Skip to contentSkip to conversation

drm/i915: Don't need a timer to wake us up

PATCH In review Tvrtko Ursulin 2 participants 5 messages Nov 26, 2015 – Nov 27, 2015

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: Don't need a timer to wake us up

From: Tvrtko Ursulin <[email protected]>

Looks like the sleeping loop in __i915_wait_request can be simplified by using io_schedule_timeout instead of setting up and destroying a timer.

---
 drivers/gpu/drm/i915/i915_gem.c | 28 ++++++++--------------------
 1 file changed, 8 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +8 -20
index 33adc8f8ab20..efb3eb8b4a5e 100644
@@ -1135,11 +1135,6 @@ i915_gem_check_wedge(struct i915_gpu_error *error,
 	return 0;
 }
 
-static void fake_irq(unsigned long data)
-{
-	wake_up_process((struct task_struct *)data);
-}
-
 static bool missed_irq(struct drm_i915_private *dev_priv,
 		       struct intel_engine_cs *ring)
 {
@@ -1231,7 +1226,7 @@ int __i915_wait_request(struct drm_i915_gem_request *req,
 	}
 
 	for (;;) {
-		struct timer_list timer;
+		long sched_timeout;
 
 		prepare_to_wait(&ring->irq_queue, &wait,
 				interruptible ? TASK_INTERRUPTIBLE : TASK_UNINTERRUPTIBLE);
@@ -1262,21 +1257,14 @@ int __i915_wait_request(struct drm_i915_gem_request *req,
 			break;
 		}
 
-		timer.function = NULL;
-		if (timeout || missed_irq(dev_priv, ring)) {
-			unsigned long expire;
-
-			setup_timer_on_stack(&timer, fake_irq, (unsigned long)current);
-			expire = missed_irq(dev_priv, ring) ? jiffies + 1 : timeout_expire;
-			mod_timer(&timer, expire);
-		}
-
-		io_schedule();
+		if (timeout)
+			sched_timeout = timeout_expire - jiffies;
+		else if (missed_irq(dev_priv, ring))
+			sched_timeout = 1;
+		else
+			sched_timeout = MAX_SCHEDULE_TIMEOUT;
 
-		if (timer.function) {
-			del_singleshot_timer_sync(&timer);
-			destroy_timer_on_stack(&timer);
-		}
+		io_schedule_timeout(sched_timeout);
 	}
 	if (!irq_test_in_progress)
 		ring->irq_put(ring);
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw
Quoted 5 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Looks like the sleeping loop in __i915_wait_request can be simplified by using io_schedule_timeout instead of setting up and destroying a timer.

Simplified by duplicating code? I liked the explicit handling for its obviousness and simplicity.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#2 lore raw
Quoted 9 lines · Chris Wilson
Quoted 5 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Looks like the sleeping loop in __i915_wait_request can be simplified by using io_schedule_timeout instead of setting up and destroying a timer.

Simplified by duplicating code? I liked the explicit handling for its obviousness and simplicity.

To be slightl more gracious, after we eliminate the irq handling from this function, the waiter does look like this, albeit I still think it is cleaner to keep the timeout handling distinct from the indefinite waits.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw
Quoted 8 lines · Chris Wilson
Quoted 7 lines · Chris Wilson
Quoted 5 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Looks like the sleeping loop in __i915_wait_request can be simplified by using io_schedule_timeout instead of setting up and destroying a timer.

Quoted 6 lines
Quoted 2 lines

Simplified by duplicating code? I liked the explicit handling for its obviousness and simplicity.

To be slightl more gracious, after we eliminate the irq handling from this function, the waiter does look like this, albeit I still think it is cleaner to keep the timeout handling distinct from the indefinite waits.

Duplicating code, fake_irq obviousness and simplicity?!?

To start with:

 drivers/gpu/drm/i915/i915_gem.c | 28 ++++++++--------------------
 1 file changed, 8 insertions(+), 20 deletions(-)

So less code, less binary text, less stack space, less kernel API usage so also fewer cycles to execute (not that it matters here but hey), and easier to understand logic.

So really I don't know what you don't like about it.

Regards,

Tvrtko
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw
Quoted 10 lines · Tvrtko Ursulin
Quoted 8 lines · Chris Wilson
Quoted 7 lines · Chris Wilson
Quoted 5 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Looks like the sleeping loop in __i915_wait_request can be simplified by using io_schedule_timeout instead of setting up and destroying a timer.

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Simplified by duplicating code? I liked the explicit handling for its obviousness and simplicity.

To be slightl more gracious, after we eliminate the irq handling from this function, the waiter does look like this, albeit I still think it is cleaner to keep the timeout handling distinct from the indefinite waits.

Duplicating code, fake_irq obviousness and simplicity?!?

Bah, the code was duplicated in Feb 2015. fake_irq is obvious and a precise description of what we are doing and why, that's why I like it. It is much easier to convert to other wakeups (say hrtimer, or napi style). This code needs to be changed to solve the thundering herd problem, if you want to improve this code, keep reviewing the prequisites.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx