kstrtox: use "unsigned int" more
PATCH 2/2 kstrtox: use "unsigned int" more
gcc does generates stupid code sign extending data back and forth. Help by using "unsigned int".
add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-61 (-61)
function old new delta
_parse_integer 128 123 -5
It _still_ does generate useless MOVSX but I don't know how to delete it:
0000000000000070 <_parse_integer>:
... a0: 89 c2 mov edx,eax a2: 83 e8 30 sub eax,0x30 a5: 83 f8 09 cmp eax,0x9 a8: 76 11 jbe bb <_parse_integer+0x4b> aa: 83 ca 20 or edx,0x20 ad: 0f be c2 ===> movsx eax,dl <=== useless b0: 8d 50 9f lea edx,[rax-0x61] b3: 83 fa 05 cmp edx,0x5
Patch also helps on embedded archs which generally only like "int". On arm "and 0xff" is generated which is waste because all values used in comparisons are positive.
Signed-off-by Alexey Dobriyan <[email protected]>
---
lib/kstrtox.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
lib/kstrtox.c +6 -4
@@ -52,12 +52,14 @@ unsigned int _parse_integer(const char *s, unsigned int base, unsigned long long res =0 ; rv =0 ;while (1 ) { +unsigned int c = *s; +unsigned int lc = c |0x20 ;/* don't tolower() this line */ unsigned int val; -if ('0' <= *s && *s <='9' ) - val = *s -'0' ; -else if ('a' <=_tolower (*s) &&_tolower (*s) <='f' ) - val =_tolower (*s) -'a' +10 ; +if ('0' <= c && c <='9' ) + val = c -'0' ; +else if ('a' <= lc && lc <='f' ) + val = lc -'a' +10 ;else break ;