_type = { .sysfs_ops = &procfs_queue_ops, .default_groups = procfs_queue_groups, }; static const struct sysfs_ops procfs_stats_ops = { .show = kfd_procfs_stats_show, }; static struct kobj_type procfs_stats_type = { .sysfs_ops = &procfs_stats_ops, .release = kfd_procfs_kobj_release, }; static const struct sysfs_ops sysfs_counters_ops = { .show = kfd_sysfs_counters_show, }; static struct kobj_type sysfs_counters_type = { .sysfs_ops = &sysfs_counters_ops, .release = kfd_procfs_kobj_release, }; int kfd_procfs_add_queue(struct queue *q) { struct kfd_process *proc; int ret; if (!q || !q->process) return -EINVAL; proc = q->process; /* Create proc//queues/ folder */ if (!proc->kobj_queues) return -EFAULT; ret = kobject_init_and_add(&q->kobj, &procfs_queue_type, proc->kobj_queues, "%u", q->properties.queue_id); if (ret < 0) { pr_warn("Creating proc//queues/%u failed", q->properties.queue_id); kobject_put(&q->kobj); return ret; } return 0; } static void kfd_sysfs_create_file(struct kobject *kobj, struct attribute *attr, char *name) { int ret; if (!kobj || !attr || !name) return; attr->name = name; attr->mode = KFD_SYSFS_FILE_MODE; sysfs_attr_init(attr); ret = sysfs_create_file(kobj, attr); if (ret) pr_warn("Create sysfs %s/%s failed %d", kobj->name, name, ret); } static void kfd_procfs_add_sysfs_stats(struct kfd_process *p) { int ret; int i; char stats_dir_filename[MAX_SYSFS_FILENAME_LEN]; if (!p || !p->kobj) return; /* * Create sysfs files for each GPU: * - proc//stats_/ * - proc//stats_/evicted_ms * - proc//stats_/cu_occupancy */ for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; snprintf(stats_dir_filename, MAX_SYSFS_FILENAME_LEN, "stats_%u", pdd->dev->id); pdd->kobj_stats = kfd_alloc_struct(pdd->kobj_stats); if (!pdd->kobj_stats) return; ret = kobject_init_and_add(pdd->kobj_stats, &procfs_stats_type, p->kobj, stats_dir_filename); if (ret) { pr_warn("Creating KFD proc/stats_%s folder failed", stats_dir_filename); kobject_put(pdd->kobj_stats); pdd->kobj_stats = NULL; return; } kfd_sysfs_create_file(pdd->kobj_stats, &pdd->attr_evict, "evicted_ms"); /* Add sysfs file to report compute unit occupancy */ if (pdd->dev->kfd2kgd->get_cu_occupancy) kfd_sysfs_create_file(pdd->kobj_stats, &pdd->attr_cu_occupancy, "cu_occupancy"); } } static void kfd_procfs_add_sysfs_counters(struct kfd_process *p) { int ret = 0; int i; char counters_dir_filename[MAX_SYSFS_FILENAME_LEN]; if (!p || !p->kobj) return; /* * Create sysfs files for each GPU which supports SVM * - proc//counters_/ * - proc//counters_/faults * - proc//counters_/page_in * - proc//counters_/page_out */ for_each_set_bit(i, p->svms.bitmap_supported, p->n_pdds) { struct kfd_process_device *pdd = p->pdds[i]; struct kobject *kobj_counters; snprintf(counters_dir_filename, MAX_SYSFS_FILENAME_LEN, "counters_%u", pdd->dev->id); kobj_counters = kfd_alloc_struct(kobj_counters); if (!kobj_counters) return; ret = kobject_init_and_add(kobj_counters, &sysfs_counters_type, p->kobj, counters_dir_filename); if (ret) { pr_warn("Creating KFD proc/%s folder failed", counters_dir_filename); kobject_put(kobj_counters); return; } pdd->kobj_counters = kobj_counters; kfd_sysfs_create_file(kobj_counters, &pdd->attr_faults, "faults"); kfd_sysfs_create_file(kobj_counters, &pdd->attr_page_in, "page_in"); kfd_sysfs_create_file(kobj_counters, &pdd->attr_page_out, "page_out"); } } static void kfd_procfs_add_sysfs_files(struct kfd_process *p) { int i; if (!p || !p->kobj) return; /* * Create sysfs files for each GPU: * - proc//vram_ * - proc//sdma_ */ for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; snprintf(pdd->vram_filename, MAX_SYSFS_FILENAME_LEN, "vram_%u", pdd->dev->id); kfd_sysfs_create_file(p->kobj, &pdd->attr_vram, pdd->vram_filename); snprintf(pdd->sdma_filename, MAX_SYSFS_FILENAME_LEN, "sdma_%u", pdd->dev->id); kfd_sysfs_create_file(p->kobj, &pdd->attr_sdma, pdd->sdma_filename); } } void kfd_procfs_del_queue(struct queue *q) { if (!q) return; kobject_del(&q->kobj); kobject_put(&q->kobj); } int kfd_process_create_wq(void) { if (!kfd_process_wq) kfd_process_wq = alloc_workqueue("kfd_process_wq", 0, 0); if (!kfd_restore_wq) kfd_restore_wq = alloc_ordered_workqueue("kfd_restore_wq", 0); if (!kfd_process_wq || !kfd_restore_wq) { kfd_process_destroy_wq(); return -ENOMEM; } return 0; } void kfd_process_destroy_wq(void) { if (kfd_process_wq) { destroy_workqueue(kfd_process_wq); kfd_process_wq = NULL; } if (kfd_restore_wq) { destroy_workqueue(kfd_restore_wq); kfd_restore_wq = NULL; } } static void kfd_process_free_gpuvm(struct kgd_mem *mem, struct kfd_process_device *pdd, void **kptr) { struct kfd_dev *dev = pdd->dev; if (kptr && *kptr) { amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem); *kptr = NULL; } amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(dev->adev, mem, pdd->drm_priv); amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->adev, mem, pdd->drm_priv, NULL); } /* kfd_process_alloc_gpuvm - Allocate GPU VM for the KFD process * This function should be only called right after the process * is created and when kfd_processes_mutex is still being held * to avoid concurrency. Because of that exclusiveness, we do * not need to take p->mutex. */ static int kfd_process_alloc_gpuvm(struct kfd_process_device *pdd, uint64_t gpu_va, uint32_t size, uint32_t flags, struct kgd_mem **mem, void **kptr) { struct kfd_dev *kdev = pdd->dev; int err; err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(kdev->adev, gpu_va, size, pdd->drm_priv, mem, NULL, flags, false); if (err) goto err_alloc_mem; err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(kdev->adev, *mem, pdd->drm_priv); if (err) goto err_map_mem; err = amdgpu_amdkfd_gpuvm_sync_memory(kdev->adev, *mem, true); if (err) { pr_debug("Sync memory failed, wait interrupted by user signal\n"); goto sync_memory_failed; } if (kptr) { 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"); goto sync_memory_failed; } } return err; sync_memory_failed: amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(kdev->adev, *mem, pdd->drm_priv); err_map_mem: amdgpu_amdkfd_gpuvm_free_memory_of_gpu(kdev->adev, *mem, pdd->drm_priv, NULL); err_alloc_mem: *mem = NULL; *kptr = NULL; return err; } /* kfd_process_device_reserve_ib_mem - Reserve memory inside the * process for IB usage The memory reserved is for KFD to submit * IB to AMDGPU from kernel. If the memory is reserved * successfully, ib_kaddr will have the CPU/kernel * address. Check ib_kaddr before accessing the memory. */ static int kfd_process_device_reserve_ib_mem(struct kfd_process_device *pdd) { struct qcm_process_device *qpd = &pdd->qpd; uint32_t flags = KFD_IOC_ALLOC_MEM_FLAGS_GTT | KFD_IOC_ALLOC_MEM_FLAGS_NO_SUBSTITUTE | KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE | KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE; struct kgd_mem *mem; void *kaddr; int ret; if (qpd->ib_kaddr || !qpd->ib_base) return 0; /* ib_base is only set for dGPU */ ret = kfd_process_alloc_gpuvm(pdd, qpd->ib_base, PAGE_SIZE, flags, &mem, &kaddr); if (ret) return ret; qpd->ib_mem = mem; qpd->ib_kaddr = kaddr; return 0; } static void kfd_process_device_destroy_ib_mem(struct kfd_process_device *pdd) { struct qcm_process_device *qpd = &pdd->qpd; if (!qpd->ib_kaddr || !qpd->ib_base) return; kfd_process_free_gpuvm(qpd->ib_mem, pdd, &qpd->ib_kaddr); } struct kfd_process *kfd_create_process(struct file *filep) { struct kfd_process *process; struct task_struct *thread = current; int ret; if (!thread->mm) return ERR_PTR(-EINVAL); /* Only the pthreads threading model is supported. */ if (thread->group_leader->mm != thread->mm) return ERR_PTR(-EINVAL); /* * take kfd processes mutex before starting of process creation * so there won't be a case where two threads of the same process * create two kfd_process structures */ mutex_lock(&kfd_processes_mutex); /* A prior open of /dev/kfd could have already created the process. */ process = find_process(thread, false); if (process) { pr_debug("Process already found\n"); } else { process = create_process(thread); if (IS_ERR(process)) goto out; ret = kfd_process_init_cwsr_apu(process, filep); if (ret) goto out_destroy; if (!procfs.kobj) goto out; process->kobj = kfd_alloc_struct(process->kobj); if (!process->kobj) { pr_warn("Creating procfs kobject failed"); goto out; } ret = kobject_init_and_add(process->kobj, &procfs_type, procfs.kobj, "%d", (int)process->lead_thread->pid); if (ret) { pr_warn("Creating procfs pid directory failed"); kobject_put(process->kobj); goto out; } kfd_sysfs_create_file(process->kobj, &process->attr_pasid, "pasid"); process->kobj_queues = kobject_create_and_add("queues", process->kobj); if (!process->kobj_queues) pr_warn("Creating KFD proc/queues folder failed"); kfd_procfs_add_sysfs_stats(process); kfd_procfs_add_sysfs_files(process); kfd_procfs_add_sysfs_counters(process); } out: if (!IS_ERR(process)) kref_get(&process->ref); mutex_unlock(&kfd_processes_mutex); return process; out_destroy: hash_del_rcu(&process->kfd_processes); mutex_unlock(&kfd_processes_mutex); synchronize_srcu(&kfd_processes_srcu); /* kfd_process_free_notifier will trigger the cleanup */ mmu_notifier_put(&process->mmu_notifier); return ERR_PTR(ret); } struct kfd_process *kfd_get_process(const struct task_struct *thread) { struct kfd_process *process; if (!thread->mm) return ERR_PTR(-EINVAL); /* Only the pthreads threading model is supported. */ if (thread->group_leader->mm != thread->mm) return ERR_PTR(-EINVAL); process = find_process(thread, false); if (!process) return ERR_PTR(-EINVAL); return process; } static struct kfd_process *find_process_by_mm(const struct mm_struct *mm) { struct kfd_process *process; hash_for_each_possible_rcu(kfd_processes_table, process, kfd_processes, (uintptr_t)mm) if (process->mm == mm) return process; return NULL; } static struct kfd_process *find_process(const struct task_struct *thread, bool ref) { struct kfd_process *p; int idx; idx = srcu_read_lock(&kfd_processes_srcu); p = find_process_by_mm(thread->mm); if (p && ref) kref_get(&p->ref); srcu_read_unlock(&kfd_processes_srcu, idx); return p; } void kfd_unref_process(struct kfd_process *p) { kref_put(&p->ref, kfd_process_ref_release); } /* This increments the process->ref counter. */ struct kfd_process *kfd_lookup_process_by_pid(struct pid *pid) { struct task_struct *task = NULL; struct kfd_process *p = NULL; if (!pid) { task = current; get_task_struct(task); } else { task = get_pid_task(pid, PIDTYPE_PID); } if (task) { p = find_process(task, true); put_task_struct(task); } return p; } static void kfd_process_device_free_bos(struct kfd_process_device *pdd) { struct kfd_process *p = pdd->process; void *mem; int id; int i; /* * Remove all handles from idr and release appropriate * local memory object */ idr_for_each_entry(&pdd->alloc_idr, mem, id) { for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *peer_pdd = p->pdds[i]; if (!peer_pdd->drm_priv) continue; amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu( peer_pdd->dev->adev, mem, peer_pdd->drm_priv); } amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->adev, mem, pdd->drm_priv, NULL); kfd_process_device_remove_obj_handle(pdd, id); } } /* * Just kunmap and unpin signal BO here. It will be freed in * kfd_process_free_outstanding_kfd_bos() */ static void kfd_process_kunmap_signal_bo(struct kfd_process *p) { struct kfd_process_device *pdd; struct kfd_dev *kdev; void *mem; kdev = kfd_device_by_id(GET_GPU_ID(p->signal_handle)); if (!kdev) return; mutex_lock(&p->mutex); pdd = kfd_get_process_device_data(kdev, p); if (!pdd) goto out; mem = kfd_process_device_translate_handle( pdd, GET_IDR_HANDLE(p->signal_handle)); if (!mem) goto out; amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem); out: mutex_unlock(&p->mutex); } static void kfd_process_free_outstanding_kfd_bos(struct kfd_process *p) { int i; for (i = 0; i < p->n_pdds; i++) kfd_process_device_free_bos(p->pdds[i]); } static void kfd_process_destroy_pdds(struct kfd_process *p) { int i; for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; pr_debug("Releasing pdd (topology id %d) for process (pasid 0x%x)\n", pdd->dev->id, p->pasid); kfd_process_device_destroy_cwsr_dgpu(pdd); kfd_process_device_destroy_ib_mem(pdd); if (pdd->drm_file) { amdgpu_amdkfd_gpuvm_release_process_vm( pdd->dev->adev, pdd->drm_priv); fput(pdd->drm_file); } if (pdd->qpd.cwsr_kaddr && !pdd->qpd.cwsr_base) free_pages((unsigned long)pdd->qpd.cwsr_kaddr, get_order(KFD_CWSR_TBA_TMA_SIZE)); bitmap_free(pdd->qpd.doorbell_bitmap); idr_destroy(&pdd->alloc_idr); kfd_free_process_doorbells(pdd->dev, pdd->doorbell_index); if (pdd->dev->shared_resources.enable_mes) amdgpu_amdkfd_free_gtt_mem(pdd->dev->adev, pdd->proc_ctx_bo); /* * before destroying pdd, make sure to report availability * for auto suspend */ if (pdd->runtime_inuse) { pm_runtime_mark_last_busy(adev_to_drm(pdd->dev->adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(pdd->dev->adev)->dev); pdd->runtime_inuse = false; } kfree(pdd); p->pdds[i] = NULL; } p->n_pdds = 0; } static void kfd_process_remove_sysfs(struct kfd_process *p) { struct kfd_process_device *pdd; int i; if (!p->kobj) return; sysfs_remove_file(p->kobj, &p->attr_pasid); kobject_del(p->kobj_queues); kobject_put(p->kobj_queues); p->kobj_queues = NULL; for (i = 0; i < p->n_pdds; i++) { pdd = p->pdds[i]; sysfs_remove_file(p->kobj, &pdd->attr_vram); sysfs_remove_file(p->kobj, &pdd->attr_sdma); sysfs_remove_file(pdd->kobj_stats, &pdd->attr_evict); if (pdd->dev->kfd2kgd->get_cu_occupancy) sysfs_remove_file(pdd->kobj_stats, &pdd->attr_cu_occupancy); kobject_del(pdd->kobj_stats); kobject_put(pdd->kobj_stats); pdd->kobj_stats = NULL; } for_each_set_bit(i, p->svms.bitmap_supported, p->n_pdds) { pdd = p->pdds[i]; sysfs_remove_file(pdd->kobj_counters, &pdd->attr_faults); sysfs_remove_file(pdd->kobj_counters, &pdd->attr_page_in); sysfs_remove_file(pdd->kobj_counters, &pdd->attr_page_out); kobject_del(pdd->kobj_counters); kobject_put(pdd->kobj_counters); pdd->kobj_counters = NULL; } kobject_del(p->kobj); kobject_put(p->kobj); p->kobj = NULL; } /* No process locking is needed in this function, because the process * is not findable any more. We must assume that no other