drm/i915: Support for 'unflushed' ring idle

When the seqno wraps around zero, the entire GPU is forced to be idle
for some reason (possibly only to work around issues with hardware
semaphores but no-one seems too sure!). This causes a problem if the
force idle occurs at an inopportune moment such as in the middle of
submitting a batch buffer. Specifically, it would lead to recursive
submits - submitting work requires a new seqno, the new seqno requires
idling the ring, idling the ring requires submitting work, submitting
work requires a new seqno...

This change adds a 'flush' parameter to the idle function call which
specifies whether the scheduler queues should be flushed out. I.e. is
the call intended to just idle the ring as it is right now (no flush)
or is it intended to force all outstanding work out of the system
(with flush).

In the seqno wrap case, pending work is not an issue because the next
operation will be to submit it. However, in other cases, the intention
is to make sure everything that could be done has been done.

For: VIZ-1587
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

4 files changed, 34 insertions, 5 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a7d91a30abbb..e3c8cabfb09f 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3834,7 +3834,7 @@ int i915_gpu_idle(struct drm_device *dev)
 			i915_add_request_no_flush(req);
 		}
 
-		ret = intel_engine_idle(engine);
+		ret = intel_engine_idle_flush(engine);
 		if (ret)
 			return ret;
 	}
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 3377df4900c1..5cef9499df66 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1095,7 +1095,7 @@ void intel_logical_ring_stop(struct intel_engine_cs *engine)
 	if (!intel_engine_initialized(engine))
 		return;
 
-	ret = intel_engine_idle(engine);
+	ret = intel_engine_idle_flush(engine);
 	if (ret && !i915_reset_in_progress(&to_i915(engine->dev)->gpu_error))
 		DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n",
 			  engine->name, ret);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 6305b86e468a..270331bbecf5 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -2372,10 +2372,37 @@ static void __wrap_ring_buffer(struct intel_ringbuffer *ringbuf)
 	intel_ring_update_space(ringbuf);
 }
 
-int intel_engine_idle(struct intel_engine_cs *engine)
+/**
+ * __intel_ring_idle - Force the engine to be idle.
+ * @engine: Engine to be idled
+ * @flush: Should queued scheduler work also be flushed
+ * Waits for all outstanding requests that have been sent to the given engine
+ * to complete. Can optionally also force all unsent requests that are queued
+ * in the scheduler to be sent first.
+ * Returns zero on success otherwise a negative error code.
+ *
+ * NB: Flushing can lead to recursion if called at the wrong time. E.g. flush
+ * causes the scheduler to submit requests to the hardware, submitting
+ * requests requires allocating a new seqno, when the seqno wraps around it
+ * idles the engine, idling with flush causes the scheduler to submit requests...
+ */
+int __intel_engine_idle(struct intel_engine_cs *engine, bool flush)
 {
 	struct drm_i915_gem_request *req;
 	uint32_t flags;
+	int ret;
+
+	/*
+	 * NB: Must not flush the scheduler if this idle request is from
+	 * within an execbuff submission (i.e. due to 'get_seqno' calling
+	 * 'wrap_seqno' calling 'idle'). As that would lead to recursive
+	 * flushes!
+	 */
+	if (flush) {
+		ret = i915_scheduler_flush(engine, true);
+		if (ret)
+			return ret;
+	}
 
 	/* Wait upon the last request to be completed */
 	if (list_empty(&engine->request_list))
@@ -3190,7 +3217,7 @@ intel_stop_engine(struct intel_engine_cs *engine)
 	if (!intel_engine_initialized(engine))
 		return;
 
-	ret = intel_engine_idle(engine);
+	ret = intel_engine_idle_flush(engine);
 	if (ret && !i915_reset_in_progress(&to_i915(engine->dev)->gpu_error))
 		DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n",
 			  engine->name, ret);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 05d757aa8d29..761347c418fd 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -472,7 +472,9 @@ void intel_ring_update_space(struct intel_ringbuffer *ringbuf);
 int intel_ring_space(struct intel_ringbuffer *ringbuf);
 bool intel_engine_stopped(struct intel_engine_cs *engine);
 
-int __must_check intel_engine_idle(struct intel_engine_cs *engine);
+#define intel_engine_idle(engine)           __intel_engine_idle((engine), false)
+#define intel_engine_idle_flush(engine)     __intel_engine_idle((engine), true)
+int __must_check __intel_engine_idle(struct intel_engine_cs *engine, bool flush);
 void intel_ring_init_seqno(struct intel_engine_cs *engine, u32 seqno);
 int intel_ring_flush_all_caches(struct drm_i915_gem_request *req);
 int intel_ring_invalidate_all_caches(struct drm_i915_gem_request *req);