gcc/ 2009-06-17 H.J. Lu PR target/40470 * config/i386/i386.h (CLASS_LIKELY_SPILLED_P): Add SSE_FIRST_REG. gcc/testsuite/ 2009-06-17 H.J. Lu PR target/40470 * gcc.dg/torture/pr40470-1.c: New. * gcc.dg/torture/pr40470-2.c: Likewise. * gcc.dg/torture/pr40470-3.c: Likewise. * gcc.dg/torture/pr40470-4.c: Likewise. --- gcc/config/i386/i386.h.xmm0 2009-06-18 09:38:16.000000000 -0700 +++ gcc/config/i386/i386.h 2009-06-18 11:00:35.000000000 -0700 @@ -1498,6 +1498,7 @@ enum reg_class || ((CLASS) == AD_REGS) \ || ((CLASS) == SIREG) \ || ((CLASS) == DIREG) \ + || ((CLASS) == SSE_FIRST_REG) \ || ((CLASS) == FP_TOP_REG) \ || ((CLASS) == FP_SECOND_REG)) --- gcc/testsuite/gcc.dg/torture/pr40470-1.c.xmm0 2009-06-18 09:38:59.000000000 -0700 +++ gcc/testsuite/gcc.dg/torture/pr40470-1.c 2009-06-18 09:38:59.000000000 -0700 @@ -0,0 +1,22 @@ +/* { dg-do compile { target i?86-*-* x86_64-*-* } } */ +/* { dg-options "-msse4" } */ +#include +__m128i load (char *); +char * +foo (char *p1, char *p2, + int bmsk, __m128i mask1, __m128i mask2) +{ + int len = 0; + __m128i frag1, frag2; + int cmp_s; + if( !p2[0]) return p1; + if( !p1[0] ) return NULL; + frag2 = load (p2); + frag1 = load (p1); + frag2 = _mm_blendv_epi8(frag2, mask2, mask1); + frag1 = _mm_blendv_epi8(frag1, mask2, mask1); + cmp_s = _mm_cmpistrs(frag2, frag1, 0x0c); + if( cmp_s ) + __asm("bsfl %[bmsk], %[len]" : [len] "=r" (len) : [bmsk] "r" (bmsk) ); + return p1 + len; +} --- gcc/testsuite/gcc.dg/torture/pr40470-2.c.xmm0 2009-06-18 09:38:59.000000000 -0700 +++ gcc/testsuite/gcc.dg/torture/pr40470-2.c 2009-06-18 09:38:59.000000000 -0700 @@ -0,0 +1,27 @@ +/* { dg-do compile { target i?86-*-* x86_64-*-* } } */ +/* { dg-options "-msse4" } */ +#include +__m128i load (char *); +char * +foo (const unsigned char *s1, const unsigned char *s2, + int bmsk, __m128i frag2) +{ + int len = 0; + char *p1 = (char *) s1; + char *p2 = (char *) s2; + __m128i frag1, fruc, mask; + int cmp_c, cmp_s; + if( !p2[0]) return (char *) s1; + if( !p1[0] ) return NULL; + if( p2[1]) frag2 = load (p2); + frag1 = load (p1); + fruc = _mm_loadu_si128 ((__m128i *) s1); + mask = _mm_cmpistrm(fruc, frag2, 0x44); + frag2 = _mm_blendv_epi8(frag2, mask, mask); + frag1 = _mm_blendv_epi8(frag1, mask, mask); + cmp_c = _mm_cmpistrc(frag2, frag1, 0x0c); + cmp_s = _mm_cmpistrs(frag2, frag1, 0x0c); + if( cmp_s & cmp_c ) + __asm("bsfl %[bmsk], %[len]" : [len] "=r" (len) : [bmsk] "r" (bmsk) ); + return p2 + len; +} --- gcc/testsuite/gcc.dg/torture/pr40470-3.c.xmm0 2009-06-18 09:38:59.000000000 -0700 +++ gcc/testsuite/gcc.dg/torture/pr40470-3.c 2009-06-18 09:38:59.000000000 -0700 @@ -0,0 +1,22 @@ +/* { dg-do compile { target i?86-*-* x86_64-*-* } } */ +/* { dg-options "-msse4" } */ +#include +__m128i load (char *); +char * +foo (char *p1, char *p2, + int bmsk, __m128i mask1, __m128i mask2) +{ + int len = 0; + __m128i frag1, frag2; + int cmp_s; + if( !p2[0]) return p1; + if( !p1[0] ) return NULL; + frag2 = load (p2); + frag1 = load (p1); + frag2 = _mm_blendv_epi8(frag2, mask2, mask1); + frag1 = _mm_blendv_epi8(frag1, mask1, mask2); + cmp_s = _mm_cmpistrs(frag2, frag1, 0x0c); + if( cmp_s ) + __asm("bsfl %[bmsk], %[len]" : [len] "=r" (len) : [bmsk] "r" (bmsk) ); + return p1 + len; +} --- gcc/testsuite/gcc.dg/torture/pr40470-4.c.xmm0 2009-06-18 09:38:59.000000000 -0700 +++ gcc/testsuite/gcc.dg/torture/pr40470-4.c 2009-06-18 09:38:59.000000000 -0700 @@ -0,0 +1,29 @@ +/* { dg-do compile { target i?86-*-* x86_64-*-* } } */ +/* { dg-options "-msse4" } */ +#include +__m128i load (char *); +char * +foo (const unsigned char *s1, const unsigned char *s2, + int bmsk, __m128i frag2) +{ + int len = 0; + char *p1 = (char *) s1; + char *p2 = (char *) s2; + __m128i frag1, fruc1, fruc2, mask1, mask2; + int cmp_c, cmp_s; + if( !p2[0]) return (char *) s1; + if( !p1[0] ) return NULL; + if( p2[1]) frag2 = load (p2); + frag1 = load (p1); + fruc1 = _mm_loadu_si128 ((__m128i *) s1); + fruc2 = _mm_loadu_si128 ((__m128i *) s2); + mask1 = _mm_cmpistrm(fruc1, frag2, 0x44); + mask2 = _mm_cmpistrm(fruc2, frag1, 0x14); + frag2 = _mm_blendv_epi8(frag2, mask1, mask2); + frag1 = _mm_blendv_epi8(frag1, mask2, mask1); + cmp_c = _mm_cmpistrc(frag2, frag1, 0x0c); + cmp_s = _mm_cmpistrs(frag2, frag1, 0x0c); + if( cmp_s & cmp_c ) + __asm("bsfl %[bmsk], %[len]" : [len] "=r" (len) : [bmsk] "r" (bmsk) ); + return p2 + len; +}