| Révision | ae1b43635802dccd9fc793474c433b5dca23841f (tree) |
|---|---|
| l'heure | 2018-01-16 05:01:44 |
| Auteur | Max Filippov <jcmvbkbc@gmai...> |
| Commiter | Waldemar Brodkorb |
xtensa: fix strcmp
Loops with 'loop forever' annotation inside strcmp are actually meant to
loop forever. Falling through the end of the first loop may result in
equal strings being compared unequal, e.g.:
#include <string.h>
int main(void)
{
char a[4096] attribute((aligned(4)));
char b[4096] attribute((aligned(4)));
memset(a, ' ', 258 * 8);
memset(b, ' ', 258 * 8);
a[255 * 8] = 0;
a[256 * 8] = 'a';
b[255 * 8] = 0;
b[256 * 8] = 'b';
return !(strcmp(a, b) == 0);
}
Falling through the end of the second loop may result in unequal strings
being compared as equal, e.g.:
#include <string.h>
int main(void)
{
char a[4096] attribute((aligned(4)));
char b[4096] attribute((aligned(4)));
memset(a, ' ', 514 * 6);
memset(b, ' ', 514 * 6);
a[514 * 6 + 0] = 'a';
a[514 * 6 + 1] = 0;
b[514 * 6 + 0] = 'b';
b[514 * 6 + 1] = 0;
return !(strcmp(a, b) != 0);
}
Use 0 as a loop counter to make 232 - 1 iterations which is enough to
cover all addressable memory. While at it drop useless nop at the end of
the first loop and use a11 for all loop counters.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
libc/string/xtensa/strcmp.S (diff)| @@ -92,8 +92,8 @@ ENTRY (strcmp) | ||
| 92 | 92 | /* (2 mod 4) alignment for loop instruction */ |
| 93 | 93 | .Lunaligned: |
| 94 | 94 | #if XCHAL_HAVE_LOOPS |
| 95 | - _movi.n a8, 0 /* set up for the maximum loop count */ | |
| 96 | - loop a8, .Lretdiff /* loop forever (almost anyway) */ | |
| 95 | + movi a11, 0 /* set up for the maximum loop count */ | |
| 96 | + loop a11, .Lretdiff /* loop forever (almost anyway) */ | |
| 97 | 97 | #endif |
| 98 | 98 | .Lnextbyte: |
| 99 | 99 | l8ui a8, a2, 0 |
| @@ -131,11 +131,10 @@ ENTRY (strcmp) | ||
| 131 | 131 | #if XCHAL_HAVE_LOOPS |
| 132 | 132 | .Laligned: |
| 133 | 133 | .begin no-transform |
| 134 | + movi a11, 0 | |
| 134 | 135 | l32r a4, .Lmask0 /* mask for byte 0 */ |
| 135 | 136 | l32r a7, .Lmask4 |
| 136 | - /* Loop forever. (a4 is more than than the maximum number | |
| 137 | - of iterations) */ | |
| 138 | - loop a4, .Laligned_done | |
| 137 | + loop a11, .Laligned_done /* Loop forever. */ | |
| 139 | 138 | |
| 140 | 139 | /* First unrolled loop body. */ |
| 141 | 140 | l32i a8, a2, 0 /* get word from s1 */ |
| @@ -156,8 +155,6 @@ ENTRY (strcmp) | ||
| 156 | 155 | addi a2, a2, 8 /* advance s1 pointer */ |
| 157 | 156 | addi a3, a3, 8 /* advance s2 pointer */ |
| 158 | 157 | .Laligned_done: |
| 159 | - or a1, a1, a1 /* nop */ | |
| 160 | - | |
| 161 | 158 | .Lprobeq2: |
| 162 | 159 | /* Adjust pointers to account for the loop unrolling. */ |
| 163 | 160 | addi a2, a2, 4 |
| @@ -198,7 +195,7 @@ ENTRY (strcmp) | ||
| 198 | 195 | #if XCHAL_HAVE_LOOPS |
| 199 | 196 | |
| 200 | 197 | /* align (1 mod 4) */ |
| 201 | - loop a4, .Leq /* loop forever (a4 is bigger than max iters) */ | |
| 198 | + loop a11, .Leq /* loop forever */ | |
| 202 | 199 | .end no-transform |
| 203 | 200 | |
| 204 | 201 | l32i a8, a2, 0 /* get word from s1 */ |
Fork