Merge tag 'amd-drm-next-5.10-2020-09-18' of git://people.freedesktop.org/~agd5f/linux into drm-next

amd-drm-next-5.10-2020-09-18:

amdgpu:
- Support for PCIe DPC recovery
- Sienna Cichlid updates
- Navy Flounder updates
- RAS fixes
- Refactor DC interrupt handling
- Display fixes
- Fix issues with OLED panels
- Mclk fixes for navi1x
- Watermark fixes for renoir and raven2
- Misc code cleanups
- Misc bug fixes

amdkfd:
- Fix a memory leak
- Fix a crach in GPU reset
- Add process eviction counters

radeon:
- expose sclk via sysfs hwmon interface
- Revert bad PLL fix

scheduler:
- Kernel doc fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200918204322.3931-1-alexander.deucher@amd.com

4 files changed, 126 insertions, 7 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 0e71a0543f98..e3fc6ed7b79c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -503,8 +503,8 @@ static const struct kfd_device_info *kfd_supported_devices[][2] = {
 #ifdef KFD_SUPPORT_IOMMU_V2
 	[CHIP_KAVERI] = {&kaveri_device_info, NULL},
 	[CHIP_CARRIZO] = {&carrizo_device_info, NULL},
-	[CHIP_RAVEN] = {&raven_device_info, NULL},
 #endif
+	[CHIP_RAVEN] = {&raven_device_info, NULL},
 	[CHIP_HAWAII] = {&hawaii_device_info, NULL},
 	[CHIP_TONGA] = {&tonga_device_info, NULL},
 	[CHIP_FIJI] = {&fiji_device_info, &fiji_vf_device_info},
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 560adc57a050..ed362ab8ec21 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -650,9 +650,10 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,
 		goto out;
 
 	pdd = qpd_to_pdd(qpd);
-	pr_info_ratelimited("Evicting PASID 0x%x queues\n",
+	pr_debug_ratelimited("Evicting PASID 0x%x queues\n",
 			    pdd->process->pasid);
 
+	pdd->last_evict_timestamp = get_jiffies_64();
 	/* Mark all queues as evicted. Deactivate all active queues on
 	 * the qpd.
 	 */
@@ -700,7 +701,7 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm,
 		goto out;
 
 	pdd = qpd_to_pdd(qpd);
-	pr_info_ratelimited("Evicting PASID 0x%x queues\n",
+	pr_debug_ratelimited("Evicting PASID 0x%x queues\n",
 			    pdd->process->pasid);
 
 	/* Mark all queues as evicted. Deactivate all active queues on
@@ -714,6 +715,7 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm,
 		q->properties.is_active = false;
 		decrement_queue_count(dqm, q->properties.type);
 	}
+	pdd->last_evict_timestamp = get_jiffies_64();
 	retval = execute_queues_cpsch(dqm,
 				qpd->is_debug ?
 				KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :
@@ -732,6 +734,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 	struct mqd_manager *mqd_mgr;
 	struct kfd_process_device *pdd;
 	uint64_t pd_base;
+	uint64_t eviction_duration;
 	int retval, ret = 0;
 
 	pdd = qpd_to_pdd(qpd);
@@ -746,7 +749,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 		goto out;
 	}
 
-	pr_info_ratelimited("Restoring PASID 0x%x queues\n",
+	pr_debug_ratelimited("Restoring PASID 0x%x queues\n",
 			    pdd->process->pasid);
 
 	/* Update PD Base in QPD */
@@ -799,6 +802,8 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 			ret = retval;
 	}
 	qpd->evicted = 0;
+	eviction_duration = get_jiffies_64() - pdd->last_evict_timestamp;
+	atomic64_add(eviction_duration, &pdd->evict_duration_counter);
 out:
 	if (mm)
 		mmput(mm);
@@ -812,6 +817,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
 	struct queue *q;
 	struct kfd_process_device *pdd;
 	uint64_t pd_base;
+	uint64_t eviction_duration;
 	int retval = 0;
 
 	pdd = qpd_to_pdd(qpd);
@@ -826,7 +832,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
 		goto out;
 	}
 
-	pr_info_ratelimited("Restoring PASID 0x%x queues\n",
+	pr_debug_ratelimited("Restoring PASID 0x%x queues\n",
 			    pdd->process->pasid);
 
 	/* Update PD Base in QPD */
@@ -845,6 +851,8 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm,
 	retval = execute_queues_cpsch(dqm,
 				KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0);
 	qpd->evicted = 0;
+	eviction_duration = get_jiffies_64() - pdd->last_evict_timestamp;
+	atomic64_add(eviction_duration, &pdd->evict_duration_counter);
 out:
 	dqm_unlock(dqm);
 	return retval;
@@ -1192,6 +1200,8 @@ static int stop_cpsch(struct device_queue_manager *dqm)
 	dqm->sched_running = false;
 	dqm_unlock(dqm);
 
+	pm_release_ib(&dqm->packets);
+
 	kfd_gtt_sa_free(dqm->dev, dqm->fence_mem);
 	pm_uninit(&dqm->packets, hanging);
 
@@ -1302,7 +1312,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	if (q->properties.is_active) {
 		increment_queue_count(dqm, q->properties.type);
 
-		retval = execute_queues_cpsch(dqm,
+		execute_queues_cpsch(dqm,
 				KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0);
 	}
 
@@ -1964,6 +1974,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,
 
 	if (!p)
 		return -EINVAL;
+	WARN(debug_evictions, "Evicting pid %d", p->lead_thread->pid);
 	pdd = kfd_get_process_device_data(dqm->dev, p);
 	if (pdd)
 		ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 023629f28495..8c2b8ccd27fb 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -631,7 +631,7 @@ enum kfd_pdd_bound {
 	PDD_BOUND_SUSPENDED,
 };
 
