drm/i915/execlists: Switch to rb_root_cached

The kernel recently gained an augmented rbtree with the purpose of
cacheing the leftmost element of the rbtree, a frequent optimisation to
avoid calls to rb_first() which is also employed by the
execlists->queue. Switch from our open-coded cache to the library.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180629075348.27358-9-chris@chris-wilson.co.uk

diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 0ac497275a51303b0c73a6307463f8f3b68fc890..220050107c486661823aaea179c8ce091b97a0ba 100644 (file)
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -467,8 +467,7 @@ static void intel_engine_init_execlist(struct intel_engine_cs *engine)
        GEM_BUG_ON(execlists_num_ports(execlists) > EXECLIST_MAX_PORTS);
 
        execlists->queue_priority = INT_MIN;
-       execlists->queue = RB_ROOT;
-       execlists->first = NULL;
+       execlists->queue = RB_ROOT_CACHED;
 }
 
 /**
@@ -1004,7 +1003,7 @@ bool intel_engine_is_idle(struct intel_engine_cs *engine)
        }
 
        /* ELSP is empty, but there are ready requests? E.g. after reset */
-       if (READ_ONCE(engine->execlists.first))
+       if (!RB_EMPTY_ROOT(&engine->execlists.queue.rb_root))
                return false;
 
        /* Ring stopped? */
@@ -1540,7 +1539,7 @@ void intel_engine_dump(struct intel_engine_cs *engine,
        last = NULL;
        count = 0;
        drm_printf(m, "\t\tQueue priority: %d\n", execlists->queue_priority);
-       for (rb = execlists->first; rb; rb = rb_next(rb)) {
+       for (rb = rb_first_cached(&execlists->queue); rb; rb = rb_next(rb)) {
                struct i915_priolist *p =
                        rb_entry(rb, typeof(*p), node);
 

diff --git a/drivers/gpu/drm/i915/intel_guc_submission.c b/drivers/gpu/drm/i915/intel_guc_submission.c
index f3945258fe1b53c1723de5a584caa66522aab774..3952656f4c9a3322099d338163b2a21596777636 100644 (file)
--- a/drivers/gpu/drm/i915/intel_guc_submission.c
+++ b/drivers/gpu/drm/i915/intel_guc_submission.c
@@ -695,9 +695,6 @@ static bool __guc_dequeue(struct intel_engine_cs *engine)
 
        lockdep_assert_held(&engine->timeline.lock);
 
-       rb = execlists->first;
-       GEM_BUG_ON(rb_first(&execlists->queue) != rb);
-
        if (port_isset(port)) {
                if (intel_engine_has_preemption(engine)) {
                        struct guc_preempt_work *preempt_work =
@@ -719,7 +716,7 @@ static bool __guc_dequeue(struct intel_engine_cs *engine)
        }
        GEM_BUG_ON(port_isset(port));
 
-       while (rb) {
+       while ((rb = rb_first_cached(&execlists->queue))) {
                struct i915_priolist *p = to_priolist(rb);
                struct i915_request *rq, *rn;
 
@@ -744,15 +741,13 @@ static bool __guc_dequeue(struct intel_engine_cs *engine)
                        submit = true;
                }
 
-               rb = rb_next(rb);
-               rb_erase(&p->node, &execlists->queue);
+               rb_erase_cached(&p->node, &execlists->queue);
                INIT_LIST_HEAD(&p->requests);
                if (p->priority != I915_PRIORITY_NORMAL)
                        kmem_cache_free(engine->i915->priorities, p);
        }
 done:
        execlists->queue_priority = rb ? to_priolist(rb)->priority : INT_MIN;
-       execlists->first = rb;
        if (submit)
                port_assign(port, last);
        if (last)
@@ -761,7 +756,8 @@ done:
        /* We must always keep the beast fed if we have work piled up */
        GEM_BUG_ON(port_isset(execlists->port) &&
                   !execlists_is_active(execlists, EXECLISTS_ACTIVE_USER));
-       GEM_BUG_ON(execlists->first && !port_isset(execlists->port));
+       GEM_BUG_ON(rb_first_cached(&execlists->queue) &&
+                  !port_isset(execlists->port));
 
        return submit;
 }

diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index d937a21da2d80d9775b3bc98e21dc873b2ef7ed3..ad436d200758cb472e230c20dde8f63e078e8716 100644 (file)
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -273,7 +273,7 @@ lookup_priolist(struct intel_engine_cs *engine, int prio)
 find_priolist:
        /* most positive priority is scheduled first, equal priorities fifo */
        rb = NULL;
-       parent = &execlists->queue.rb_node;
+       parent = &execlists->queue.rb_root.rb_node;
        while (*parent) {
                rb = *parent;
                p = to_priolist(rb);
@@ -311,10 +311,7 @@ find_priolist:
        p->priority = prio;
        INIT_LIST_HEAD(&p->requests);
        rb_link_node(&p->node, rb, parent);
-       rb_insert_color(&p->node, &execlists->queue);
-
-       if (first)
-               execlists->first = &p->node;
+       rb_insert_color_cached(&p->node, &execlists->queue, first);
 
        return p;
 }
@@ -602,9 +599,6 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
         * and context switches) submission.
         */
 
