patch: sequence points in memset

patch: sequence points in memset

[Pascal Cuoq]

[-- Attachment #1: Type: text/plain, Size: 435 bytes --]

Following a recent discussion of statements shaped like x = x = c;, the attached patch introduces additional sequence points in musl's implementation of memset, because no realistic target platform benefits from the lack of sequence points in the current version, and if a hypothetical target (VLIW maybe?) was designed to benefit from the lack of sequence point, the lack of sequence point might just as well cause harm.

Pascal


[-- Attachment #2: memset_sp.patch --]
[-- Type: application/octet-stream, Size: 632 bytes --]

diff --git a/src/string/memset.c b/src/string/memset.c
index f438b07..5613a14 100644
--- a/src/string/memset.c
+++ b/src/string/memset.c
@@ -11,12 +11,16 @@ void *memset(void *dest, int c, size_t n)
 	 * offsets are well-defined and in the dest region. */
 
 	if (!n) return dest;
-	s[0] = s[n-1] = c;
+	s[0] = c;
+	s[n-1] = c;
 	if (n <= 2) return dest;
-	s[1] = s[n-2] = c;
-	s[2] = s[n-3] = c;
+	s[1] = c;
+	s[2] = c;
+	s[n-2] = c;
+	s[n-3] = c;
 	if (n <= 6) return dest;
-	s[3] = s[n-4] = c;
+	s[3] = c;
+	s[n-4] = c;
 	if (n <= 8) return dest;
 
 	/* Advance pointer to align it at a 4-byte boundary,

Re: patch: sequence points in memset

[Alexander Monakov]

Alternatively, the middle group could read:

	s[1] = s[2] = c;
	s[n-3] = s[n-2] = c;

(it doesn't really matter, but spelling it this way at least leaves
a hint in the source that some assignments are not chained for a reason)

Alexander