thread is * using it any more, otherwise we couldn't safely free the process * structure in the end. */ 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); kfd_process_kunmap_signal_bo(p); kfd_process_free_outstanding_kfd_bos(p); svm_range_list_fini(p); kfd_process_destroy_pdds(p); dma_fence_put(p->ef); kfd_event_free_process(p); kfd_pasid_free(p->pasid); mutex_destroy(&p->mutex); put_task_struct(p->lead_thread); kfree(p); } static void kfd_process_ref_release(struct kref *ref) { struct kfd_process *p = container_of(ref, struct kfd_process, ref); INIT_WORK(&p->release_work, kfd_process_wq_release); queue_work(kfd_process_wq, &p->release_work); } static struct mmu_notifier *kfd_process_alloc_notifier(struct mm_struct *mm) { int idx = srcu_read_lock(&kfd_processes_srcu); struct kfd_process *p = find_process_by_mm(mm); srcu_read_unlock(&kfd_processes_srcu, idx); return p ? &p->mmu_notifier : ERR_PTR(-ESRCH); } static void kfd_process_free_notifier(struct mmu_notifier *mn) { kfd_unref_process(container_of(mn, struct kfd_process, mmu_notifier)); } static void kfd_process_notifier_release_internal(struct kfd_process *p) { cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); /* Indicate to other users that MM is no longer valid */ p->mm = NULL; mmu_notifier_put(&p->mmu_notifier); } static void kfd_process_notifier_release(struct mmu_notifier *mn, struct mm_struct *mm) { struct kfd_process *p; /* * The kfd_process structure can not be free because the * mmu_notifier srcu is read locked */ p = container_of(mn, struct kfd_process, mmu_notifier); if (WARN_ON(p->mm != mm)) return; mutex_lock(&kfd_processes_mutex); /* * Do early return if table is empty. * * This could potentially happen if this function is called concurrently * by mmu_notifier and by kfd_cleanup_pocesses. * */ if (hash_empty(kfd_processes_table)) { mutex_unlock(&kfd_processes_mutex); return; } hash_del_rcu(&p->kfd_processes); mutex_unlock(&kfd_processes_mutex); synchronize_srcu(&kfd_processes_srcu); kfd_process_notifier_release_internal(p); } static const struct mmu_notifier_ops kfd_process_mmu_notifier_ops = { .release = kfd_process_notifier_release, .alloc_notifier = kfd_process_alloc_notifier, .free_notifier = kfd_process_free_notifier, }; /* * This code handles the case when driver is being unloaded before all * mm_struct are released. We need to safely free the kfd_process and * avoid race conditions with mmu_notifier that might try to free them. * */ void kfd_cleanup_processes(void) { struct kfd_process *p; struct hlist_node *p_temp; unsigned int temp; HLIST_HEAD(cleanup_list); /* * Move all remaining kfd_process from the process table to a * temp list for processing. Once done, callback from mmu_notifier * release will not see the kfd_process in the table and do early return, * avoiding double free issues. */ mutex_lock(&kfd_processes_mutex); hash_for_each_safe(kfd_processes_table, temp, p_temp, p, kfd_processes) { hash_del_rcu(&p->kfd_processes); synchronize_srcu(&kfd_processes_srcu); hlist_add_head(&p->kfd_processes, &cleanup_list); } mutex_unlock(&kfd_processes_mutex); hlist_for_each_entry_safe(p, p_temp, &cleanup_list, kfd_processes) kfd_process_notifier_release_internal(p); /* * Ensures that all outstanding free_notifier get called, triggering * the release of the kfd_process struct. */ mmu_notifier_synchronize(); } static int kfd_process_init_cwsr_apu(struct kfd_process *p, struct file *filep) { unsigned long offset; int i; for (i = 0; i < p->n_pdds; i++) { struct kfd_dev *dev = p->pdds[i]->dev; struct qcm_process_device *qpd = &p->pdds[i]->qpd; if (!dev->cwsr_enabled || qpd->cwsr_kaddr || qpd->cwsr_base) continue; offset = KFD_MMAP_TYPE_RESERVED_MEM | KFD_MMAP_GPU_ID(dev->id); qpd->tba_addr = (int64_t)vm_mmap(filep, 0, KFD_CWSR_TBA_TMA_SIZE, PROT_READ | PROT_EXEC, MAP_SHARED, offset); if (IS_ERR_VALUE(qpd->tba_addr)) { int err = qpd->tba_addr; pr_err("Failure to set tba address. error %d.\n", err); qpd->tba_addr = 0; qpd->cwsr_kaddr = NULL; return err; } memcpy(qpd->cwsr_kaddr, dev->cwsr_isa, dev->cwsr_isa_size); qpd->tma_addr = qpd->tba_addr + KFD_CWSR_TMA_OFFSET; pr_debug("set tba :0x%llx, tma:0x%llx, cwsr_kaddr:%p for pqm.\n", qpd->tba_addr, qpd->tma_addr, qpd->cwsr_kaddr); } return 0; } static int kfd_process_device_init_cwsr_dgpu(struct kfd_process_device *pdd) { struct kfd_dev *dev = pdd->dev; struct qcm_process_device *qpd = &pdd->qpd; uint32_t flags = KFD_IOC_ALLOC_MEM_FLAGS_GTT | KFD_IOC_ALLOC_MEM_FLAGS_NO_SUBSTITUTE | KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE; struct kgd_mem *mem; void *kaddr; int ret; if (!dev->cwsr_enabled || qpd->cwsr_kaddr || !qpd->cwsr_base) return 0; /* cwsr_base is only set for dGPU */ ret = kfd_process_alloc_gpuvm(pdd, qpd->cwsr_base, KFD_CWSR_TBA_TMA_SIZE, flags, &mem, &kaddr); if (ret) return ret; qpd->cwsr_mem = mem; qpd->cwsr_kaddr = kaddr; qpd->tba_addr = qpd->cwsr_base; memcpy(qpd->cwsr_kaddr, dev->cwsr_isa, dev->cwsr_isa_size); qpd->tma_addr = qpd->tba_addr + KFD_CWSR_TMA_OFFSET; pr_debug("set tba :0x%llx, tma:0x%llx, cwsr_kaddr:%p for pqm.