drm/i915/gt: Expose busywait duration to sysfs

We busywait on an inflight request (one that is currently executing on
HW, and so might complete quickly) prior to setting up an interrupt and
sleeping. The trade off is that we keep an expensive CPU core busy in
order to avoid wake up latency: where that trade off should lie is best
left to the sysadmin.

The busywait mechanism can be compiled out with

	./scripts/config --set-val DRM_I915_SPIN_REQUEST 0

The maximum busywait duration can be adjusted per-engine using,

	/sys/class/drm/card?/engine/*/ms_busywait_duration_ns

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Steve Carbonari <steven.carbonari@intel.com>
Tested-by: Steve Carbonari <steven.carbonari@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200228131716.3243616-4-chris@chris-wilson.co.uk

1 files changed, 10 insertions, 9 deletions

diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index e5a55801f753..feccf29381aa 100644
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -1404,7 +1404,7 @@ void i915_request_add(struct i915_request *rq)
 	mutex_unlock(&tl->mutex);
 }
 
-static unsigned long local_clock_us(unsigned int *cpu)
+static unsigned long local_clock_ns(unsigned int *cpu)
 {
 	unsigned long t;
 
@@ -1421,7 +1421,7 @@ static unsigned long local_clock_us(unsigned int *cpu)
 	 * stop busywaiting, see busywait_stop().
 	 */
 	*cpu = get_cpu();
-	t = local_clock() >> 10;
+	t = local_clock();
 	put_cpu();
 
 	return t;
@@ -1431,15 +1431,15 @@ static bool busywait_stop(unsigned long timeout, unsigned int cpu)
 {
 	unsigned int this_cpu;
 
-	if (time_after(local_clock_us(&this_cpu), timeout))
+	if (time_after(local_clock_ns(&this_cpu), timeout))
 		return true;
 
 	return this_cpu != cpu;
 }
 
-static bool __i915_spin_request(const struct i915_request * const rq,
-				int state, unsigned long timeout_us)
+static bool __i915_spin_request(const struct i915_request * const rq, int state)
 {
+	unsigned long timeout_ns;
 	unsigned int cpu;
 
 	/*
@@ -1467,7 +1467,8 @@ static bool __i915_spin_request(const struct i915_request * const rq,
 	 * takes to sleep on a request, on the order of a microsecond.
 	 */
 
-	timeout_us += local_clock_us(&cpu);
+	timeout_ns = READ_ONCE(rq->engine->props.max_busywait_duration_ns);
+	timeout_ns += local_clock_ns(&cpu);
 	do {
 		if (i915_request_completed(rq))
 			return true;
@@ -1475,7 +1476,7 @@ static bool __i915_spin_request(const struct i915_request * const rq,
 		if (signal_pending_state(state, current))
 			break;
 
-		if (busywait_stop(timeout_us, cpu))
+		if (busywait_stop(timeout_ns, cpu))
 			break;
 
 		cpu_relax();
@@ -1561,8 +1562,8 @@ long i915_request_wait(struct i915_request *rq,
 	 * completion. That requires having a good predictor for the request
 	 * duration, which we currently lack.
 	 */
-	if (IS_ACTIVE(CONFIG_DRM_I915_SPIN_REQUEST) &&
-	    __i915_spin_request(rq, state, CONFIG_DRM_I915_SPIN_REQUEST)) {
+	if (IS_ACTIVE(CONFIG_DRM_I915_MAX_REQUEST_BUSYWAIT) &&
+	    __i915_spin_request(rq, state)) {
 		dma_fence_signal(&rq->fence);
 		goto out;
 	}