Improve my implementations of strcmp() and memcpy() since Nick's profiler
indicates that KDE apps spend 20% of their simulated insns in these two
functions alone.
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@192 a5019735-40e9-0310-863c-91ae7b9d1cf9
diff --git a/coregrind/vg_clientfuncs.c b/coregrind/vg_clientfuncs.c
index e27fa30..436ee25 100644
--- a/coregrind/vg_clientfuncs.c
+++ b/coregrind/vg_clientfuncs.c
@@ -419,16 +419,17 @@
int strcmp ( const char* s1, const char* s2 )
{
+ register char c1, c2;
while (True) {
- if (*s1 == 0 && *s2 == 0) return 0;
- if (*s1 == 0) return -1;
- if (*s2 == 0) return 1;
-
- if (*(char*)s1 < *(char*)s2) return -1;
- if (*(char*)s1 > *(char*)s2) return 1;
-
+ c1 = *s1;
+ c2 = *s2;
+ if (c1 != c2) break;
+ if (c1 == 0) break;
s1++; s2++;
}
+ if (c1 < c2) return -1;
+ if (c1 > c2) return 1;
+ return 0;
}
void* memchr(const void *s, int c, unsigned int n)
@@ -448,13 +449,29 @@
if ( dst > src ) {
d = (char *)dst + len - 1;
s = (char *)src + len - 1;
- while ( len-- )
+ while ( len >= 4 ) {
*d-- = *s--;
+ *d-- = *s--;
+ *d-- = *s--;
+ *d-- = *s--;
+ len -= 4;
+ }
+ while ( len-- ) {
+ *d-- = *s--;
+ }
} else if ( dst < src ) {
d = (char *)dst;
s = (char *)src;
- while ( len-- )
+ while ( len >= 4 ) {
*d++ = *s++;
+ *d++ = *s++;
+ *d++ = *s++;
+ *d++ = *s++;
+ len -= 4;
+ }
+ while ( len-- ) {
+ *d++ = *s++;
+ }
}
return dst;
}
diff --git a/vg_clientfuncs.c b/vg_clientfuncs.c
index e27fa30..436ee25 100644
--- a/vg_clientfuncs.c
+++ b/vg_clientfuncs.c
@@ -419,16 +419,17 @@
int strcmp ( const char* s1, const char* s2 )
{
+ register char c1, c2;
while (True) {
- if (*s1 == 0 && *s2 == 0) return 0;
- if (*s1 == 0) return -1;
- if (*s2 == 0) return 1;
-
- if (*(char*)s1 < *(char*)s2) return -1;
- if (*(char*)s1 > *(char*)s2) return 1;
-
+ c1 = *s1;
+ c2 = *s2;
+ if (c1 != c2) break;
+ if (c1 == 0) break;
s1++; s2++;
}
+ if (c1 < c2) return -1;
+ if (c1 > c2) return 1;
+ return 0;
}
void* memchr(const void *s, int c, unsigned int n)
@@ -448,13 +449,29 @@
if ( dst > src ) {
d = (char *)dst + len - 1;
s = (char *)src + len - 1;
- while ( len-- )
+ while ( len >= 4 ) {
*d-- = *s--;
+ *d-- = *s--;
+ *d-- = *s--;
+ *d-- = *s--;
+ len -= 4;
+ }
+ while ( len-- ) {
+ *d-- = *s--;
+ }
} else if ( dst < src ) {
d = (char *)dst;
s = (char *)src;
- while ( len-- )
+ while ( len >= 4 ) {
*d++ = *s++;
+ *d++ = *s++;
+ *d++ = *s++;
+ *d++ = *s++;
+ len -= 4;
+ }
+ while ( len-- ) {
+ *d++ = *s++;
+ }
}
return dst;
}