drm/i915: Disable EXEC_OBJECT_ASYNC when doing relocations

If we write a relocation into the buffer, we require our own implicit
synchronisation added after the start of the execbuf, outside of the
user's control. As we may end up clflushing, or doing the patch itself
on the GPU, asynchronously we need to look at the implicit serialisation
on obj->resv and hence need to disable EXEC_OBJECT_ASYNC for this
object.

If the user does trigger a stall for relocations, we make sure the stall
is complete enough so that the batch is not submitted before we complete
those relocations.

Fixes: 77ae9957897d ("drm/i915: Enable userspace to opt-out of implicit fencing")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index d7154688eba9c6c82f2de5e07ca7185d54ea9b87..9c3f6c40270f1a0c1d0215ad65f947a5db411b29 100644 (file)
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -699,6 +699,16 @@ eb_relocate_entry(struct i915_vma *vma,
                return -EINVAL;
        }
 
+       /*
+        * If we write into the object, we need to force the synchronisation
+        * barrier, either with an asynchronous clflush or if we executed the
+        * patching using the GPU (though that should be serialised by the
+        * timeline). To be completely sure, and since we are required to
+        * do relocations we are already stalling, disable the user's opt
+        * of our synchronisation.
+        */
+       vma->exec_entry->flags &= ~EXEC_OBJECT_ASYNC;
+
        ret = relocate_entry(vma->obj, reloc, &eb->reloc_cache, target_offset);
        if (ret)
                return ret;