drm/xe: Use usleep_range for accurate long-running workload timeslicing

commit 80f9c60 upstream.

msleep is not very accurate in terms of how long it actually sleeps,
whereas usleep_range is precise. Replace the timeslice sleep for
long-running workloads with the more accurate usleep_range to avoid
jitter if the sleep period is less than 20ms.

Fixes: dd08ebf ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patch.msgid.link/20251212182847.1683222-3-matthew.brost@intel.com
(cherry picked from commit ca415c4)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Author: mbrost05

drivers/gpu/drm/xe/xe_guc_submit.c

@@ -670,6 +670,24 @@ static u32 wq_space_until_wrap(struct xe_exec_queue *q)
 	return (WQ_SIZE - q->guc->wqi_tail);
 }
 
+static inline void relaxed_ms_sleep(unsigned int delay_ms)
+{
+	unsigned long min_us, max_us;
+
+	if (!delay_ms)
+		return;
+
+	if (delay_ms > 20) {
+		msleep(delay_ms);
+		return;
+	}
+
+	min_us = mul_u32_u32(delay_ms, 1000);
+	max_us = min_us + 500;
+
+	usleep_range(min_us, max_us);
+}
+
 static int wq_wait_for_space(struct xe_exec_queue *q, u32 wqi_size)
 {
 	struct xe_guc *guc = exec_queue_to_guc(q);
@@ -1559,7 +1577,7 @@ static void __guc_exec_queue_process_msg_suspend(struct xe_sched_msg *msg)
 				since_resume_ms;
 
 			if (wait_ms > 0 && q->guc->resume_time)
-				msleep(wait_ms);
+				relaxed_ms_sleep(wait_ms);
 
 			set_exec_queue_suspended(q);
 			disable_scheduling(q, false);