-       rb = execlists->first;
-       GEM_BUG_ON(rb_first(&execlists->queue) != rb);
-
        if (last) {
                /*
                 * Don't resubmit or switch until all outstanding
@@ -666,7 +660,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
                last->tail = last->wa_tail;
        }
 
-       while (rb) {
+       while ((rb = rb_first_cached(&execlists->queue))) {
                struct i915_priolist *p = to_priolist(rb);
                struct i915_request *rq, *rn;
 
@@ -725,8 +719,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
                        submit = true;
                }
 
-               rb = rb_next(rb);
-               rb_erase(&p->node, &execlists->queue);
+               rb_erase_cached(&p->node, &execlists->queue);
                INIT_LIST_HEAD(&p->requests);
                if (p->priority != I915_PRIORITY_NORMAL)
                        kmem_cache_free(engine->i915->priorities, p);
@@ -752,14 +745,14 @@ done:
        execlists->queue_priority =
                port != execlists->port ? rq_prio(last) : INT_MIN;
 
-       execlists->first = rb;
        if (submit) {
                port_assign(port, last);
                execlists_submit_ports(engine);
        }
 
        /* We must always keep the beast fed if we have work piled up */
-       GEM_BUG_ON(execlists->first && !port_isset(execlists->port));
+       GEM_BUG_ON(rb_first_cached(&execlists->queue) &&
+                  !port_isset(execlists->port));
 
        /* Re-evaluate the executing context setup after each preemptive kick */
        if (last)
@@ -922,8 +915,7 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
        }
 
        /* Flush the queued requests to the timeline list (for retiring). */
-       rb = execlists->first;
-       while (rb) {
+       while ((rb = rb_first_cached(&execlists->queue))) {
                struct i915_priolist *p = to_priolist(rb);
 
                list_for_each_entry_safe(rq, rn, &p->requests, sched.link) {
@@ -933,8 +925,7 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
                        __i915_request_submit(rq);
                }
 
-               rb = rb_next(rb);
-               rb_erase(&p->node, &execlists->queue);
+               rb_erase_cached(&p->node, &execlists->queue);
                INIT_LIST_HEAD(&p->requests);
                if (p->priority != I915_PRIORITY_NORMAL)
                        kmem_cache_free(engine->i915->priorities, p);
@@ -943,8 +934,7 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
        /* Remaining _unready_ requests will be nop'ed when submitted */
 
        execlists->queue_priority = INT_MIN;
-       execlists->queue = RB_ROOT;
-       execlists->first = NULL;
+       execlists->queue = RB_ROOT_CACHED;
        GEM_BUG_ON(port_isset(execlists->port));
 
        spin_unlock_irqrestore(&engine->timeline.lock, flags);
@@ -1192,7 +1182,7 @@ static void execlists_submit_request(struct i915_request *request)
 
        queue_request(engine, &request->sched, rq_prio(request));
 
-       GEM_BUG_ON(!engine->execlists.first);
+       GEM_BUG_ON(RB_EMPTY_ROOT(&engine->execlists.queue.rb_root));
        GEM_BUG_ON(list_empty(&request->sched.link));
 
        submit_queue(engine, rq_prio(request));
@@ -2044,7 +2034,7 @@ static void execlists_reset_finish(struct intel_engine_cs *engine)
        struct intel_engine_execlists * const execlists = &engine->execlists;
 
        /* After a GPU reset, we may have requests to replay */
-       if (execlists->first)
+       if (!RB_EMPTY_ROOT(&execlists->queue.rb_root))
                tasklet_schedule(&execlists->tasklet);
 
        /*

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index ce6cc2a6cf7afb545c3aec13bb7fbaff61b1e4d7..d1eee08e5f6bb9d1334a48ad7c36d07990fea149 100644 (file)
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -292,12 +292,7 @@ struct intel_engine_execlists {
        /**
         * @queue: queue of requests, in priority lists
         */
-       struct rb_root queue;
-
-       /**
-        * @first: leftmost level in priority @queue
-        */
-       struct rb_node *first;
+       struct rb_root_cached queue;
 
        /**
         * @csb_read: control register for Context Switch buffer