drm/amdgpu: add userq create and destroy tracepoints

Add ftrace events around user queue creation and destruction to profile
queue setup and teardown latency.

Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 85724ec6aaf8499534eb42340e9e2e3aa6fed5de..0d5fb5daddfe832a4d6c9d911805e190e1508a7d 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -582,6 +582,64 @@ TRACE_EVENT(amdgpu_reset_reg_dumps,
                      __entry->value)
 );
 
+DECLARE_EVENT_CLASS(amdgpu_userq_queue,
+           TP_PROTO(struct amdgpu_usermode_queue *queue),
+           TP_ARGS(queue),
+           TP_STRUCT__entry(
+                            __field(void *, queue)
+                            __field(u64, doorbell_index)
+                            __field(int, queue_type)
+                            __field(int, state)
+                            __field(u32, xcp_id)
+                            ),
+           TP_fast_assign(
+                          __entry->queue = queue;
+                          __entry->doorbell_index = queue->doorbell_index;
+                          __entry->queue_type = queue->queue_type;
+                          __entry->state = queue->state;
+                          __entry->xcp_id = queue->xcp_id;
+                          ),
+           TP_printk("queue=%p, doorbell=%llu, type=%d, state=%d, xcp_id=%u",
+                     __entry->queue, __entry->doorbell_index,
+                     __entry->queue_type, __entry->state, __entry->xcp_id)
+);
+DEFINE_EVENT(amdgpu_userq_queue, amdgpu_userq_create_start,
+            TP_PROTO(struct amdgpu_usermode_queue *queue),
+            TP_ARGS(queue));
+DEFINE_EVENT(amdgpu_userq_queue, amdgpu_userq_destroy_start,
+            TP_PROTO(struct amdgpu_usermode_queue *queue),
+            TP_ARGS(queue));
+DECLARE_EVENT_CLASS(amdgpu_userq_queue_result,
+           TP_PROTO(struct amdgpu_usermode_queue *queue, int result),
+           TP_ARGS(queue, result),
+           TP_STRUCT__entry(
+                            __field(void *, queue)
+                            __field(u64, doorbell_index)
+                            __field(int, queue_type)
+                            __field(int, state)
+                            __field(u32, xcp_id)
+                            __field(int, result)
+                            ),
+           TP_fast_assign(
+                          __entry->queue = queue;
+                          __entry->doorbell_index = queue->doorbell_index;
+                          __entry->queue_type = queue->queue_type;
+                          __entry->state = queue->state;
+                          __entry->xcp_id = queue->xcp_id;
+                          __entry->result = result;
+                          ),
+           TP_printk("queue=%p, doorbell=%llu, type=%d, state=%d, xcp_id=%u, result=%d",
+                     __entry->queue, __entry->doorbell_index,
+                     __entry->queue_type, __entry->state,
+                     __entry->xcp_id, __entry->result)
+);
+DEFINE_EVENT(amdgpu_userq_queue_result, amdgpu_userq_create_end,
+            TP_PROTO(struct amdgpu_usermode_queue *queue, int result),
+            TP_ARGS(queue, result));
+DEFINE_EVENT(amdgpu_userq_queue_result, amdgpu_userq_destroy_end,
+            TP_PROTO(struct amdgpu_usermode_queue *queue, int result),
+            TP_ARGS(queue, result));
+
 #undef AMDGPU_JOB_GET_TIMELINE_NAME
 #endif
 

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
index 285dcadb9342bd8d492c869f9678412d53429833..b8aa2adc399eb078738ea5970347fbd07c344d4f 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
@@ -33,6 +33,7 @@
 #include "amdgpu_userq.h"
 #include "amdgpu_hmm.h"
 #include "amdgpu_userq_fence.h"
+#include "amdgpu_trace.h"
 
 u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev)
 {
@@ -505,6 +506,8 @@ amdgpu_userq_destroy(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_usermode_que
        const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
        int r = 0;
 
+       trace_amdgpu_userq_destroy_start(queue);
+
        cancel_delayed_work_sync(&uq_mgr->resume_work);
 
        /* Cancel any pending hang detection work and cleanup */
@@ -530,6 +533,7 @@ amdgpu_userq_destroy(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_usermode_que
        amdgpu_bo_unreserve(queue->db_obj.obj);
        amdgpu_bo_unref(&queue->db_obj.obj);
 
+       trace_amdgpu_userq_destroy_end(queue, r);
        kfree(queue);
 
        pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
@@ -671,6 +675,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
        }
 
        queue->doorbell_index = index;
+       trace_amdgpu_userq_create_start(queue);
        r = uq_funcs->mqd_create(queue, &args->in);
        if (r) {
                drm_file_err(uq_mgr->file, "Failed to create Queue\n");
@@ -694,6 +699,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
                r = amdgpu_userq_map_helper(queue);
                if (r) {
                        drm_file_err(uq_mgr->file, "Failed to map Queue\n");
+                       trace_amdgpu_userq_create_end(queue, r);
                        mutex_unlock(&uq_mgr->userq_mutex);
                        goto erase_doorbell;
                }
@@ -710,11 +716,13 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
                 * This drops the last reference which should take care of
                 * all cleanup.
                 */
+               trace_amdgpu_userq_create_end(queue, r);
                amdgpu_userq_put(queue);
                return r;
        }
 
        amdgpu_debugfs_userq_init(filp, queue, qid);
+       trace_amdgpu_userq_create_end(queue, 0);
        args->out.queue_id = qid;
        return 0;
 
@@ -730,6 +738,7 @@ clean_doorbell_bo:
 free_fence_drv:
        amdgpu_userq_fence_driver_free(queue);
 free_queue:
+       trace_amdgpu_userq_create_end(queue, r);
        kfree(queue);
 err_pm_runtime:
        pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);