diff options
Diffstat (limited to 'libc/string/xtensa')
-rw-r--r-- | libc/string/xtensa/memcpy.S | 22 | ||||
-rw-r--r-- | libc/string/xtensa/memset.S | 12 | ||||
-rw-r--r-- | libc/string/xtensa/strcmp.S | 148 | ||||
-rw-r--r-- | libc/string/xtensa/strcpy.S | 72 | ||||
-rw-r--r-- | libc/string/xtensa/strlen.S | 56 | ||||
-rw-r--r-- | libc/string/xtensa/strncpy.S | 150 |
6 files changed, 230 insertions, 230 deletions
diff --git a/libc/string/xtensa/memcpy.S b/libc/string/xtensa/memcpy.S index 19f3a6818..fc04c023e 100644 --- a/libc/string/xtensa/memcpy.S +++ b/libc/string/xtensa/memcpy.S @@ -83,7 +83,7 @@ __memcpy_aux: loopnez a4, 2f #else beqz a4, 2f - add a7, a3, a4 // a7 = end address for source + add a7, a3, a4 /* a7 = end address for source */ #endif 1: l8ui a6, a3, 0 addi a3, a3, 1 @@ -98,7 +98,7 @@ __memcpy_aux: /* Destination is unaligned. */ .align 4 -.Ldst1mod2: // dst is only byte aligned +.Ldst1mod2: /* dst is only byte aligned */ /* Do short copies byte-by-byte. */ _bltui a4, 7, .Lbytecopy @@ -113,7 +113,7 @@ __memcpy_aux: /* Return to main algorithm if dst is now aligned. */ _bbci.l a5, 1, .Ldstaligned -.Ldst2mod4: // dst has 16-bit alignment +.Ldst2mod4: /* dst has 16-bit alignment */ /* Do short copies byte-by-byte. */ _bltui a4, 6, .Lbytecopy @@ -134,7 +134,7 @@ __memcpy_aux: ENTRY (memcpy) /* a2 = dst, a3 = src, a4 = len */ - mov a5, a2 // copy dst so that a2 is return value + mov a5, a2 /* copy dst so that a2 is return value */ _bbsi.l a2, 0, .Ldst1mod2 _bbsi.l a2, 1, .Ldst2mod4 .Ldstaligned: @@ -152,7 +152,7 @@ ENTRY (memcpy) #else beqz a7, 2f slli a8, a7, 4 - add a8, a8, a3 // a8 = end of last 16B source chunk + add a8, a8, a3 /* a8 = end of last 16B source chunk */ #endif 1: l32i a6, a3, 0 l32i a7, a3, 4 @@ -218,18 +218,18 @@ ENTRY (memcpy) /* Copy 16 bytes per iteration for word-aligned dst and unaligned src. */ - ssa8 a3 // set shift amount from byte offset + ssa8 a3 /* set shift amount from byte offset */ #if UNALIGNED_ADDRESSES_CHECKED - and a11, a3, a8 // save unalignment offset for below - sub a3, a3, a11 // align a3 + and a11, a3, a8 /* save unalignment offset for below */ + sub a3, a3, a11 /* align a3 */ #endif - l32i a6, a3, 0 // load first word + l32i a6, a3, 0 /* load first word */ #if XCHAL_HAVE_LOOPS loopnez a7, 2f #else beqz a7, 2f slli a10, a7, 4 - add a10, a10, a3 // a10 = end of last 16B source chunk + add a10, a10, a3 /* a10 = end of last 16B source chunk */ #endif 1: l32i a7, a3, 4 l32i a8, a3, 8 @@ -273,7 +273,7 @@ ENTRY (memcpy) mov a6, a7 4: #if UNALIGNED_ADDRESSES_CHECKED - add a3, a3, a11 // readjust a3 with correct misalignment + add a3, a3, a11 /* readjust a3 with correct misalignment */ #endif bbsi.l a4, 1, 5f bbsi.l a4, 0, 6f diff --git a/libc/string/xtensa/memset.S b/libc/string/xtensa/memset.S index c0928825d..076b8f001 100644 --- a/libc/string/xtensa/memset.S +++ b/libc/string/xtensa/memset.S @@ -29,7 +29,7 @@ The algorithm is as follows: Create a word with c in all byte positions. - + If the destination is aligned, set 16B chunks with a loop, and then finish up with 8B, 4B, 2B, and 1B stores conditional on the length. @@ -57,7 +57,7 @@ __memset_aux: loopnez a4, 2f #else beqz a4, 2f - add a6, a5, a4 // a6 = ending address + add a6, a5, a4 /* a6 = ending address */ #endif 1: s8i a3, a5, 0 addi a5, a5, 1 @@ -71,7 +71,7 @@ __memset_aux: .align 4 -.Ldst1mod2: // dst is only byte aligned +.Ldst1mod2: /* dst is only byte aligned */ /* Do short sizes byte-by-byte. */ bltui a4, 8, .Lbyteset @@ -84,7 +84,7 @@ __memset_aux: /* Now retest if dst is aligned. */ _bbci.l a5, 1, .Ldstaligned -.Ldst2mod4: // dst has 16-bit alignment +.Ldst2mod4: /* dst has 16-bit alignment */ /* Do short sizes byte-by-byte. */ bltui a4, 8, .Lbyteset @@ -108,7 +108,7 @@ ENTRY (memset) slli a7, a3, 16 or a3, a3, a7 - mov a5, a2 // copy dst so that a2 is return value + mov a5, a2 /* copy dst so that a2 is return value */ /* Check if dst is unaligned. */ _bbsi.l a2, 0, .Ldst1mod2 @@ -124,7 +124,7 @@ ENTRY (memset) #else beqz a7, 2f slli a6, a7, 4 - add a6, a6, a5 // a6 = end of last 16B chunk + add a6, a6, a5 /* a6 = end of last 16B chunk */ #endif /* Set 16 bytes per iteration. */ 1: s32i a3, a5, 0 diff --git a/libc/string/xtensa/strcmp.S b/libc/string/xtensa/strcmp.S index 622bb27ed..ac058a2bf 100644 --- a/libc/string/xtensa/strcmp.S +++ b/libc/string/xtensa/strcmp.S @@ -45,35 +45,35 @@ 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 @@ -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 @@ -186,49 +186,49 @@ 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 +.Leq: movi a2, 0 /* return equal */ retw .Lwne2: /* Words are not equal. On big-endian processors, if none of the @@ -243,18 +243,18 @@ ENTRY (strcmp) .Lposreturn: movi a2, 1 retw -.Lsomezero: // There is probably some zero byte. +.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: diff --git a/libc/string/xtensa/strcpy.S b/libc/string/xtensa/strcpy.S index 108070384..dc0a15175 100644 --- a/libc/string/xtensa/strcpy.S +++ b/libc/string/xtensa/strcpy.S @@ -36,7 +36,7 @@ ENTRY (strcpy) /* a2 = dst, a3 = src */ - mov a10, a2 // leave dst in return value register + mov a10, a2 /* leave dst in return value register */ movi a4, MASK0 movi a5, MASK1 movi a6, MASK2 @@ -51,23 +51,23 @@ ENTRY (strcpy) j .Ldstunaligned -.Lsrc1mod2: // src address is odd - l8ui a8, a3, 0 // get byte 0 - addi a3, a3, 1 // advance src pointer - s8i a8, a10, 0 // store byte 0 - beqz a8, 1f // if byte 0 is zero - addi a10, a10, 1 // advance dst pointer - bbci.l a3, 1, .Lsrcaligned // if src is now word-aligned +.Lsrc1mod2: /* src address is odd */ + l8ui a8, a3, 0 /* get byte 0 */ + addi a3, a3, 1 /* advance src pointer */ + s8i a8, a10, 0 /* store byte 0 */ + beqz a8, 1f /* if byte 0 is zero */ + addi a10, a10, 1 /* advance dst pointer */ + bbci.l a3, 1, .Lsrcaligned /* if src is now word-aligned */ -.Lsrc2mod4: // src address is 2 mod 4 - l8ui a8, a3, 0 // get byte 0 +.Lsrc2mod4: /* src address is 2 mod 4 */ + l8ui a8, a3, 0 /* get byte 0 */ /* 1-cycle interlock */ - s8i a8, a10, 0 // store byte 0 - beqz a8, 1f // if byte 0 is zero - l8ui a8, a3, 1 // get byte 0 - addi a3, a3, 2 // advance src pointer - s8i a8, a10, 1 // store byte 0 - addi a10, a10, 2 // advance dst pointer + s8i