\n", qpd->tba_addr, qpd->tma_addr, qpd->cwsr_kaddr); return 0; } static void kfd_process_device_destroy_cwsr_dgpu(struct kfd_process_device *pdd) { struct kfd_dev *dev = pdd->dev; struct qcm_process_device *qpd = &pdd->qpd; if (!dev->cwsr_enabled || !qpd->cwsr_kaddr || !qpd->cwsr_base) return; kfd_process_free_gpuvm(qpd->cwsr_mem, pdd, &qpd->cwsr_kaddr); } void kfd_process_set_trap_handler(struct qcm_process_device *qpd, uint64_t tba_addr, uint64_t tma_addr) { if (qpd->cwsr_kaddr) { /* KFD trap handler is bound, record as second-level TBA/TMA * in first-level TMA. First-level trap will jump to second. */ uint64_t *tma = (uint64_t *)(qpd->cwsr_kaddr + KFD_CWSR_TMA_OFFSET); tma[0] = tba_addr; tma[1] = tma_addr; } else { /* No trap handler bound, bind as first-level TBA/TMA. */ qpd->tba_addr = tba_addr; qpd->tma_addr = tma_addr; } } bool kfd_process_xnack_mode(struct kfd_process *p, bool supported) { int i; /* On most GFXv9 GPUs, the retry mode in the SQ must match the * boot time retry setting. Mixing processes with different * XNACK/retry settings can hang the GPU. * * Different GPUs can have different noretry settings depending * on HW bugs or limitations. We need to find at least one * XNACK mode for this process that's compatible with all GPUs. * Fortunately GPUs with retry enabled (noretry=0) can run code * built for XNACK-off. On GFXv9 it may perform slower. * * Therefore applications built for XNACK-off can always be * supported and will be our fallback if any GPU does not * support retry. */ for (i = 0; i < p->n_pdds; i++) { struct kfd_dev *dev = p->pdds[i]->dev; /* Only consider GFXv9 and higher GPUs. Older GPUs don't * support the SVM APIs and don't need to be considered * for the XNACK mode selection. */ if (!KFD_IS_SOC15(dev)) continue; /* Aldebaran can always support XNACK because it can support * per-process XNACK mode selection. But let the dev->noretry * setting still influence the default XNACK mode. */ if (supported && KFD_SUPPORT_XNACK_PER_PROCESS(dev)) continue; /* GFXv10 and later GPUs do not support shader preemption * during page faults. This can lead to poor QoS for queue * management and memory-manager-related preemptions or * even deadlocks. */ if (KFD_GC_VERSION(dev) >= IP_VERSION(10, 1, 1)) return false; if (dev->noretry) return false; } return true; } /* * On return the kfd_process is fully operational and will be freed when the * mm is released */ static struct kfd_process *create_process(const struct task_struct *thread) { struct kfd_process *process; struct mmu_notifier *mn; int err = -ENOMEM; process = kzalloc(sizeof(*process), GFP_KERNEL); if (!process) goto err_alloc_process; kref_init(&process->ref); mutex_init(&process->mutex); process->mm = thread->mm; process->lead_thread = thread->group_leader; process->n_pdds = 0; process->queues_paused = false; INIT_DELAYED_WORK(&process->eviction_work, evict_process_worker); INIT_DELAYED_WORK(&process->restore_work, restore_process_worker); process->last_restore_timestamp = get_jiffies_64(); err = kfd_event_init_process(process); if (err) goto err_event_init; process->is_32bit_user_mode = in_compat_syscall(); process->pasid = kfd_pasid_alloc(); if (process->pasid == 0) { err = -ENOSPC; goto err_alloc_pasid; } err = pqm_init(&process->pqm, process); if (err != 0) goto err_process_pqm_init; /* init process apertures*/ err = kfd_init_apertures(process); if (err != 0) goto err_init_apertures; /* Check XNACK support after PDDs are created in kfd_init_apertures */ process->xnack_enabled = kfd_process_xnack_mode(process, false); err = svm_range_list_init(process); if (err) goto err_init_svm_range_list; /* alloc_notifier needs to find the process in the hash table */ 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 * dropping the last process reference in the free_notifier. */ mn = mmu_notifier_get(&kfd_process_mmu_notifier_ops, process->mm); if (IS_ERR(mn)) { err = PTR_ERR(mn); goto err_register_notifier; } BUG_ON(mn != &process->mmu_notifier); kfd_unref_process(process); get_task_struct(process->lead_thread); return process; err_register_notifier: hash_del_rcu(&process->kfd_processes); svm_range_list_fini(process); err_init_svm_range_list: kfd_process_free_outstanding_kfd_bos(process); kfd_process_destroy_pdds(process); err_init_apertures: pqm_uninit(&process->pqm); err_process_pqm_init: kfd_pasid_free(process->pasid); err_alloc_pasid: