drm/i915: Wake-up wait_request() from elapsed hang-check (v2)

If our watchdog fires and we see that the GPU is idle, but that we
are still waiting on an interrupt, forcibly wake-up the waiter.

i915_do_wait_request() should not be racy, yet there are persistent
reports that 945GM hangs whilst the GPU is idle. This implies that the
hardware is not quite as coherent as the documentation claims - a write
followed by a flush is supposed to be coherent in main memory before the
flush is retired and the irq is emitted. This seems to be a sensible and
elegant guard to force the wait to timeout.

v2: Daniel Vetter pointed out that a warning would be useful to explain
why the machine appeared to stall.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>

diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 5161cea7a4ef43ce48b9d1743f4b7e9981d0b5fa..69a36fc035dc92a892baf3929b1dbb360540f982 100644 (file)
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1304,6 +1304,16 @@ void i915_hangcheck_elapsed(unsigned long data)
                                &dev_priv->render_ring),
                        i915_get_tail_request(dev)->seqno)) {
                dev_priv->hangcheck_count = 0;
+
+               /* Issue a wake-up to catch stuck h/w. */
+               if (dev_priv->render_ring.waiting_gem_seqno |
+                   dev_priv->bsd_ring.waiting_gem_seqno) {
+                       DRM_ERROR("Hangcheck timer elapsed... GPU idle, missed IRQ.\n");
+                       if (dev_priv->render_ring.waiting_gem_seqno)
+                               DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
+                       if (dev_priv->bsd_ring.waiting_gem_seqno)
+                               DRM_WAKEUP(&dev_priv->bsd_ring.irq_queue);
+               }
                return;
        }