[PATCH 52/52] bad-user-semaphore-round-robinning
Chris Wilson
chris at chris-wilson.co.uk
Tue Apr 23 15:32:58 UTC 2019
---
drivers/gpu/drm/i915/gt/intel_engine_types.h | 3 +
drivers/gpu/drm/i915/gt/intel_lrc.c | 92 ++++++++++++++++++--
drivers/gpu/drm/i915/i915_scheduler.c | 1 +
drivers/gpu/drm/i915/i915_scheduler_types.h | 1 +
4 files changed, 92 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h
index 1a3c8e2bf93e..d1f0de9ac738 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
@@ -12,6 +12,7 @@
#include <linux/kref.h>
#include <linux/list.h>
#include <linux/llist.h>
+#include <linux/timer.h>
#include <linux/types.h>
#include "i915_gem.h"
@@ -137,6 +138,8 @@ struct intel_engine_execlists {
*/
struct tasklet_struct tasklet;
+ struct timer_list timer;
+
/**
* @default_priolist: priority list for I915_PRIORITY_NORMAL
*/
diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
index 223a6754520a..3a0586913ef6 100644
--- a/drivers/gpu/drm/i915/gt/intel_lrc.c
+++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
@@ -719,6 +719,9 @@ static void execlists_submit_ports(struct intel_engine_cs *engine)
/* we need to manually load the submit queue */
if (execlists->ctrl_reg)
writel(EL_CTRL_LOAD, execlists->ctrl_reg);
+
+ if (execlists->pending[1])
+ mod_timer(&execlists->timer, jiffies + 1);
}
static bool ctx_single_port_submission(const struct intel_context *ce)
@@ -831,12 +834,12 @@ last_active(const struct intel_engine_execlists *execlists)
return *last;
}
-static bool execlists_dequeue(struct intel_engine_cs *engine)
+static bool execlists_dequeue(struct intel_engine_cs *engine,
+ struct i915_request *last)
{
struct intel_engine_execlists * const execlists = &engine->execlists;
struct i915_request **port = execlists->pending;
struct i915_request ** const last_port = port + execlists->port_mask;
- struct i915_request *last;
struct rb_node *rb;
bool submit = false;
@@ -890,7 +893,6 @@ static bool execlists_dequeue(struct intel_engine_cs *engine)
* i.e. we will retrigger preemption following the ack in case
* of trouble.
*/
- last = last_active(execlists);
if (last) {
if (need_preempt(engine, last, rb)) {
GEM_TRACE("%s: preempting last=%llx:%lld, prio=%d, hint=%d\n",
@@ -938,8 +940,12 @@ static bool execlists_dequeue(struct intel_engine_cs *engine)
&engine->active.requests))
return false;
- if (i915_request_started(last))
+#if 0
+ if (i915_request_started(last)) {
+ mod_timer(&execlists->timer, jiffies + 1);
return false;
+ }
+#endif
/*
* WaIdleLiteRestore:bdw,skl
@@ -1306,7 +1312,8 @@ static void __execlists_submission_tasklet(struct intel_engine_cs *const engine)
submit = false;
if (!engine->execlists.pending[0])
- submit = execlists_dequeue(engine);
+ submit = execlists_dequeue(engine,
+ last_active(&engine->execlists));
} while (submit);
}
@@ -1324,6 +1331,79 @@ static void execlists_submission_tasklet(unsigned long data)
spin_unlock_irqrestore(&engine->active.lock, flags);
}
+static struct i915_request *
+node_to_request(struct i915_sched_node *node)
+{
+ return container_of(node, struct i915_request, sched);
+}
+
+static void defer_request(struct i915_request * const rq,
+ struct list_head * const pl)
+{
+ struct i915_dependency *p;
+
+ /*
+ * We want to move the interrupted request to the back of
+ * the round-robin list (i.e. its priority level), but
+ * in doing so, we must then move all requests that were in
+ * flight and were waiting for the interrupted request to
+ * after it again.
