If we are forced to do an uninterruptible wait for a result from the GPU, boost the qos of DMA operations to minimise the latency and minimise the amount of time spent waiting for the operation whilst uninterruptible.
This shaves a modicum of time off waiting for GPU operations to complete before performing a MMIO flip, for example. Signed-off-by: Chris Wilson <[email protected]> --- drivers/gpu/drm/i915/i915_gem.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index f532a01..f5da4a0 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -1013,14 +1013,22 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno, i915_reset_in_progress(&dev_priv->gpu_error) || \ reset_counter != atomic_read(&dev_priv->gpu_error.reset_counter)) do { - if (interruptible) + if (interruptible) { end = wait_event_interruptible_timeout(ring->irq_queue, EXIT_COND, timeout_jiffies); - else + } else { + struct pm_qos_request qos; + + memset(&qos, 0, sizeof(qos)); + pm_qos_add_request(&qos, PM_QOS_CPU_DMA_LATENCY, 0); + end = wait_event_timeout(ring->irq_queue, EXIT_COND, timeout_jiffies); + pm_qos_remove_request(&qos); + } + /* We need to check whether any gpu reset happened in between * the caller grabbing the seqno and now ... */ if (reset_counter != atomic_read(&dev_priv->gpu_error.reset_counter)) -- 1.8.3.2 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
