[Intel-gfx] [RFC 29/39] drm/i915: Added trace points to scheduler
John.C.Harrison at Intel.com
John.C.Harrison at Intel.com
Fri Jul 17 07:33:38 PDT 2015
From: John Harrison <John.C.Harrison at Intel.com>
Added trace points to the scheduler to track all the various events, node state
transitions and other interesting things that occur.
Change-Id: I9886390cfc7897bc1faf50a104bc651d8baed8a5
For: VIZ-1587
Signed-off-by: John Harrison <John.C.Harrison at Intel.com>
---
drivers/gpu/drm/i915/i915_gem_execbuffer.c | 2 +
drivers/gpu/drm/i915/i915_scheduler.c | 34 ++++-
drivers/gpu/drm/i915/i915_trace.h | 208 +++++++++++++++++++++++++++++
drivers/gpu/drm/i915/intel_lrc.c | 2 +
4 files changed, 244 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 1325b19..f90a2c8 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -1291,6 +1291,8 @@ i915_gem_ringbuffer_submission(struct i915_execbuffer_params *params,
i915_gem_execbuffer_move_to_active(vmas, params->request);
+ trace_i915_gem_ring_queue(ring, params);
+
qe = container_of(params, typeof(*qe), params);
ret = i915_scheduler_queue_execbuffer(qe);
if (ret)
diff --git a/drivers/gpu/drm/i915/i915_scheduler.c b/drivers/gpu/drm/i915/i915_scheduler.c
index 66dbc20..408bedc 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.c
+++ b/drivers/gpu/drm/i915/i915_scheduler.c
@@ -101,6 +101,8 @@ int i915_scheduler_queue_execbuffer(struct i915_scheduler_queue_entry *qe)
qe->scheduler_index = scheduler->index++;
+ trace_i915_scheduler_queue(qe->params.ring, qe);
+
WARN_ON(qe->params.fence_wait &&
(atomic_read(&qe->params.fence_wait->status) == 0));
@@ -253,6 +255,9 @@ int i915_scheduler_queue_execbuffer(struct i915_scheduler_queue_entry *qe)
not_flying = i915_scheduler_count_flying(scheduler, ring) <
scheduler->min_flying;
+ trace_i915_scheduler_queue(ring, node);
+ trace_i915_scheduler_node_state_change(ring, node);
+
spin_unlock_irqrestore(&scheduler->lock, flags);
if (not_flying)
@@ -280,6 +285,9 @@ static int i915_scheduler_fly_node(struct i915_scheduler_queue_entry *node)
node->status = i915_sqs_flying;
+ trace_i915_scheduler_fly(ring, node);
+ trace_i915_scheduler_node_state_change(ring, node);
+
if (!(scheduler->flags[ring->id] & i915_sf_interrupts_enabled)) {
bool success = true;
@@ -344,6 +352,8 @@ static void i915_scheduler_node_requeue(struct i915_scheduler_queue_entry *node)
BUG_ON(!I915_SQS_IS_FLYING(node));
node->status = i915_sqs_queued;
+ trace_i915_scheduler_unfly(node->params.ring, node);
+ trace_i915_scheduler_node_state_change(node->params.ring, node);
}
/* Give up on a popped node completely. For example, because it is causing the
@@ -354,6 +364,8 @@ static void i915_scheduler_node_kill(struct i915_scheduler_queue_entry *node)
BUG_ON(!I915_SQS_IS_FLYING(node));
node->status = i915_sqs_dead;
+ trace_i915_scheduler_unfly(node->params.ring, node);
+ trace_i915_scheduler_node_state_change(node->params.ring, node);
}
/* Abandon a queued node completely. For example because the driver is being
@@ -365,6 +377,7 @@ static void i915_scheduler_node_kill_queued(struct i915_scheduler_queue_entry *n
BUG_ON(!I915_SQS_IS_QUEUED(node));
node->status = i915_sqs_dead;
+ trace_i915_scheduler_node_state_change(node->params.ring, node);
}
/* The system is toast. Terminate all nodes with extreme prejudice. */
@@ -429,8 +442,10 @@ static void i915_scheduler_seqno_complete(struct intel_engine_cs *ring, uint32_t
* if a completed entry is found then there is no need to scan further.
*/
list_for_each_entry(node, &scheduler->node_queue[ring->id], link) {
- if (I915_SQS_IS_COMPLETE(node))
+ if (I915_SQS_IS_COMPLETE(node)) {
+ trace_i915_scheduler_landing(ring, seqno, node);
return;
+ }
if (seqno == node->params.request->seqno)
break;
@@ -441,8 +456,12 @@ static void i915_scheduler_seqno_complete(struct intel_engine_cs *ring, uint32_t
* like cache flushes and page flips. So don't complain about if
* no node was found.
