drm/i915: Move the irq_counter inside the spinlock

Rather than have multiple locked instructions inside the notify_ring()
irq handler, move them inside the spinlock and reduce their intrinsic
locking.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180627201304.15817-3-chris@chris-wilson.co.uk

diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 4f137eeaf3951a42f150e4612623422a942b7a2f..c81b4c1877cce4030f6af7955f2215f1763e4cef 100644 (file)
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1162,8 +1162,6 @@ static void notify_ring(struct intel_engine_cs *engine)
        if (unlikely(!engine->breadcrumbs.irq_armed))
                return;
 
-       atomic_inc(&engine->irq_count);
-
        rcu_read_lock();
 
        spin_lock(&engine->breadcrumbs.irq_lock);
@@ -1198,6 +1196,8 @@ static void notify_ring(struct intel_engine_cs *engine)
                                tsk = wait->tsk;
                        }
                }
+
+               engine->breadcrumbs.irq_count++;
        } else {
                if (engine->breadcrumbs.irq_armed)
                        __intel_engine_disarm_breadcrumbs(engine);

diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index e1dbb544046fef9c6b10b28a05922f10a155f9dc..39b296878ba2872a47da64bf2190a97777311f39 100644 (file)
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -1196,7 +1196,7 @@ static bool __i915_spin_request(const struct i915_request *rq,
         * takes to sleep on a request, on the order of a microsecond.
         */
 
-       irq = atomic_read(&engine->irq_count);
+       irq = READ_ONCE(engine->breadcrumbs.irq_count);
        timeout_us += local_clock_us(&cpu);
        do {
                if (i915_seqno_passed(intel_engine_get_seqno(engine), seqno))
@@ -1208,7 +1208,7 @@ static bool __i915_spin_request(const struct i915_request *rq,
                 * assume we won't see one in the near future but require
                 * the engine->seqno_barrier() to fixup coherency.
                 */
-               if (atomic_read(&engine->irq_count) != irq)
+               if (READ_ONCE(engine->breadcrumbs.irq_count) != irq)
                        break;
 
                if (signal_pending_state(state, current))

diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index 86a987b8ac667b6988a64d446b76ee594046005d..1db6ba7d926ee3b27b0c00e9bf82f956173e8b59 100644 (file)
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -98,12 +98,14 @@ static void intel_breadcrumbs_hangcheck(struct timer_list *t)
        struct intel_engine_cs *engine =
                from_timer(engine, t, breadcrumbs.hangcheck);
        struct intel_breadcrumbs *b = &engine->breadcrumbs;
+       unsigned int irq_count;
 
        if (!b->irq_armed)
                return;
 
-       if (b->hangcheck_interrupts != atomic_read(&engine->irq_count)) {
-               b->hangcheck_interrupts = atomic_read(&engine->irq_count);
+       irq_count = READ_ONCE(b->irq_count);
+       if (b->hangcheck_interrupts != irq_count) {
+               b->hangcheck_interrupts = irq_count;
                mod_timer(&b->hangcheck, wait_timeout());
                return;
        }
@@ -272,13 +274,14 @@ static bool use_fake_irq(const struct intel_breadcrumbs *b)
        if (!test_bit(engine->id, &engine->i915->gpu_error.missed_irq_rings))
                return false;
 
-       /* Only start with the heavy weight fake irq timer if we have not
+       /*
+        * Only start with the heavy weight fake irq timer if we have not
         * seen any interrupts since enabling it the first time. If the
         * interrupts are still arriving, it means we made a mistake in our
         * engine->seqno_barrier(), a timing error that should be transient
         * and unlikely to reoccur.
         */
-       return atomic_read(&engine->irq_count) == b->hangcheck_interrupts;
+       return READ_ONCE(b->irq_count) == b->hangcheck_interrupts;
 }
 
 static void enable_fake_irq(struct intel_breadcrumbs *b)

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index a0bc7a8222b4091509e2f04e7ececbe0048c98d4..44ac90ec540c21fadc6749ed2d7e4b05740480bb 100644 (file)
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -345,7 +345,6 @@ struct intel_engine_cs {
        struct drm_i915_gem_object *default_state;
        void *pinned_default_state;
 
-       atomic_t irq_count;
        unsigned long irq_posted;
 #define ENGINE_IRQ_BREADCRUMB 0
 #define ENGINE_IRQ_EXECLIST 1
@@ -380,6 +379,7 @@ struct intel_engine_cs {
 
                unsigned int hangcheck_interrupts;
                unsigned int irq_enabled;
+               unsigned int irq_count;
 
                bool irq_armed : 1;
                I915_SELFTEST_DECLARE(bool mock : 1);