90 lines
3.2 KiB
Diff
90 lines
3.2 KiB
Diff
|
From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
|
||
|
|
Date: Wed, 8 Sep 2021 19:03:41 +0200
|
||
|
|
Subject: [PATCH 5/8] drm/i915/gt: Use spin_lock_irq() instead of
|
||
|
|
local_irq_disable() + spin_lock()
|
||
|
|
Origin: https://www.kernel.org/pub/linux/kernel/projects/rt/6.11/older/patches-6.11-rt7.tar.xz
|
||
|
|
|
||
|
|
execlists_dequeue() is invoked from a function which uses
|
||
|
|
local_irq_disable() to disable interrupts so the spin_lock() behaves
|
||
|
|
like spin_lock_irq().
|
||
|
|
This breaks PREEMPT_RT because local_irq_disable() + spin_lock() is not
|
||
|
|
the same as spin_lock_irq().
|
||
|
|
|
||
|
|
execlists_dequeue_irq() and execlists_dequeue() has each one caller
|
||
|
|
only. If intel_engine_cs::active::lock is acquired and released with the
|
||
|
|
_irq suffix then it behaves almost as if execlists_dequeue() would be
|
||
|
|
invoked with disabled interrupts. The difference is the last part of the
|
||
|
|
function which is then invoked with enabled interrupts.
|
||
|
|
I can't tell if this makes a difference. From looking at it, it might
|
||
|
|
work to move the last unlock at the end of the function as I didn't find
|
||
|
|
anything that would acquire the lock again.
|
||
|
|
|
||
|
|
Reported-by: Clark Williams <williams@redhat.com>
|
||
|
|
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
|
||
|
|
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
|
||
|
|
---
|
||
|
|
drivers/gpu/drm/i915/gt/intel_execlists_submission.c | 17 +++++------------
|
||
|
|
1 file changed, 5 insertions(+), 12 deletions(-)
|
||
|
|
|
||
|
|
--- a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
|
||
|
|
+++ b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
|
||
|
|
@@ -1303,7 +1303,7 @@ static void execlists_dequeue(struct int
|
||
|
|
* and context switches) submission.
|
||
|
|
*/
|
||
|
|
|
||
|
|
- spin_lock(&sched_engine->lock);
|
||
|
|
+ spin_lock_irq(&sched_engine->lock);
|
||
|
|
|
||
|
|
/*
|
||
|
|
* If the queue is higher priority than the last
|
||
|
|
@@ -1403,7 +1403,7 @@ static void execlists_dequeue(struct int
|
||
|
|
* Even if ELSP[1] is occupied and not worthy
|
||
|
|
* of timeslices, our queue might be.
|
||
|
|
*/
|
||
|
|
- spin_unlock(&sched_engine->lock);
|
||
|
|
+ spin_unlock_irq(&sched_engine->lock);
|
||
|
|
return;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
@@ -1429,7 +1429,7 @@ static void execlists_dequeue(struct int
|
||
|
|
|
||
|
|
if (last && !can_merge_rq(last, rq)) {
|
||
|
|
spin_unlock(&ve->base.sched_engine->lock);
|
||
|
|
- spin_unlock(&engine->sched_engine->lock);
|
||
|
|
+ spin_unlock_irq(&engine->sched_engine->lock);
|
||
|
|
return; /* leave this for another sibling */
|
||
|
|
}
|
||
|
|
|
||
|
|
@@ -1591,7 +1591,7 @@ static void execlists_dequeue(struct int
|
||
|
|
*/
|
||
|
|
sched_engine->queue_priority_hint = queue_prio(sched_engine);
|
||
|
|
i915_sched_engine_reset_on_empty(sched_engine);
|
||
|
|
- spin_unlock(&sched_engine->lock);
|
||
|
|
+ spin_unlock_irq(&sched_engine->lock);
|
||
|
|
|
||
|
|
/*
|
||
|
|
* We can skip poking the HW if we ended up with exactly the same set
|
||
|
|
@@ -1617,13 +1617,6 @@ static void execlists_dequeue(struct int
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
-static void execlists_dequeue_irq(struct intel_engine_cs *engine)
|
||
|
|
-{
|
||
|
|
- local_irq_disable(); /* Suspend interrupts across request submission */
|
||
|
|
- execlists_dequeue(engine);
|
||
|
|
- local_irq_enable(); /* flush irq_work (e.g. breadcrumb enabling) */
|
||
|
|
-}
|
||
|
|
-
|
||
|
|
static void clear_ports(struct i915_request **ports, int count)
|
||
|
|
{
|
||
|
|
memset_p((void **)ports, NULL, count);
|
||
|
|
@@ -2478,7 +2471,7 @@ static void execlists_submission_tasklet
|
||
|
|
}
|
||
|
|
|
||
|
|
if (!engine->execlists.pending[0]) {
|
||
|
|
- execlists_dequeue_irq(engine);
|
||
|
|
+ execlists_dequeue(engine);
|
||
|
|
start_timeslice(engine);
|
||
|
|
}
|
||
|
|
|