a8, a10, 0 /* store byte 0 */ + beqz a8, 1f /* if byte 0 is zero */ + l8ui a8, a3, 1 /* get byte 0 */ + addi a3, a3, 2 /* advance src pointer */ + s8i a8, a10, 1 /* store byte 0 */ + addi a10, a10, 2 /* advance dst pointer */ bnez a8, .Lsrcaligned 1: retw @@ -78,28 +78,28 @@ ENTRY (strcpy) #if XCHAL_HAVE_LOOPS /* (2 mod 4) alignment for loop instruction */ .Laligned: - _movi.n a8, 0 // set up for the maximum loop count - loop a8, .Lz3 // loop forever (almost anyway) - l32i a8, a3, 0 // get word from src - addi a3, a3, 4 // advance src pointer - bnone a8, a4, .Lz0 // if byte 0 is zero - bnone a8, a5, .Lz1 // if byte 1 is zero - bnone a8, a6, .Lz2 // if byte 2 is zero - s32i a8, a10, 0 // store word to dst - bnone a8, a7, .Lz3 // if byte 3 is zero - addi a10, a10, 4 // advance dst pointer + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, .Lz3 /* loop forever (almost anyway) */ + l32i a8, a3, 0 /* get word from src */ + addi a3, a3, 4 /* advance src pointer */ + bnone a8, a4, .Lz0 /* if byte 0 is zero */ + bnone a8, a5, .Lz1 /* if byte 1 is zero */ + bnone a8, a6, .Lz2 /* if byte 2 is zero */ + s32i a8, a10, 0 /* store word to dst */ + bnone a8, a7, .Lz3 /* if byte 3 is zero */ + addi a10, a10, 4 /* advance dst pointer */ #else /* !XCHAL_HAVE_LOOPS */ -1: addi a10, a10, 4 // advance dst pointer +1: addi a10, a10, 4 /* advance dst pointer */ .Laligned: - l32i a8, a3, 0 // get word from src - addi a3, a3, 4 // advance src pointer - bnone a8, a4, .Lz0 // if byte 0 is zero - bnone a8, a5, .Lz1 // if byte 1 is zero - bnone a8, a6, .Lz2 // if byte 2 is zero - s32i a8, a10, 0 // store word to dst - bany a8, a7, 1b // if byte 3 is zero + l32i a8, a3, 0 /* get word from src */ + addi a3, a3, 4 /* advance src pointer */ + bnone a8, a4, .Lz0 /* if byte 0 is zero */ + bnone a8, a5, .Lz1 /* if byte 1 is zero */ + bnone a8, a6, .Lz2 /* if byte 2 is zero */ + s32i a8, a10, 0 /* store word to dst */ + bany a8, a7, 1b /* if byte 3 is zero */ #endif /* !XCHAL_HAVE_LOOPS */ .Lz3: /* Byte 3 is zero. */ @@ -133,8 +133,8 @@ ENTRY (strcpy) .Ldstunaligned: #if XCHAL_HAVE_LOOPS - _movi.n a8, 0 // set up for the maximum loop count - loop a8, 2f // loop forever (almost anyway) + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, 2f /* loop forever (almost anyway) */ #endif 1: l8ui a8, a3, 0 addi a3, a3, 1 diff --git a/libc/string/xtensa/strlen.S b/libc/string/xtensa/strlen.S index dd72c16fa..9ee4995f4 100644 --- a/libc/string/xtensa/strlen.S +++ b/libc/string/xtensa/strlen.S @@ -36,7 +36,7 @@ ENTRY (strlen) /* a2 = s */ - addi a3, a2, -4 // because we overincrement at the end + addi a3, a2, -4 /* because we overincrement at the end */ movi a4, MASK0 movi a5, MASK1 movi a6, MASK2 @@ -45,21 +45,21 @@ ENTRY (strlen) bbsi.l a2, 1, .L2mod4 j .Laligned -.L1mod2: // address is odd - l8ui a8, a3, 4 // get byte 0 - addi a3, a3, 1 // advance string pointer - beqz a8, .Lz3 // if byte 0 is zero - bbci.l a3, 1, .Laligned // if string pointer is now word-aligned +.L1mod2: /* address is odd */ + l8ui a8, a3, 4 /* get byte 0 */ + addi a3, a3, 1 /* advance string pointer */ + beqz a8, .Lz3 /* if byte 0 is zero */ + bbci.l a3, 1, .Laligned /* if string pointer is now word-aligned */ -.L2mod4: // address is 2 mod 4 - addi a3, a3, 2 // advance ptr for aligned access - l32i a8, a3, 0 // get word with first two bytes of string - bnone a8, a6, .Lz2 // if byte 2 (of word, not string) is zero - bany a8, a7, .Laligned // if byte 3 (of word, not string) is nonzero +.L2mod4: /* address is 2 mod 4 */ + addi a3, a3, 2 /* advance ptr for aligned access */ + l32i a8, a3, 0 /* get word with first two bytes of string */ + bnone a8, a6, .Lz2 /* if byte 2 (of word, not string) is zero */ + bany a8, a7, .Laligned /* if byte 3 (of word, not string) is nonzero */ /* Byte 3 is zero. */ - addi a3, a3, 3 // point to zero byte - sub a2, a3, a2 // subtract to get length + addi a3, a3, 3 /* point to zero byte */ + sub a2, a3, a2 /* subtract to get length */ retw @@ -69,36 +69,36 @@ ENTRY (strlen) /* (2 mod 4) alignment for loop instruction */ .Laligned: #if XCHAL_HAVE_LOOPS - _movi.n a8, 0 // set up for the maximum loop count - loop a8, .Lz3 // loop forever (almost anyway) + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, .Lz3 /* loop forever (almost anyway) */ #endif -1: l32i a8, a3, 4 // get next word of string - addi a3, a3, 4 // advance string pointer - bnone a8, a4, .Lz0 // if byte 0 is zero - bnone a8, a5, .Lz1 // if byte 1 is zero - bnone a8, a6, .Lz2 // if byte 2 is zero +1: l32i a8, a3, 4 /* get next word of string */ + addi a3, a3, 4 /* advance string pointer */ + bnone a8, a4, .Lz0 /* if byte 0 is zero */ + bnone a8, a5, .Lz1 /* if byte 1 is zero */ + bnone a8, a6, .Lz2 /* if byte 2 is zero */ #if XCHAL_HAVE_LOOPS - bnone a8, a7, .Lz3 // if byte 3 is zero + bnone a8, a7, .Lz3 /* if byte 3 is zero */ #else - bany a8, a7, 1b // repeat if byte 3 is non-zero + bany a8, a7, 1b /* repeat if byte 3 is non-zero */ #endif .Lz3: /* Byte 3 is zero. */ - addi a3, a3, 3 // point to zero byte + addi a3, a3, 3 /* point to zero byte */ /* Fall through.... */ .Lz0: /* Byte 0 is zero. */ - sub a2, a3, a2 // subtract to get length + sub a2, a3, a2 /* subtract to get length */ retw .Lz1: /* Byte 1 is zero. */ - addi a3, a3, 1 // point to zero byte - sub a2, a3, a2 // subtract to get length + addi a3, a3, 1 /* point to zero byte */ + sub a2, a3, a2 /* subtract to get length */ retw .Lz2: /* Byte 2 is zero. */ - addi a3, a3, 2 // point to zero byte - sub a2, a3, a2 // subtract to get length + addi a3, a3, 2 /* point to zero byte */ + sub a2, a3, a2 /* subtract to get length */ retw libc_hidden_def (strlen) diff --git a/libc/string/xtensa/strncpy.S b/libc/string/xtensa/strncpy.S index 7ba2ef77d..fe3ec894c 100644 --- a/libc/string/xtensa/strncpy.S +++ b/libc/string/xtensa/strncpy.S @@ -41,29 +41,29 @@ .literal_position __strncpy_aux: -.Lsrc1mod2: // src address is odd - l8ui a8, a3, 0 // get byte 0 - addi a3, a3, 1 // advance src pointer - s8i a8, a10, 0 // store byte 0 - addi a4, a4, -1 // decrement n - beqz a4, .Lret // if n is zero - addi a10, a10, 1 // advance dst pointer - beqz a8, .Lfill // if byte 0 is zero - bbci.l a3, 1, .Lsrcaligned // if src is now word-aligned - -.Lsrc2mod4: // src address is 2 mod 