-#define MAX_SYSFS_FILENAME_LEN 11
+#define MAX_SYSFS_FILENAME_LEN 15
 
 /*
  * SDMA counter runs at 100MHz frequency.
@@ -692,6 +692,13 @@ struct kfd_process_device {
 	uint64_t sdma_past_activity_counter;
 	struct attribute attr_sdma;
 	char sdma_filename[MAX_SYSFS_FILENAME_LEN];
+
+	/* Eviction activity tracking */
+	uint64_t last_evict_timestamp;
+	atomic64_t evict_duration_counter;
+	struct attribute attr_evict;
+
+	struct kobject *kobj_stats;
 };
 
 #define qpd_to_pdd(x) container_of(x, struct kfd_process_device, qpd)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index a0e12a79ab7d..ad53b2668221 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -344,6 +344,26 @@ static ssize_t kfd_procfs_queue_show(struct kobject *kobj,
 
 	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));
+	} else
+		pr_err("Invalid attribute");
+
+	return 0;
+}
 
 static struct attribute attr_queue_size = {
 	.name = "size",
@@ -376,6 +396,19 @@ static struct kobj_type procfs_queue_type = {
 	.default_attrs = procfs_queue_attrs,
 };
 
+static const struct sysfs_ops procfs_stats_ops = {
+	.show = kfd_procfs_stats_show,
+};
+
+static struct attribute *procfs_stats_attrs[] = {
+	NULL
+};
+
+static struct kobj_type procfs_stats_type = {
+	.sysfs_ops = &procfs_stats_ops,
+	.default_attrs = procfs_stats_attrs,
+};
+
 int kfd_procfs_add_queue(struct queue *q)
 {
 	struct kfd_process *proc;
@@ -417,6 +450,58 @@ static int kfd_sysfs_create_file(struct kfd_process *p, struct attribute *attr,
 	return ret;
 }
 
+static int kfd_procfs_add_sysfs_stats(struct kfd_process *p)
+{
+	int ret = 0;
+	struct kfd_process_device *pdd;
+	char stats_dir_filename[MAX_SYSFS_FILENAME_LEN];
+
+	if (!p)
+		return -EINVAL;
+
+	if (!p->kobj)
+		return -EFAULT;
+
+	/*
+	 * Create sysfs files for each GPU:
+	 * - proc/<pid>/stats_<gpuid>/
+	 * - proc/<pid>/stats_<gpuid>/evicted_ms
+	 */
+	list_for_each_entry(pdd, &p->per_device_data, per_device_list) {
+		struct kobject *kobj_stats;
+
+		snprintf(stats_dir_filename, MAX_SYSFS_FILENAME_LEN,
+				"stats_%u", pdd->dev->id);
+		kobj_stats = kfd_alloc_struct(kobj_stats);
+		if (!kobj_stats)
+			return -ENOMEM;
+
+		ret = kobject_init_and_add(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(kobj_stats);
+			goto err;
+		}
+
+		pdd->kobj_stats = kobj_stats;
+		pdd->attr_evict.name = "evicted_ms";
+		pdd->attr_evict.mode = KFD_SYSFS_FILE_MODE;
+		sysfs_attr_init(&pdd->attr_evict);
+		ret = sysfs_create_file(kobj_stats, &pdd->attr_evict);
+		if (ret)
+			pr_warn("Creating eviction stats for gpuid %d failed",
+					(int)pdd->dev->id);
+	}
+err:
+	return ret;
+}
+
+
 static int kfd_procfs_add_sysfs_files(struct kfd_process *p)
 {
 	int ret = 0;
@@ -660,6 +745,16 @@ struct kfd_process *kfd_create_process(struct file *filep)
 		if (!process->kobj_queues)
 			pr_warn("Creating KFD proc/queues folder failed");
 
+		ret = kfd_procfs_add_sysfs_stats(process);
+		if (ret)
+			pr_warn("Creating sysfs stats dir for pid %d failed",
+				(int)process->lead_thread->pid);
+
+		ret = kfd_procfs_add_sysfs_stats(process);
+		if (ret)
+			pr_warn("Creating sysfs stats dir for pid %d failed",
+				(int)process->lead_thread->pid);
+
 		ret = kfd_procfs_add_sysfs_files(process);
 		if (ret)
 			pr_warn("Creating sysfs usage file for pid %d failed",
@@ -816,6 +911,10 @@ static void kfd_process_wq_release(struct work_struct *work)
 		list_for_each_entry(pdd, &p->per_device_data, per_device_list) {
 			sysfs_remove_file(p->kobj, &pdd->attr_vram);
 			sysfs_remove_file(p->kobj, &pdd->attr_sdma);
+			sysfs_remove_file(p->kobj, &pdd->attr_evict);
+			kobject_del(pdd->kobj_stats);
+			kobject_put(pdd->kobj_stats);
+			pdd->kobj_stats = NULL;
 		}
 
 		kobject_del(p->kobj);
@@ -1125,6 +1224,7 @@ struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
 	pdd->runtime_inuse = false;
 	pdd->vram_usage = 0;
 	pdd->sdma_past_activity_counter = 0;
+	atomic64_set(&pdd->evict_duration_counter, 0);
 	list_add(&pdd->per_device_list, &p->per_device_data);
 
 	/* Init idr used for memory handle translation */
@@ -1488,6 +1588,7 @@ void kfd_suspend_all_processes(void)
 	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);