kfd_event_free_process(process); err_event_init: mutex_destroy(&process->mutex); kfree(process); err_alloc_process: return ERR_PTR(err); } static int init_doorbell_bitmap(struct qcm_process_device *qpd, struct kfd_dev *dev) { unsigned int i; int range_start = dev->shared_resources.non_cp_doorbells_start; int range_end = dev->shared_resources.non_cp_doorbells_end; if (!KFD_IS_SOC15(dev)) return 0; qpd->doorbell_bitmap = bitmap_zalloc(KFD_MAX_NUM_OF_QUEUES_PER_PROCESS, GFP_KERNEL); if (!qpd->doorbell_bitmap) return -ENOMEM; /* Mask out doorbells reserved for SDMA, IH, and VCN on SOC15. */ pr_debug("reserved doorbell 0x%03x - 0x%03x\n", range_start, range_end); pr_debug("reserved doorbell 0x%03x - 0x%03x\n", range_start + KFD_QUEUE_DOORBELL_MIRROR_OFFSET, range_end + KFD_QUEUE_DOORBELL_MIRROR_OFFSET); for (i = 0; i < KFD_MAX_NUM_OF_QUEUES_PER_PROCESS / 2; i++) { if (i >= range_start && i <= range_end) { __set_bit(i, qpd->doorbell_bitmap); __set_bit(i + KFD_QUEUE_DOORBELL_MIRROR_OFFSET, qpd->doorbell_bitmap); } } return 0; } struct kfd_process_device *kfd_get_process_device_data(struct kfd_dev *dev, struct kfd_process *p) { int i; for (i = 0; i < p->n_pdds; i++) if (p->pdds[i]->dev == dev) return p->pdds[i]; return NULL; } struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev, struct kfd_process *p) { struct kfd_process_device *pdd = NULL; int retval = 0; if (WARN_ON_ONCE(p->n_pdds >= MAX_GPU_INSTANCE)) return NULL; pdd = kzalloc(sizeof(*pdd), GFP_KERNEL); if (!pdd) return NULL; if (init_doorbell_bitmap(&pdd->qpd, dev)) { pr_err("Failed to init doorbell for process\n"); goto err_free_pdd; } pdd->dev = dev; INIT_LIST_HEAD(&pdd->qpd.queues_list); INIT_LIST_HEAD(&pdd->qpd.priv_queue_list); pdd->qpd.dqm = dev->dqm; pdd->qpd.pqm = &p->pqm; pdd->qpd.evicted = 0; pdd->qpd.mapped_gws_queue = false; pdd->process = p; pdd->bound = PDD_UNBOUND; pdd->already_dequeued = false; pdd->runtime_inuse = false; pdd->vram_usage = 0; pdd->sdma_past_activity_counter = 0; pdd->user_gpu_id = dev->id; atomic64_set(&pdd->evict_duration_counter, 0); if (dev->shared_resources.enable_mes) { retval = amdgpu_amdkfd_alloc_gtt_mem(dev->adev, AMDGPU_MES_PROC_CTX_SIZE, &pdd->proc_ctx_bo, &pdd->proc_ctx_gpu_addr, &pdd->proc_ctx_cpu_ptr, false); if (retval) { pr_err("failed to allocate process context bo\n"); goto err_free_pdd; } memset(pdd->proc_ctx_cpu_ptr, 0, AMDGPU_MES_PROC_CTX_SIZE); } p->pdds[p->n_pdds++] = pdd; /* Init idr used for memory handle translation */ idr_init(&pdd->alloc_idr); return pdd; err_free_pdd: kfree(pdd); return NULL; } /** * kfd_process_device_init_vm - Initialize a VM for a process-device * * @pdd: The process-device * @drm_file: Optional pointer to a DRM file descriptor * * If @drm_file is specified, it will be used to acquire the VM from * that file descriptor. If successful, the @pdd takes ownership of * the file descriptor. * * If @drm_file is NULL, a new VM is created. * * Returns 0 on success, -errno on failure. */ int kfd_process_device_init_vm(struct kfd_process_device *pdd, struct file *drm_file) { struct amdgpu_fpriv *drv_priv; struct amdgpu_vm *avm; struct kfd_process *p; struct kfd_dev *dev; int ret; if (!drm_file) return -EINVAL; if (pdd->drm_priv) return -EBUSY; ret = amdgpu_file_to_fpriv(drm_file, &drv_priv); if (ret) return ret; avm = &drv_priv->vm; p = pdd->process; dev = pdd->dev; ret = amdgpu_amdkfd_gpuvm_acquire_process_vm(dev->adev, avm, &p->kgd_process_info, &p->ef); if (ret) { pr_err("Failed to create process VM object\n"); return ret; } pdd->drm_priv = drm_file->private_data; atomic64_set(&pdd->tlb_seq, 0); ret = kfd_process_device_reserve_ib_mem(pdd); if (ret) goto err_reserve_ib_mem; ret = kfd_process_device_init_cwsr_dgpu(pdd); if (ret) goto err_init_cwsr; ret = amdgpu_amdkfd_gpuvm_set_vm_pasid(dev->adev, avm, p->pasid); if (ret) goto err_set_pasid; pdd->drm_file = drm_file; return 0; err_set_pasid: kfd_process_device_destroy_cwsr_dgpu(pdd); err_init_cwsr: kfd_process_device_destroy_ib_mem(pdd); err_reserve_ib_mem: pdd->drm_priv = NULL; amdgpu_amdkfd_gpuvm_destroy_cb(dev->adev, avm); return ret; } /* * Direct the IOMMU to bind the process (specifically the pasid->mm) * to the device. * Unbinding occurs when the process dies or the device is removed. * * Assumes that the process lock is held. */ struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev, struct kfd_process *p) { struct kfd_process_device *pdd; int err; pdd = kfd_get_process_device_data(dev, p); if (!pdd) { pr_err("Process device data doesn't exist\n"); return ERR_PTR(-ENOMEM); } if (!pdd->drm_priv) return ERR_PTR(-ENODEV); /* * signal runtime-pm system to auto resume and prevent * further runtime suspend once device pdd is created until * pdd