drm/i915/guc: Do not use 0 for GuC doorbell cookie
authorMichał Winiarski <michal.winiarski@intel.com>
Wed, 25 Oct 2017 20:00:09 +0000 (22:00 +0200)
committerChris Wilson <chris@chris-wilson.co.uk>
Thu, 26 Oct 2017 20:35:21 +0000 (21:35 +0100)
Apparently, this value is reserved and may be interpreted as changing
doorbell ownership. Even though we're not observing any side effects
now, let's skip over it to be consistent with the spec.

v2: Apply checkpatch (Sagar)

Suggested-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171025200020.16636-2-michal.winiarski@intel.com
drivers/gpu/drm/i915/i915_guc_submission.c

index 07289a4..0b61bf1 100644 (file)
@@ -475,9 +475,12 @@ static void guc_ring_doorbell(struct i915_guc_client *client)
        /* pointer of current doorbell cacheline */
        db = __get_doorbell(client);
 
-       /* we're not expecting the doorbell cookie to change behind our back */
+       /*
+        * We're not expecting the doorbell cookie to change behind our back,
+        * we also need to treat 0 as a reserved value.
+        */
        cookie = READ_ONCE(db->cookie);
-       WARN_ON_ONCE(xchg(&db->cookie, cookie + 1) != cookie);
+       WARN_ON_ONCE(xchg(&db->cookie, cookie + 1 ?: cookie + 2) != cookie);
 
        /* XXX: doorbell was lost and need to acquire it again */
        GEM_BUG_ON(db->db_status != GUC_DOORBELL_ENABLED);