4 - l8ui a8, a3, 0 // get byte 0 - addi a4, a4, -1 // decrement n - s8i a8, a10, 0 // store byte 0 - beqz a4, .Lret // if n is zero - addi a10, a10, 1 // advance dst pointer - beqz a8, .Lfill // if byte 0 is zero - l8ui a8, a3, 1 // get byte 0 - addi a3, a3, 2 // advance src pointer - s8i a8, a10, 0 // store byte 0 - addi a4, a4, -1 // decrement n - beqz a4, .Lret // if n is zero - addi a10, a10, 1 // advance dst pointer +.Lsrc1mod2: /* src address is odd */ + l8ui a8, a3, 0 /* get byte 0 */ + addi a3, a3, 1 /* advance src pointer */ + s8i a8, a10, 0 /* store byte 0 */ + addi a4, a4, -1 /* decrement n */ + beqz a4, .Lret /* if n is zero */ + addi a10, a10, 1 /* advance dst pointer */ + beqz a8, .Lfill /* if byte 0 is zero */ + bbci.l a3, 1, .Lsrcaligned /* if src is now word-aligned */ + +.Lsrc2mod4: /* src address is 2 mod 4 */ + l8ui a8, a3, 0 /* get byte 0 */ + addi a4, a4, -1 /* decrement n */ + s8i a8, a10, 0 /* store byte 0 */ + beqz a4, .Lret /* if n is zero */ + addi a10, a10, 1 /* advance dst pointer */ + beqz a8, .Lfill /* if byte 0 is zero */ + l8ui a8, a3, 1 /* get byte 0 */ + addi a3, a3, 2 /* advance src pointer */ + s8i a8, a10, 0 /* store byte 0 */ + addi a4, a4, -1 /* decrement n */ + beqz a4, .Lret /* if n is zero */ + addi a10, a10, 1 /* advance dst pointer */ bnez a8, .Lsrcaligned j .Lfill @@ -74,8 +74,8 @@ __strncpy_aux: ENTRY (strncpy) /* a2 = dst, a3 = src */ - mov a10, a2 // leave dst in return value register - beqz a4, .Lret // if n is zero + mov a10, a2 /* leave dst in return value register */ + beqz a4, .Lret /* if n is zero */ movi a11, MASK0 movi a5, MASK1 @@ -125,28 +125,28 @@ ENTRY (strncpy) .Lfillcleanup: /* Fill leftover (1 to 3) bytes with zero. */ - s8i a9, a10, 0 // store byte 0 - addi a4, a4, -1 // decrement n + s8i a9, a10, 0 /* store byte 0 */ + addi a4, a4, -1 /* decrement n */ addi a10, a10, 1 - bnez a4, .Lfillcleanup + bnez a4, .Lfillcleanup 2: retw - -.Lfill1mod2: // dst address is odd - s8i a9, a10, 0 // store byte 0 - addi a4, a4, -1 // decrement n - beqz a4, 2b // if n is zero - addi a10, a10, 1 // advance dst pointer - bbci.l a10, 1, .Lfillaligned // if dst is now word-aligned - -.Lfill2mod4: // dst address is 2 mod 4 - s8i a9, a10, 0 // store byte 0 - addi a4, a4, -1 // decrement n - beqz a4, 2b // if n is zero - s8i a9, a10, 1 // store byte 1 - addi a4, a4, -1 // decrement n - beqz a4, 2b // if n is zero - addi a10, a10, 2 // advance dst pointer + +.Lfill1mod2: /* dst address is odd */ + s8i a9, a10, 0 /* store byte 0 */ + addi a4, a4, -1 /* decrement n */ + beqz a4, 2b /* if n is zero */ + addi a10, a10, 1 /* advance dst pointer */ + bbci.l a10, 1, .Lfillaligned /* if dst is now word-aligned */ + +.Lfill2mod4: /* dst address is 2 mod 4 */ + s8i a9, a10, 0 /* store byte 0 */ + addi a4, a4, -1 /* decrement n */ + beqz a4, 2b /* if n is zero */ + s8i a9, a10, 1 /* store byte 1 */ + addi a4, a4, -1 /* decrement n */ + beqz a4, 2b /* if n is zero */ + addi a10, a10, 2 /* advance dst pointer */ j .Lfillaligned @@ -156,32 +156,32 @@ ENTRY (strncpy) /* (2 mod 4) alignment for loop instruction */ .Laligned: #if XCHAL_HAVE_LOOPS - _movi.n a8, 0 // set up for the maximum loop count - loop a8, 1f // loop forever (almost anyway) - blti a4, 5, .Ldstunaligned // n is near limit; do one at a time - l32i a8, a3, 0 // get word from src - addi a3, a3, 4 // advance src pointer - bnone a8, a11, .Lz0 // if byte 0 is zero - bnone a8, a5, .Lz1 // if byte 1 is zero - bnone a8, a6, .Lz2 // if byte 2 is zero - s32i a8, a10, 0 // store word to dst - addi a4, a4, -4 // decrement n - addi a10, a10, 4 // advance dst pointer - bnone a8, a7, .Lfill // if byte 3 is zero -1: + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, 1f /* loop forever (almost anyway) */ + blti a4, 5, .Ldstunaligned /* n is near limit; do one at a time */ + l32i a8, a3, 0 /* get word from src */ + addi a3, a3, 4 /* advance src pointer */ + bnone a8, a11, .Lz0 /* if byte 0 is zero */ + bnone a8, a5, .Lz1 /* if byte 1 is zero */ + bnone a8, a6, .Lz2 /* if byte 2 is zero */ + s32i a8, a10, 0 /* store word to dst */ + addi a4, a4, -4 /* decrement n */ + addi a10, a10, 4 /* advance dst pointer */ + bnone a8, a7, .Lfill /* if byte 3 is zero */ +1: #else /* !XCHAL_HAVE_LOOPS */ -1: blti a4, 5, .Ldstunaligned // n is near limit; do one at a time - l32i a8, a3, 0 // get word from src - addi a3, a3, 4 // advance src pointer - bnone a8, a11, .Lz0 // if byte 0 is zero - bnone a8, a5, .Lz1 // if byte 1 is zero - bnone a8, a6, .Lz2 // if byte 2 is zero - s32i a8, a10, 0 // store word to dst - addi a4, a4, -4 // decrement n - addi a10, a10, 4 // advance dst pointer - bany a8, a7, 1b // no zeroes +1: blti a4, 5, .Ldstunaligned /* n is near limit; do one at a time */ + l32i a8, a3, 0 /* get word from src */ + addi a3, a3, 4 /* advance src pointer */ + bnone a8, a11, .Lz0 /* if byte 0 is zero */ + bnone a8, a5, .Lz1 /* if byte 1 is zero */ + bnone a8, a6, .Lz2 /* if byte 2 is zero */ + s32i a8, a10, 0 /* store word to dst */ + addi a4, a4, -4 /* decrement n */ + addi a10, a10, 4 /* advance dst pointer */ + bany a8, a7, 1b /* no zeroes */ #endif /* !XCHAL_HAVE_LOOPS */ j .Lfill @@ -191,8 +191,8 @@ ENTRY (strncpy) movi a8, 0 #endif s8i a8, a10, 0 - addi a4, a4, -1 // decrement n - addi a10, a10, 1 // advance dst pointer + addi a4, a4, -1 /* decrement n */ + addi a10, a10, 1 /* advance dst pointer */ j .Lfill .Lz1: /* Byte 1 is zero. */ @@ -200,8 +200,8 @@ ENTRY (strncpy) extui a8, a8, 16, 16 #endif s16i a8, a10, 0 - addi a4, a4, -2 // decrement n - addi a10, a10, 2 // advance dst pointer + addi a4, a4, -2 /* decrement n */ + addi a10, a10, 2 /* advance dst pointer */ j .Lfill .Lz2: /* Byte 2 is zero. */ @@ -211,8 +211,8 @@ ENTRY (strncpy) s16i a8, a10, 0 movi a8, 0 s8i a8, a10, 2 - addi a4, a4, -3 // decrement n - addi a10, a10, 3 // advance dst pointer + addi a4, a4, -3 /* decrement n */ + addi a10, a10, 3 /* advance dst pointer */ j .Lfill .align 4 @@ -220,8 +220,8 @@ ENTRY (strncpy) .Ldstunaligned: #if XCHAL_HAVE_LOOPS - _movi.n a8, 0 // set up for the maximum loop count - loop a8, 2f // loop forever (almost anyway) + _movi.n a8, 0 /* set up for the maximum loop count */ + loop a8, 2f /* loop forever (almost anyway) */ #endif 1: l8ui a8, a3, 0 addi a3, a3, 1 |