is destroyed. */ if (!pdd->runtime_inuse) { err = pm_runtime_get_sync(adev_to_drm(dev->adev)->dev); if (err < 0) { pm_runtime_put_autosuspend(adev_to_drm(dev->adev)->dev); return ERR_PTR(err); } } err = kfd_iommu_bind_process_to_device(pdd); if (err) goto out; /* * make sure that runtime_usage counter is incremented just once * per pdd */ pdd->runtime_inuse = true; return pdd; out: /* balance runpm reference count and exit with error */ if (!pdd->runtime_inuse) { pm_runtime_mark_last_busy(adev_to_drm(dev->adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(dev->adev)->dev); } return ERR_PTR(err); } /* Create specific handle mapped to mem from process local memory idr * Assumes that the process lock is held. */ int kfd_process_device_create_obj_handle(struct kfd_process_device *pdd, void *mem) { return idr_alloc(&pdd->alloc_idr, mem, 0, 0, GFP_KERNEL); } /* Translate specific handle from process local memory idr * Assumes that the process lock is held. */ void *kfd_process_device_translate_handle(struct kfd_process_device *pdd, int handle) { if (handle < 0) return NULL; return idr_find(&pdd->alloc_idr, handle); } /* Remove specific handle from process local memory idr * Assumes that the process lock is held. */ void kfd_process_device_remove_obj_handle(struct kfd_process_device *pdd, int handle) { if (handle >= 0) idr_remove(&pdd->alloc_idr, handle); } /* This increments the process->ref counter. */ struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid) { struct kfd_process *p, *ret_p = NULL; unsigned int temp; int idx = srcu_read_lock(&kfd_processes_srcu); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { if (p->pasid == pasid) { kref_get(&p->ref); ret_p = p; break; } } srcu_read_unlock(&kfd_processes_srcu, idx); return ret_p; } /* This increments the process->ref counter. */ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm) { struct kfd_process *p; int idx = srcu_read_lock(&kfd_processes_srcu); p = find_process_by_mm(mm); if (p) kref_get(&p->ref); srcu_read_unlock(&kfd_processes_srcu, idx); return p; } /* kfd_process_evict_queues - Evict all user queues of a process * * 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, uint32_t trigger) { int r = 0; int i; unsigned int n_evicted = 0; 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 * we would like to set all the queues to be in evicted state to prevent * them been add back since they actually not be saved right now. */ if (r && r != -EIO) { pr_err("Failed to evict process queues\n"); goto fail; } n_evicted++; } return r; fail: /* To keep state consistent, roll back partial eviction by * restoring queues */ for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; 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"); n_evicted--; } return r; } /* kfd_process_restore_queues - Restore all user queues of a process */ int kfd_process_restore_queues(struct kfd_process *p) { int r, ret = 0; int i; 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) { pr_err("Failed to restore process queues\n"); if (!ret) ret = r; } } return ret; } int kfd_process_gpuidx_from_gpuid(struct kfd_process *p, uint32_t gpu_id) { int i; for (i = 0; i < p->n_pdds; i++) if (p->pdds[i] && gpu_id == p->pdds[i]->user_gpu_id) return i; return -EINVAL; } int kfd_process_gpuid_from_adev(struct kfd_process *p, struct amdgpu_device *adev, uint32_t *gpuid, uint32_t *gpuidx) { int i; for (i = 0; i < p->n_pdds; i++) if (p->pdds[i] && p->pdds[i]->dev->adev == adev) { *gpuid = p->pdds[i]->user_gpu_id; *gpuidx = i; return 0; } return -EINVAL; } static void evict_process_worker(struct work_struct *work) { int ret; struct kfd_process *p; struct delayed_work *dwork; dwork = to_delayed_work(work); /* Process termination destroys this worker thread. So during the * lifetime of this thread, kfd_process p will be valid */ p = container_of(dwork, struct kfd_process, eviction_work); WARN_ONCE(p->last_eviction_seqno != p->ef->seqno, "Eviction fence mismatch\n"); /* Narrow window of overlap between restore and evict work * item is possible. Once amdgpu_amdkfd_gpuvm_restore_process_bos * unreserves KFD BOs, it is possible to evicted again. But * restore has few more steps of finish. So lets wait for any * previous restore work to complete */ flush_delayed_work(&p->restore_work); pr_debug("Started evicting pasid 0x%x\n", p->pasid); ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_TTM); if (!ret) { dma_fence_signal(p->ef); dma_fence_put(p->ef); p->ef = NULL; queue_delayed_work(kfd_restore_wq, &p->restore_work, msecs_to_jiffies(PROCESS_RESTORE_TIME_MS)); pr_debug("Finished evicting pasid 0x%x\n", p->pasid); } else pr_err("Failed to evict queues of pasid 0x%x\n", p->pasid); } static void restore_process_worker(struct