glibc/turn-REP_STOSB_THRESHOLD-from-2k-to-1M.patch

From dc8c5d3ba8ec3c2de8ca0898d682d89492d275b3 Mon Sep 17 00:00:00 2001
From: Shuo Wang <wangshuo_1994@foxmail.com>
Date: Tue, 2 Mar 2021 10:41:09 +0800
Subject: [PATCH] turn REP_STOSB_THRESHOLD from 2k to 1M

REP_STOSB_THRESHOLD is designed to choose vec mov or stosb.
The default threshold (2k) will lead to performance degradation if the
memcpy size is between 2k and 1M.
---
 sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S b/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
index faa40856..76f84748 100644
--- a/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
+++ b/sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
@@ -58,6 +58,16 @@
 # endif
 #endif
 
+/* Threshold to use Enhanced REP STOSB.  Since there is overhead to set
+   up REP STOSB operation, REP STOSB isn't faster on short data.  The
+   memset micro benchmark in glibc shows that 2KB is the approximate
+   value above which REP STOSB becomes faster on processors with
+   Enhanced REP STOSB.  Since the stored value is fixed, larger register
+   size has minimal impact on threshold.  */
+#ifndef REP_STOSB_THRESHOLD
+# define REP_STOSB_THRESHOLD           1048576
+#endif
+
 #ifndef SECTION
 # error SECTION is not defined!
 #endif
@@ -171,7 +181,7 @@ ENTRY (MEMSET_SYMBOL (__memset, unaligned_erms))
 	ret
 
 L(stosb_more_2x_vec):
-	cmp	__x86_rep_stosb_threshold(%rip), %RDX_LP
+	cmp	$REP_STOSB_THRESHOLD, %RDX_LP
 	ja	L(stosb)
 #endif
 L(more_2x_vec):
-- 
2.23.0