diff options
Diffstat (limited to 'libc/string/xtensa/strcmp.S')
| -rw-r--r-- | libc/string/xtensa/strcmp.S | 180 |
1 files changed, 90 insertions, 90 deletions
diff --git a/libc/string/xtensa/strcmp.S b/libc/string/xtensa/strcmp.S index 622bb27ed..2dce590db 100644 --- a/libc/string/xtensa/strcmp.S +++ b/libc/string/xtensa/strcmp.S @@ -13,11 +13,10 @@ Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public - License along with the GNU C Library; if not, write to the Free - Software Foundation, Inc., 51 Franklin Street - Fifth Floor, - Boston, MA 02110-1301, USA. */ + License along with the GNU C Library; if not, see + <http://www.gnu.org/licenses/>. */ -#include "../../sysdeps/linux/xtensa/sysdep.h" +#include <sysdep.h> #include <bits/xtensa-config.h> #include <features.h> @@ -35,45 +34,46 @@ #define MASK4 0x40404040 + .text + .align 4 + .literal_position .literal .Lmask0, MASK0 .literal .Lmask1, MASK1 .literal .Lmask2, MASK2 .literal .Lmask3, MASK3 .literal .Lmask4, MASK4 - - .text ENTRY (strcmp) /* a2 = s1, a3 = s2 */ - l8ui a8, a2, 0 // byte 0 from s1 - l8ui a9, a3, 0 // byte 0 from s2 - movi a10, 3 // mask + l8ui a8, a2, 0 /* byte 0 from s1 */ + l8ui a9, a3, 0 /* byte 0 from s2 */ + movi a10, 3 /* mask */ bne a8, a9, .Lretdiff or a11, a2, a3 bnone a11, a10, .Laligned - xor a11, a2, a3 // compare low two bits of s1 and s2 - bany a11, a10, .Lunaligned // if they have different alignment + xor a11, a2, a3 /* compare low two bits of s1 and s2 */ + bany a11, a10, .Lunaligned /* if they have different alignment */ /* s1/s2 are not word-aligned. */ - addi a2, a2, 1 // advance s1 - beqz a8, .Leq // bytes equal, if zero, strings are equal - addi a3, a3, 1 // advance s2 - bnone a2, a10, .Laligned // if s1/s2 now aligned - l8ui a8, a2, 0 // byte 1 from s1 - l8ui a9, a3, 0 // byte 1 from s2 - addi a2, a2, 1 // advance s1 - bne a8, a9, .Lretdiff // if different, return difference - beqz a8, .Leq // bytes equal, if zero, strings are equal - addi a3, a3, 1 // advance s2 - bnone a2, a10, .Laligned // if s1/s2 now aligned - l8ui a8, a2, 0 // byte 2 from s1 - l8ui a9, a3, 0 // byte 2 from s2 - addi a2, a2, 1 // advance s1 - bne a8, a9, .Lretdiff // if different, return difference - beqz a8, .Leq // bytes equal, if zero, strings are equal - addi a3, a3, 1 // advance s2 + addi a2, a2, 1 /* advance s1 */ + beqz a8, .Leq /* bytes equal, if zero, strings are equal */ + addi a3, a3, 1 /* advance s2 */ + bnone a2, a10, .Laligned /* if s1/s2 now aligned */ + l8ui a8, a2, 0 /* byte 1 from s1 */ + l8ui a9, a3, 0 /* byte 1 from s2 */ + addi a2, a2, 1 /* advance s1 */ + bne a8, a9, .Lretdiff /* if different, return difference */ + beqz a8, .Leq /* bytes equal, if zero, strings are equal */ + addi a3, a3, 1 /* advance s2 */ + bnone a2, a10, .Laligned /* if s1/s2 now aligned */ + l8ui a8, a2, 0 /* byte 2 from s1 */ + l8ui a9, a3, 0 /* byte 2 from s2 */ + addi a2, a2, 1 /* advance s1 */ + bne a8, a9, .Lretdiff /* if different, return difference */ + beqz a8, .Leq /* bytes equal, if zero, strings are equal */ + addi a3, a3, 1 /* advance s2 */ j .Laligned /* s1 and s2 have different alignment. @@ -92,8 +92,8 @@ ENTRY (strcmp) /* (2 mod 4) alignment for loop instruction */ .Lunaligned: #if XCHAL_HAVE_LOOPS - _movi.n a8, 0 // set up for the maximum loop count - loop a8, .Lretdiff // loop forever (almost anyway) + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, .Lretdiff /* loop forever (almost anyway) */ #endif .Lnextbyte: l8ui a8, a2, 0 @@ -108,7 +108,7 @@ ENTRY (strcmp) #endif .Lretdiff: sub a2, a8, a9 - retw + abi_ret /* s1 is word-aligned; s2 is word-aligned. @@ -131,32 +131,32 @@ ENTRY (strcmp) #if XCHAL_HAVE_LOOPS .Laligned: .begin no-transform - l32r a4, .Lmask0 // mask for byte 0 + l32r a4, .Lmask0 /* mask for byte 0 */ l32r a7, .Lmask4 /* Loop forever. (a4 is more than than the maximum number of iterations) */ loop a4, .Laligned_done /* First unrolled loop body. */ - l32i a8, a2, 0 // get word from s1 - l32i a9, a3, 0 // get word from s2 + l32i a8, a2, 0 /* get word from s1 */ + l32i a9, a3, 0 /* get word from s2 */ slli a5, a8, 1 bne a8, a9, .Lwne2 or a9, a8, a5 bnall a9, a7, .Lprobeq /* Second unrolled loop body. */ - l32i a8, a2, 4 // get word from s1+4 - l32i a9, a3, 4 // get word from s2+4 + l32i a8, a2, 4 /* get word from s1+4 */ + l32i a9, a3, 4 /* get word from s2+4 */ slli a5, a8, 1 bne a8, a9, .Lwne2 or a9, a8, a5 bnall a9, a7, .Lprobeq2 - addi a2, a2, 8 // advance s1 pointer - addi a3, a3, 8 // advance s2 pointer + addi a2, a2, 8 /* advance s1 pointer */ + addi a3, a3, 8 /* advance s2 pointer */ .Laligned_done: - or a1, a1, a1 // nop + or a1, a1, a1 /* nop */ .Lprobeq2: /* Adjust pointers to account for the loop unrolling. */ @@ -166,15 +166,15 @@ ENTRY (strcmp) #else /* !XCHAL_HAVE_LOOPS */ .Laligned: - movi a4, MASK0 // mask for byte 0 + movi a4, MASK0 /* mask for byte 0 */ movi a7, MASK4 j .Lfirstword .Lnextword: - addi a2, a2, 4 // advance s1 pointer - addi a3, a3, 4 // advance s2 pointer + addi a2, a2, 4 /* advance s1 pointer */ + addi a3, a3, 4 /* advance s2 pointer */ .Lfirstword: - l32i a8, a2, 0 // get word from s1 - l32i a9, a3, 0 // get word from s2 + l32i a8, a2, 0 /* get word from s1 */ + l32i a9, a3, 0 /* get word from s2 */ slli a5, a8, 1 bne a8, a9, .Lwne2 or a9, a8, a5 @@ -186,50 +186,50 @@ ENTRY (strcmp) /* Words are probably equal, but check for sure. If not, loop over the rest of string using normal algorithm. */ - bnone a8, a4, .Leq // if byte 0 is zero - l32r a5, .Lmask1 // mask for byte 1 - l32r a6, .Lmask2 // mask for byte 2 - bnone a8, a5, .Leq // if byte 1 is zero - l32r a7, .Lmask3 // mask for byte 3 - bnone a8, a6, .Leq // if byte 2 is zero - bnone a8, a7, .Leq // if byte 3 is zero - addi.n a2, a2, 4 // advance s1 pointer - addi.n a3, a3, 4 // advance s2 pointer + bnone a8, a4, .Leq /* if byte 0 is zero */ + l32r a5, .Lmask1 /* mask for byte 1 */ + l32r a6, .Lmask2 /* mask for byte 2 */ + bnone a8, a5, .Leq /* if byte 1 is zero */ + l32r a7, .Lmask3 /* mask for byte 3 */ + bnone a8, a6, .Leq /* if byte 2 is zero */ + bnone a8, a7, .Leq /* if byte 3 is zero */ + addi.n a2, a2, 4 /* advance s1 pointer */ + addi.n a3, a3, 4 /* advance s2 pointer */ #if XCHAL_HAVE_LOOPS /* align (1 mod 4) */ - loop a4, .Leq // loop forever (a4 is bigger than max iters) + loop a4, .Leq /* loop forever (a4 is bigger than max iters) */ .end no-transform - l32i a8, a2, 0 // get word from s1 - l32i a9, a3, 0 // get word from s2 - addi a2, a2, 4 // advance s1 pointer + l32i a8, a2, 0 /* get word from s1 */ + l32i a9, a3, 0 /* get word from s2 */ + addi a2, a2, 4 /* advance s1 pointer */ bne a8, a9, .Lwne - bnone a8, a4, .Leq // if byte 0 is zero - bnone a8, a5, .Leq // if byte 1 is zero - bnone a8, a6, .Leq // if byte 2 is zero - bnone a8, a7, .Leq // if byte 3 is zero - addi a3, a3, 4 // advance