| Message ID | 20190926105644.16703-3-bigeasy@linutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/i915: Acquire locks with interrupts disabled | expand |
Quoting Sebastian Andrzej Siewior (2019-09-26 11:56:44) > The lockdep_assert_irqs_disabled() check is needless. The previous > lockdep_assert_held() check ensures that the lock is acquired and while > the lock is acquired lockdep also prints a warning if the interrupts are > not disabled if they have to be. > These IRQ-off asserts trigger on PREEMPT_RT because the locks become > sleeping locks and do not really disable interrupts. The intent was to document the entry points in were to be common dma-fence functions where irqs were expected to be off. Similarly for breadcrumbs to indicate that they were inner functions dealing with irq-sensitive locks that the caller had to disable irqs for. lockdep is not the clearest when it comes to explaining irq-inversions. > Remove lockdep_assert_irqs_disabled(). > > Reported-by: Clark Williams <williams@redhat.com> > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Given the context though, they are moot. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> -Chris
diff --git a/drivers/gpu/drm/i915/gt/intel_breadcrumbs.c b/drivers/gpu/drm/i915/gt/intel_breadcrumbs.c index f750375056de4..55317081d48b0 100644 --- a/drivers/gpu/drm/i915/gt/intel_breadcrumbs.c +++ b/drivers/gpu/drm/i915/gt/intel_breadcrumbs.c @@ -120,7 +120,6 @@ __dma_fence_signal__notify(struct dma_fence *fence, struct dma_fence_cb *cur, *tmp; lockdep_assert_held(fence->lock); - lockdep_assert_irqs_disabled(); list_for_each_entry_safe(cur, tmp, list, node) { INIT_LIST_HEAD(&cur->node); @@ -269,7 +268,6 @@ void intel_engine_fini_breadcrumbs(struct intel_engine_cs *engine) bool i915_request_enable_breadcrumb(struct i915_request *rq) { lockdep_assert_held(&rq->lock); - lockdep_assert_irqs_disabled(); if (test_bit(I915_FENCE_FLAG_ACTIVE, &rq->fence.flags)) { struct intel_breadcrumbs *b = &rq->engine->breadcrumbs; @@ -319,7 +317,6 @@ void i915_request_cancel_breadcrumb(struct i915_request *rq) struct intel_breadcrumbs *b = &rq->engine->breadcrumbs; lockdep_assert_held(&rq->lock); - lockdep_assert_irqs_disabled(); /* * We must wait for b->irq_lock so that we know the interrupt handler
The lockdep_assert_irqs_disabled() check is needless. The previous lockdep_assert_held() check ensures that the lock is acquired and while the lock is acquired lockdep also prints a warning if the interrupts are not disabled if they have to be. These IRQ-off asserts trigger on PREEMPT_RT because the locks become sleeping locks and do not really disable interrupts. Remove lockdep_assert_irqs_disabled(). Reported-by: Clark Williams <williams@redhat.com> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- drivers/gpu/drm/i915/gt/intel_breadcrumbs.c | 3 --- 1 file changed, 3 deletions(-)