From 8a7413f9b036da83ffde491a37d9d2340bc321a7 Mon Sep 17 00:00:00 2001 From: Alan Modra Date: Sat, 17 Aug 2013 18:41:17 +0930 Subject: PowerPC LE strcmp and strncmp http://sourceware.org/ml/libc-alpha/2013-08/msg00099.html More little-endian support. I leave the main strcmp loops unchanged, (well, except for renumbering rTMP to something other than r0 since it's needed in an addi insn) and modify the tail for little-endian. I noticed some of the big-endian tail code was a little untidy so have cleaned that up too. * sysdeps/powerpc/powerpc64/strcmp.S (rTMP2): Define as r0. (rTMP): Define as r11. (strcmp): Add little-endian support. Optimise tail. * sysdeps/powerpc/powerpc32/strcmp.S: Similarly. * sysdeps/powerpc/powerpc64/strncmp.S: Likewise. * sysdeps/powerpc/powerpc32/strncmp.S: Likewise. * sysdeps/powerpc/powerpc64/power4/strncmp.S: Likewise. * sysdeps/powerpc/powerpc32/power4/strncmp.S: Likewise. * sysdeps/powerpc/powerpc64/power7/strncmp.S: Likewise. * sysdeps/powerpc/powerpc32/power7/strncmp.S: Likewise. --- sysdeps/powerpc/powerpc32/power4/strncmp.S | 56 ++++++++++++++++++++++-------- 1 file changed, 41 insertions(+), 15 deletions(-) (limited to 'sysdeps/powerpc/powerpc32/power4/strncmp.S') diff --git a/sysdeps/powerpc/powerpc32/power4/strncmp.S b/sysdeps/powerpc/powerpc32/power4/strncmp.S index 724d9084a9..89b961e78d 100644 --- a/sysdeps/powerpc/powerpc32/power4/strncmp.S +++ b/sysdeps/powerpc/powerpc32/power4/strncmp.S @@ -24,7 +24,7 @@ EALIGN (strncmp, 4, 0) -#define rTMP r0 +#define rTMP2 r0 #define rRTN r3 #define rSTR1 r3 /* first string arg */ #define rSTR2 r4 /* second string arg */ @@ -37,6 +37,7 @@ EALIGN (strncmp, 4, 0) #define r7F7F r9 /* constant 0x7f7f7f7f */ #define rNEG r10 /* ~(word in s1 | 0x7f7f7f7f) */ #define rBITDIF r11 /* bits that differ in s1 & s2 words */ +#define rTMP r12 dcbt 0,rSTR1 or rTMP, rSTR2, rSTR1 @@ -75,12 +76,45 @@ L(g1): add rTMP, rFEFE, rWORD1 we don't compare two strings as different because of gunk beyond the end of the strings... */ +#ifdef __LITTLE_ENDIAN__ +L(endstring): + slwi rTMP, rTMP, 1 + addi rTMP2, rTMP, -1 + andc rTMP2, rTMP2, rTMP + and rWORD2, rWORD2, rTMP2 /* Mask off gunk. */ + and rWORD1, rWORD1, rTMP2 + rlwinm rTMP2, rWORD2, 8, 0xffffffff /* Byte reverse word. */ + rlwinm rTMP, rWORD1, 8, 0xffffffff + rldimi rTMP2, rWORD2, 24, 32 + rldimi rTMP, rWORD1, 24, 32 + rlwimi rTMP2, rWORD2, 24, 16, 23 + rlwimi rTMP, rWORD1, 24, 16, 23 + xor. rBITDIF, rTMP, rTMP2 + sub rRTN, rTMP, rTMP2 + bgelr+ + ori rRTN, rTMP2, 1 + blr + +L(different): + lwz rWORD1, -4(rSTR1) + rlwinm rTMP2, rWORD2, 8, 0xffffffff /* Byte reverse word. */ + rlwinm rTMP, rWORD1, 8, 0xffffffff + rldimi rTMP2, rWORD2, 24, 32 + rldimi rTMP, rWORD1, 24, 32 + rlwimi rTMP2, rWORD2, 24, 16, 23 + rlwimi rTMP, rWORD1, 24, 16, 23 + xor. rBITDIF, rTMP, rTMP2 + sub rRTN, rTMP, rTMP2 + bgelr+ + ori rRTN, rTMP2, 1 + blr + +#else L(endstring): and rTMP, r7F7F, rWORD1 beq cr1, L(equal) add rTMP, rTMP, r7F7F xor. rBITDIF, rWORD1, rWORD2 - andc rNEG, rNEG, rTMP blt- L(highbit) cntlzw rBITDIF, rBITDIF @@ -88,28 +122,20 @@ L(endstring): addi rNEG, rNEG, 7 cmpw cr1, rNEG, rBITDIF sub rRTN, rWORD1, rWORD2 - blt- cr1, L(equal) - srawi rRTN, rRTN, 31 - ori rRTN, rRTN, 1 - blr + bgelr+ cr1 L(equal): li rRTN, 0 blr L(different): - lwzu rWORD1, -4(rSTR1) + lwz rWORD1, -4(rSTR1) xor. rBITDIF, rWORD1, rWORD2 sub rRTN, rWORD1, rWORD2 - blt- L(highbit) - srawi rRTN, rRTN, 31 - ori rRTN, rRTN, 1 - blr + bgelr+ L(highbit): - srwi rWORD2, rWORD2, 24 - srwi rWORD1, rWORD1, 24 - sub rRTN, rWORD1, rWORD2 + ori rRTN, rWORD2, 1 blr - +#endif /* Oh well. In this case, we just do a byte-by-byte comparison. */ .align 4 -- cgit 1.4.1