drm/i915: Set sync_seqno properly after seqno wrap

i915_gem_handle_seqno_wrap() will zero all sync_seqnos but as the
wrap can happen inside ring->sync_to(), pre wrap seqno was
carried over and overwrote the zeroed sync_seqno.

When wrap is handled, all outstanding requests will be retired and
objects moved to inactive queue, causing their last_read_seqno to be zero.
Use this to update the sync_seqno correctly.

RING_SYNC registers after wrap will contain pre wrap values which
are >= seqno. So injecting the semaphore wait into ring completes
immediately.

Original idea for using last_read_seqno from Chris Wilson.

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

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 85a09482c2a9b8bd9ca396551eb15b711ca96c9e..4ef8b5714337f20f7fb3a44a69c69a6cce574d1c 100644 (file)
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2397,7 +2397,11 @@ i915_gem_object_sync(struct drm_i915_gem_object *obj,
 
        ret = to->sync_to(to, from, seqno);
        if (!ret)
-               from->sync_seqno[idx] = seqno;
+               /* We use last_read_seqno because sync_to()
+                * might have just caused seqno wrap under
+                * the radar.
+                */
+               from->sync_seqno[idx] = obj->last_read_seqno;
 
        return ret;
 }