+ */
+ list_move_tail(&rq->sched.link, pl);
+
+ list_for_each_entry(p, &rq->sched.waiters_list, wait_link) {
+ struct i915_request *w = node_to_request(p->waiter);
+
+ if (!i915_sw_fence_done(&w->submit))
+ continue;
+
+ GEM_BUG_ON(w->engine != rq->engine);
+ GEM_BUG_ON(rq_prio(w) > rq_prio(rq));
+
+ if (rq_prio(w) < rq_prio(rq))
+ continue;
+
+ /*
+ * This should be very shallow as it is limited by the
+ * number of requests that can fit in a ring (<64) and
+ * the number of contexts that can be in flight.
+ */
+ defer_request(w, pl);
+ }
+}
+
+static void interrupt_context(struct intel_engine_cs *engine)
+{
+ struct i915_request *rq = __unwind_incomplete_requests(engine);
+
+ defer_request(rq, i915_sched_lookup_priolist(engine, rq_prio(rq)));
+}
+
+static void execlists_submission_timer(struct timer_list *timer)
+{
+ struct intel_engine_cs *engine =
+ from_timer(engine, timer, execlists.timer);
+ struct intel_engine_execlists *execlists = &engine->execlists;
+ unsigned long flags;
+
+ spin_lock_irqsave(&engine->active.lock, flags);
+
+ process_csb(engine);
+ if (execlists->pending[0]) {
+ mod_timer(&execlists->timer, jiffies + 1);
+ } else if (execlists->active[1]) {
+ trace_ports(execlists, "interrupted", execlists->active);
+
+ ring_suspend(engine) = 1;
+ interrupt_context(engine);
+ execlists_dequeue(engine, NULL);
+ } else {
+ execlists_dequeue(engine, last_active(&engine->execlists));
+ }
+
+ spin_unlock_irqrestore(&engine->active.lock, flags);
+}
+
static void queue_request(struct intel_engine_cs *engine,
struct i915_sched_node *node,
int prio)
@@ -2572,6 +2652,7 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
if (WARN_ON(test_bit(TASKLET_STATE_SCHED,
&engine->execlists.tasklet.state)))
tasklet_kill(&engine->execlists.tasklet);
+ del_timer_sync(&engine->execlists.timer);
dev_priv = engine->i915;
@@ -2676,6 +2757,7 @@ int intel_execlists_submission_setup(struct intel_engine_cs *engine)
tasklet_init(&engine->execlists.tasklet,
execlists_submission_tasklet, (unsigned long)engine);
+ timer_setup(&engine->execlists.timer, execlists_submission_timer, 0);
logical_ring_default_vfuncs(engine);
logical_ring_default_irqs(engine);
diff --git a/drivers/gpu/drm/i915/i915_scheduler.c b/drivers/gpu/drm/i915/i915_scheduler.c
index 601aae909491..7d0c14a5e687 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.c
+++ b/drivers/gpu/drm/i915/i915_scheduler.c
@@ -71,6 +71,7 @@ bool __i915_sched_node_add_dependency(struct i915_sched_node *node,
list_add(&dep->wait_link, &signal->waiters_list);
list_add(&dep->signal_link, &node->signalers_list);
dep->signaler = signal;
+ dep->waiter = node;
dep->flags = flags;
/* Keep track of whether anyone on this chain has a semaphore */
diff --git a/drivers/gpu/drm/i915/i915_scheduler_types.h b/drivers/gpu/drm/i915/i915_scheduler_types.h
index 166a457884b2..21fb9cd81fcb 100644
--- a/drivers/gpu/drm/i915/i915_scheduler_types.h
+++ b/drivers/gpu/drm/i915/i915_scheduler_types.h
@@ -62,6 +62,7 @@ struct i915_sched_node {
struct i915_dependency {
struct i915_sched_node *signaler;
+ struct i915_sched_node *waiter;
struct list_head signal_link;
struct list_head wait_link;
struct list_head dfs_link;
--
2.20.1
More information about the Intel-gfx-trybot
mailing list