drm/i915: Open-code i915_gpu_idle() for handling seqno wrapping
authorChris Wilson <chris@chris-wilson.co.uk>
Mon, 10 Dec 2012 11:56:17 +0000 (13:56 +0200)
committerDaniel Vetter <daniel.vetter@ffwll.ch>
Tue, 11 Dec 2012 13:07:03 +0000 (14:07 +0100)
The complication is that during seqno wrapping we must be extremely
careful not to write to any ring as that will require a new seqno, and
so would recurse back into the seqno wrap handler. So we cannot call
i915_gpu_idle() as that does additional work beyond simply retiring the
current set of requests, and instead must do the minimal work ourselves
during seqno wrapping.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drivers/gpu/drm/i915/i915_gem.c

index a81b78a59bd94c94a212e747a8fda50742dd40c7..bc73f98c3631f54b2aa74d9787764148c3ea6429 100644 (file)
@@ -1938,11 +1938,15 @@ i915_gem_handle_seqno_wrap(struct drm_device *dev)
        if (ret == 0)
                return ret;
 
-       ret = i915_gpu_idle(dev);
-       if (ret)
-               return ret;
-
+       /* Carefully retire all requests without writing to the rings */
+       for_each_ring(ring, dev_priv, i) {
+               ret = intel_ring_idle(ring);
+               if (ret)
+                       return ret;
+       }
        i915_gem_retire_requests(dev);
+
+       /* Finally reset hw state */
        for_each_ring(ring, dev_priv, i) {
                ret = intel_ring_handle_seqno_wrap(ring);
                if (ret)