drm/amdkfd: Check HIQ's MQD for queue preemption status

MEC firmware can silently fail the queue preemption request
without time out. In this case, HIQ's MQD's queue_doorbell_id
will be set. Check this field to see whether last queue preemption
was successful or not.
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Suggested-by: Jay Cornwall <Jay.Cornwall@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c

@@ -1393,6 +1393,7 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm,
 				uint32_t filter_param)
 {
 	int retval = 0;
+	struct mqd_manager *mqd_mgr;
 
 	if (!dqm->sched_running)
 		return 0;
@@ -1424,6 +1425,22 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm,
 		return retval;
 	}
 
+	/* In the current MEC firmware implementation, if compute queue
+	 * doesn't response to the preemption request in time, HIQ will
+	 * abandon the unmap request without returning any timeout error
+	 * to driver. Instead, MEC firmware will log the doorbell of the
+	 * unresponding compute queue to HIQ.MQD.queue_doorbell_id fields.
+	 * To make sure the queue unmap was successful, driver need to
+	 * check those fields
+	 */
+	mqd_mgr = dqm->mqd_mgrs[KFD_MQD_TYPE_HIQ];
+	if (mqd_mgr->read_doorbell_id(dqm->packets.priv_queue->queue->mqd)) {
+		pr_err("HIQ MQD's queue_doorbell_id0 is not 0, Queue preemption time out\n");
+		while (halt_if_hws_hang)
+			schedule();
+		return -ETIME;
+	}
+
 	pm_release_ib(&dqm->packets);
 	dqm->active_runlist = false;
 

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.h

@@ -101,6 +101,7 @@ struct mqd_manager {
 #if defined(CONFIG_DEBUG_FS)
 	int	(*debugfs_show_mqd)(struct seq_file *m, void *data);
 #endif
+	uint32_t (*read_doorbell_id)(void *mqd);
 
 	struct mutex	mqd_mutex;
 	struct kfd_dev	*dev;

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_cik.c

@@ -226,6 +226,13 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 	__update_mqd(mm, mqd, q, 1);
 }
 
+static uint32_t read_doorbell_id(void *mqd)
+{
+	struct cik_mqd *m = (struct cik_mqd *)mqd;
+
+	return m->queue_doorbell_id0;
+}
+
 static void update_mqd_hawaii(struct mqd_manager *mm, void *mqd,
 			struct queue_properties *q)
 {
@@ -398,6 +405,7 @@ struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
+		mqd->read_doorbell_id = read_doorbell_id;
 		break;
 	case KFD_MQD_TYPE_DIQ:
 		mqd->allocate_mqd = allocate_mqd;

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c

@@ -224,6 +224,13 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 	q->is_active = QUEUE_IS_ACTIVE(*q);
 }
 
+static uint32_t read_doorbell_id(void *mqd)
+{
+	struct v10_compute_mqd *m = (struct v10_compute_mqd *)mqd;
+
+	return m->queue_doorbell_id0;
+}
+
 static int destroy_mqd(struct mqd_manager *mm, void *mqd,
 		       enum kfd_preempt_type type,
 		       unsigned int timeout, uint32_t pipe_id,
@@ -425,6 +432,7 @@ struct mqd_manager *mqd_manager_init_v10(enum KFD_MQD_TYPE type,
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
+		mqd->read_doorbell_id = read_doorbell_id;
 		pr_debug("%s@%i\n", __func__, __LINE__);
 		break;
 	case KFD_MQD_TYPE_DIQ:

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c

@@ -276,6 +276,13 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 }
 
 
+static uint32_t read_doorbell_id(void *mqd)
+{
+	struct v9_mqd *m = (struct v9_mqd *)mqd;
+
+	return m->queue_doorbell_id0;
+}
+
 static int destroy_mqd(struct mqd_manager *mm, void *mqd,
 			enum kfd_preempt_type type,
 			unsigned int timeout, uint32_t pipe_id,
@@ -477,6 +484,7 @@ struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
+		mqd->read_doorbell_id = read_doorbell_id;
 		break;
 	case KFD_MQD_TYPE_DIQ:
 		mqd->allocate_mqd = allocate_mqd;

drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c

@@ -243,6 +243,13 @@ static void update_mqd(struct mqd_manager *mm, void *mqd,
 	__update_mqd(mm, mqd, q, MTYPE_CC, 1);
 }
 
+static uint32_t read_doorbell_id(void *mqd)
+{
+	struct vi_mqd *m = (struct vi_mqd *)mqd;
+
+	return m->queue_doorbell_id0;
+}
+
 static void update_mqd_tonga(struct mqd_manager *mm, void *mqd,
 			struct queue_properties *q)
 {
@@ -446,6 +453,7 @@ struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
 #if defined(CONFIG_DEBUG_FS)
 		mqd->debugfs_show_mqd = debugfs_show_mqd;
 #endif
+		mqd->read_doorbell_id = read_doorbell_id;
 		break;
 	case KFD_MQD_TYPE_DIQ:
 		mqd->allocate_mqd = allocate_mqd;

drivers/gpu/drm/amd/include/vi_structs.h

@@ -397,22 +397,22 @@ struct vi_mqd {
 	uint32_t reserved60;
 	uint32_t reserved61;
 	uint32_t reserved62;
-	uint32_t reserved63;
-	uint32_t reserved64;
-	uint32_t reserved65;
-	uint32_t reserved66;
-	uint32_t reserved67;
-	uint32_t reserved68;
-	uint32_t reserved69;
-	uint32_t reserved70;
-	uint32_t reserved71;
-	uint32_t reserved72;
-	uint32_t reserved73;
-	uint32_t reserved74;
-	uint32_t reserved75;
-	uint32_t reserved76;
-	uint32_t reserved77;
-	uint32_t reserved78;
+	uint32_t queue_doorbell_id0;
+	uint32_t queue_doorbell_id1;
+	uint32_t queue_doorbell_id2;
+	uint32_t queue_doorbell_id3;
+	uint32_t queue_doorbell_id4;
+	uint32_t queue_doorbell_id5;
+	uint32_t queue_doorbell_id6;
+	uint32_t queue_doorbell_id7;
+	uint32_t queue_doorbell_id8;
+	uint32_t queue_doorbell_id9;
+	uint32_t queue_doorbell_id10;
+	uint32_t queue_doorbell_id11;
+	uint32_t queue_doorbell_id12;
+	uint32_t queue_doorbell_id13;
+	uint32_t queue_doorbell_id14;
+	uint32_t queue_doorbell_id15;
 	uint32_t reserved_t[256];
 };