s2 pointer + bnone a8, a4, .Leq /* if byte 0 is zero */ + bnone a8, a5, .Leq /* if byte 1 is zero */ + bnone a8, a6, .Leq /* if byte 2 is zero */ + bnone a8, a7, .Leq /* if byte 3 is zero */ + addi a3, a3, 4 /* advance s2 pointer */ #else /* !XCHAL_HAVE_LOOPS */ j .Lfirstword2 .Lnextword2: - addi a3, a3, 4 // advance s2 pointer + addi a3, a3, 4 /* advance s2 pointer */ .Lfirstword2: - l32i a8, a2, 0 // get word from s1 - l32i a9, a3, 0 // get word from s2 - addi a2, a2, 4 // advance s1 pointer + l32i a8, a2, 0 /* get word from s1 */ + l32i a9, a3, 0 /* get word from s2 */ + addi a2, a2, 4 /* advance s1 pointer */ bne a8, a9, .Lwne - bnone a8, a4, .Leq // if byte 0 is zero - bnone a8, a5, .Leq // if byte 1 is zero - bnone a8, a6, .Leq // if byte 2 is zero - bany a8, a7, .Lnextword2 // if byte 3 is zero + bnone a8, a4, .Leq /* if byte 0 is zero */ + bnone a8, a5, .Leq /* if byte 1 is zero */ + bnone a8, a6, .Leq /* if byte 2 is zero */ + bany a8, a7, .Lnextword2 /* if byte 3 is zero */ #endif /* !XCHAL_HAVE_LOOPS */ /* Words are equal; some byte is zero. */ -.Leq: movi a2, 0 // return equal - retw +.Leq: movi a2, 0 /* return equal */ + abi_ret .Lwne2: /* Words are not equal. On big-endian processors, if none of the bytes are zero, the return value can be determined by a simple @@ -239,22 +239,22 @@ ENTRY (strcmp) bnall a10, a7, .Lsomezero bgeu a8, a9, .Lposreturn movi a2, -1 - retw + abi_ret .Lposreturn: movi a2, 1 - retw -.Lsomezero: // There is probably some zero byte. + abi_ret +.Lsomezero: /* There is probably some zero byte. */ #endif /* __XTENSA_EB__ */ .Lwne: /* Words are not equal. */ - xor a2, a8, a9 // get word with nonzero in byte that differs - bany a2, a4, .Ldiff0 // if byte 0 differs - movi a5, MASK1 // mask for byte 1 - bnone a8, a4, .Leq // if byte 0 is zero - bany a2, a5, .Ldiff1 // if byte 1 differs - movi a6, MASK2 // mask for byte 2 - bnone a8, a5, .Leq // if byte 1 is zero - bany a2, a6, .Ldiff2 // if byte 2 differs - bnone a8, a6, .Leq // if byte 2 is zero + xor a2, a8, a9 /* get word with nonzero in byte that differs */ + bany a2, a4, .Ldiff0 /* if byte 0 differs */ + movi a5, MASK1 /* mask for byte 1 */ + bnone a8, a4, .Leq /* if byte 0 is zero */ + bany a2, a5, .Ldiff1 /* if byte 1 differs */ + movi a6, MASK2 /* mask for byte 2 */ + bnone a8, a5, .Leq /* if byte 1 is zero */ + bany a2, a6, .Ldiff2 /* if byte 2 differs */ + bnone a8, a6, .Leq /* if byte 2 is zero */ #ifdef __XTENSA_EB__ .Ldiff3: .Ldiff2: @@ -263,14 +263,14 @@ ENTRY (strcmp) byte. Just subtract words to get the return value. The high order equal bytes cancel, leaving room for the sign. */ sub a2, a8, a9 - retw + abi_ret .Ldiff0: /* Need to make room for the sign, so can't subtract whole words. */ extui a10, a8, 24, 8 extui a11, a9, 24, 8 sub a2, a10, a11 - retw + abi_ret #else /* !__XTENSA_EB__ */ /* Little-endian is a little more difficult because can't subtract @@ -281,28 +281,28 @@ ENTRY (strcmp) extui a10, a8, 24, 8 extui a11, a9, 24, 8 sub a2, a10, a11 - retw + abi_ret .Ldiff0: /* Byte 0 is different. */ extui a10, a8, 0, 8 extui a11, a9, 0, 8 sub a2, a10, a11 - retw + abi_ret .Ldiff1: /* Byte 0 is equal; byte 1 is different. */ extui a10, a8, 8, 8 extui a11, a9, 8, 8 sub a2, a10, a11 - retw + abi_ret .Ldiff2: /* Bytes 0-1 are equal; byte 2 is different. */ extui a10, a8, 16, 8 extui a11, a9, 16, 8 sub a2, a10, a11 - retw + abi_ret #endif /* !__XTENSA_EB */ |