*/
- if (&node->link == &scheduler->node_queue[ring->id])
+ if (&node->link == &scheduler->node_queue[ring->id]) {
+ trace_i915_scheduler_landing(ring, seqno, NULL);
return;
+ }
+
+ trace_i915_scheduler_landing(ring, seqno, node);
WARN_ON(!I915_SQS_IS_FLYING(node));
@@ -457,6 +476,7 @@ static void i915_scheduler_seqno_complete(struct intel_engine_cs *ring, uint32_t
/* Node was in flight so mark it as complete. */
node->status = i915_sqs_complete;
+ trace_i915_scheduler_node_state_change(ring, node);
got_changes = true;
}
@@ -481,6 +501,8 @@ int i915_scheduler_handle_irq(struct intel_engine_cs *ring)
seqno = ring->get_seqno(ring, false);
+ trace_i915_scheduler_irq(ring, seqno);
+
if (i915.scheduler_override & i915_so_direct_submit)
return 0;
@@ -625,6 +647,8 @@ static int i915_scheduler_remove(struct intel_engine_cs *ring)
/* Launch more packets now? */
do_submit = (queued > 0) && (flying < scheduler->min_flying);
+ trace_i915_scheduler_remove(ring, min_seqno, do_submit);
+
spin_unlock_irqrestore(&scheduler->lock, flags);
if (do_submit)
@@ -634,6 +658,8 @@ static int i915_scheduler_remove(struct intel_engine_cs *ring)
node = list_first_entry(&remove, typeof(*node), link);
list_del(&node->link);
+ trace_i915_scheduler_destroy(ring, node);
+
#ifdef CONFIG_SYNC
if (node->params.fence_wait)
sync_fence_put(node->params.fence_wait);
@@ -1016,6 +1042,8 @@ static int i915_scheduler_pop_from_queue_locked(struct intel_engine_cs *ring,
INIT_LIST_HEAD(&best->link);
best->status = i915_sqs_popped;
+ trace_i915_scheduler_node_state_change(ring, best);
+
ret = 0;
} else {
/* Can only get here if:
@@ -1073,6 +1101,8 @@ static int i915_scheduler_pop_from_queue_locked(struct intel_engine_cs *ring,
#endif
}
+ trace_i915_scheduler_pop_from_queue(ring, best);
+
*pop_node = best;
return ret;
}
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 796c630..8774192 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -9,6 +9,7 @@
#include "i915_drv.h"
#include "intel_drv.h"
#include "intel_ringbuffer.h"
+#include "i915_scheduler.h"
#undef TRACE_SYSTEM
#define TRACE_SYSTEM i915
@@ -786,6 +787,213 @@ TRACE_EVENT(switch_mm,
__entry->dev, __entry->ring, __entry->to, __entry->vm)
);
+TRACE_EVENT(i915_scheduler_queue,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node ? node->params.request->seqno : 0;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d",
+ __entry->ring, __entry->uniq, __entry->seqno)
+);
+
+TRACE_EVENT(i915_scheduler_fly,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node ? node->params.request->seqno : 0;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d",
+ __entry->ring, __entry->uniq, __entry->seqno)
+);
+
+TRACE_EVENT(i915_scheduler_unfly,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node ? node->params.request->seqno : 0;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d",
+ __entry->ring, __entry->uniq, __entry->seqno)
+);
+
+TRACE_EVENT(i915_scheduler_landing,
+ TP_PROTO(struct intel_engine_cs *ring, u32 seqno,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, seqno, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ __field(u32, status)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = seqno;
+ __entry->status = node ? node->status : ~0U;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d, status=%d",
+ __entry->ring, __entry->uniq, __entry->seqno, __entry->status)
+);
+
+TRACE_EVENT(i915_scheduler_remove,
+ TP_PROTO(struct intel_engine_cs *ring,
+ u32 min_seqno, bool do_submit),
+ TP_ARGS(ring, min_seqno, do_submit),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, min_seqno)
+ __field(bool, do_submit)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->min_seqno = min_seqno;
+ __entry->do_submit = do_submit;
+ ),
+
+ TP_printk("ring=%d, min_seqno = %d, do_submit=%d",
+ __entry->ring, __entry->min_seqno, __entry->do_submit)
+);
+
+TRACE_EVENT(i915_scheduler_destroy,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node ? node->params.request->seqno : 0;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d",
+ __entry->ring, __entry->uniq, __entry->seqno)
+);
+
+TRACE_EVENT(i915_scheduler_pop_from_queue,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node ? node->params.request->seqno : 0;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d",
+ __entry->ring, __entry->uniq, __entry->seqno)
+);
+
+TRACE_EVENT(i915_scheduler_node_state_change,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_scheduler_queue_entry *node),
+ TP_ARGS(ring, node),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, uniq)
+ __field(u32, seqno)
+ __field(u32, status)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->uniq = node ? node->params.request->uniq : 0;
+ __entry->seqno = node->params.request->seqno;
+ __entry->status = node->status;
+ ),
+
+ TP_printk("ring=%d, uniq=%d, seqno=%d, status=%d",
+ __entry->ring, __entry->uniq, __entry->seqno, __entry->status)
+);
+
+TRACE_EVENT(i915_scheduler_irq,
+ TP_PROTO(struct intel_engine_cs *ring, uint32_t seqno),
+ TP_ARGS(ring, seqno),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->seqno = seqno;
+ ),
+
+ TP_printk("ring=%d, seqno=%d", __entry->ring, __entry->seqno)
+);
+
+TRACE_EVENT(i915_gem_ring_queue,
+ TP_PROTO(struct intel_engine_cs *ring,
+ struct i915_execbuffer_params *params),
+ TP_ARGS(ring, params),
+
+ TP_STRUCT__entry(
+ __field(u32, ring)
+ __field(u32, seqno)
+ ),
+
+ TP_fast_assign(
+ __entry->ring = ring->id;
+ __entry->seqno = params->request->seqno;
+ ),
+
+ TP_printk("ring=%d, seqno=%d", __entry->ring, __entry->seqno)
+);
+
#endif /* _I915_TRACE_H_ */
/* This part must be outside protection */
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index d3e7399..41dca2a 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -886,6 +886,8 @@ int intel_execlists_submission(struct i915_execbuffer_params *params,
i915_gem_execbuffer_move_to_active(vmas, params->request);
+ trace_i915_gem_ring_queue(ring, params);
+
qe = container_of(params, typeof(*qe), params);
ret = i915_scheduler_queue_execbuffer(qe);
if (ret)
--
1.9.1
More information about the Intel-gfx
mailing list