work_struct *work) { struct delayed_work *dwork; struct kfd_process *p; int ret = 0; dwork = to_delayed_work(work); /* Process termination destroys this worker thread. So during the * lifetime of this thread, kfd_process p will be valid */ p = container_of(dwork, struct kfd_process, restore_work); pr_debug("Started restoring pasid 0x%x\n", p->pasid); /* Setting last_restore_timestamp before successful restoration. * Otherwise this would have to be set by KGD (restore_process_bos) * before KFD BOs are unreserved. If not, the process can be evicted * again before the timestamp is set. * If restore fails, the timestamp will be set again in the next * attempt. This would mean that the minimum GPU quanta would be * PROCESS_ACTIVE_TIME_MS - (time to execute the following two * functions) */ p->last_restore_timestamp = get_jiffies_64(); ret = amdgpu_amdkfd_gpuvm_restore_process_bos(p->kgd_process_info, &p->ef); if (ret) { pr_debug("Failed to restore BOs of pasid 0x%x, retry after %d ms\n", p->pasid, PROCESS_BACK_OFF_TIME_MS); ret = queue_delayed_work(kfd_restore_wq, &p->restore_work, msecs_to_jiffies(PROCESS_BACK_OFF_TIME_MS)); WARN(!ret, "reschedule restore work failed\n"); return; } ret = kfd_process_restore_queues(p); if (!ret) pr_debug("Finished restoring pasid 0x%x\n", p->pasid); else pr_err("Failed to restore queues of pasid 0x%x\n", p->pasid); } void kfd_suspend_all_processes(void) { struct kfd_process *p; unsigned int temp; int idx = srcu_read_lock(&kfd_processes_srcu); WARN(debug_evictions, "Evicting all processes"); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); 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); p->ef = NULL; } srcu_read_unlock(&kfd_processes_srcu, idx); } int kfd_resume_all_processes(void) { struct kfd_process *p; unsigned int temp; int ret = 0, idx = srcu_read_lock(&kfd_processes_srcu); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { if (!queue_delayed_work(kfd_restore_wq, &p->restore_work, 0)) { pr_err("Restore process %d failed during resume\n", p->pasid); ret = -EFAULT; } } srcu_read_unlock(&kfd_processes_srcu, idx); return ret; } int kfd_reserved_mem_mmap(struct kfd_dev *dev, struct kfd_process *process, struct vm_area_struct *vma) { struct kfd_process_device *pdd; struct qcm_process_device *qpd; if ((vma->vm_end - vma->vm_start) != KFD_CWSR_TBA_TMA_SIZE) { pr_err("Incorrect CWSR mapping size.\n"); return -EINVAL; } pdd = kfd_get_process_device_data(dev, process); if (!pdd) return -EINVAL; qpd = &pdd->qpd; qpd->cwsr_kaddr = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, get_order(KFD_CWSR_TBA_TMA_SIZE)); if (!qpd->cwsr_kaddr) { pr_err("Error allocating per process CWSR buffer.\n"); return -ENOMEM; } vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE | VM_DONTDUMP | VM_PFNMAP); /* Mapping pages to user process */ return remap_pfn_range(vma, vma->vm_start, PFN_DOWN(__pa(qpd->cwsr_kaddr)), KFD_CWSR_TBA_TMA_SIZE, vma->vm_page_prot); } void kfd_flush_tlb(struct kfd_process_device *pdd, enum TLB_FLUSH_TYPE type) { struct amdgpu_vm *vm = drm_priv_to_vm(pdd->drm_priv); uint64_t tlb_seq = amdgpu_vm_tlb_seq(vm); struct kfd_dev *dev = pdd->dev; /* * It can be that we race and lose here, but that is extremely unlikely * and the worst thing which could happen is that we flush the changes * into the TLB once more which is harmless. */ if (atomic64_xchg(&pdd->tlb_seq, tlb_seq) == tlb_seq) return; if (dev->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) { /* Nothing to flush until a VMID is assigned, which * only happens when the first queue is created. */ if (pdd->qpd.vmid) amdgpu_amdkfd_flush_gpu_tlb_vmid(dev->adev, pdd->qpd.vmid); } else { amdgpu_amdkfd_flush_gpu_tlb_pasid(dev->adev, pdd->process->pasid, type); } } struct kfd_process_device *kfd_process_device_data_by_id(struct kfd_process *p, uint32_t gpu_id) { int i; if (gpu_id) { for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; if (pdd->user_gpu_id == gpu_id) return pdd; } } return NULL; } int kfd_process_get_user_gpu_id(struct kfd_process *p, uint32_t actual_gpu_id) { int i; if (!actual_gpu_id) return 0; for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; if (pdd->dev->id == actual_gpu_id) return pdd->user_gpu_id; } return -EINVAL; } #if defined(CONFIG_DEBUG_FS) int kfd_debugfs_mqds_by_process(struct seq_file *m, void *data) { struct kfd_process *p; unsigned int temp; int r = 0; int idx = srcu_read_lock(&kfd_processes_srcu); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { seq_printf(m, "Process %d PASID 0x%x:\n", p->lead_thread->tgid, p->pasid); mutex_lock(&p->mutex); r = pqm_debugfs_mqds(m, &p->pqm); mutex_unlock(&p->mutex); if (r) break; } srcu_read_unlock(&kfd_processes_srcu, idx); return r; } #endif