From: Rafael Antognolli <rafael.antognolli@intel.com>
Date: Wed, 24 Jun 2020 18:52:55 +0000 (+0000)
Subject: iris: Align last_seqnos to 64 bits.
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=293221dddaedb410781d39fdecf3c93bb111475b;p=mesa.git

iris: Align last_seqnos to 64 bits.

last_seqnos is used in atomic operations. Specially on 32 bit platorms,
it tends to be slower if it's not aligned to 64 bits (see
cdc331c6f9f6b2ffc035018de4445dba9b67c1f7). This fixes a small regression
on Bioshock.

Fixes: aba3aed96e4 ("iris: fix export of GEM handles")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5637>
---

diff --git a/src/gallium/drivers/iris/iris_bufmgr.h b/src/gallium/drivers/iris/iris_bufmgr.h
index d1594ff3d62..e69b5a55e3f 100644
--- a/src/gallium/drivers/iris/iris_bufmgr.h
+++ b/src/gallium/drivers/iris/iris_bufmgr.h
@@ -203,8 +203,11 @@ struct iris_bo {
     *
     * Although this is a global field, use in multiple contexts should be
     * safe, see iris_emit_buffer_barrier_for() for details.
+    *
+    * Also align it to 64 bits. This will make atomic operations faster on 32
+    * bit platforms.
     */
-   uint64_t last_seqnos[NUM_IRIS_DOMAINS];
+   uint64_t last_seqnos[NUM_IRIS_DOMAINS] __attribute__ ((aligned (8)));
 
    /**
     * Boolean of whether the GPU is definitely not accessing the buffer.