diff options
Diffstat (limited to 'sysdeps/x86_64/multiarch/memcmp-sse4.S')
| -rw-r--r-- | sysdeps/x86_64/multiarch/memcmp-sse4.S | 84 |
1 files changed, 40 insertions, 44 deletions
diff --git a/sysdeps/x86_64/multiarch/memcmp-sse4.S b/sysdeps/x86_64/multiarch/memcmp-sse4.S index 1ed4200f4c..d7b147e5ce 100644 --- a/sysdeps/x86_64/multiarch/memcmp-sse4.S +++ b/sysdeps/x86_64/multiarch/memcmp-sse4.S @@ -25,10 +25,6 @@ # define MEMCMP __memcmp_sse4_1 # endif -# ifndef ALIGN -# define ALIGN(n) .p2align n -# endif - # define JMPTBL(I, B) (I - B) # define BRANCH_TO_JMPTBL_ENTRY(TABLE, INDEX, SCALE) \ @@ -60,7 +56,7 @@ ENTRY (MEMCMP) BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) # ifndef USE_AS_WMEMCMP - ALIGN (4) + .p2align 4 L(firstbyte): movzbl (%rdi), %eax movzbl (%rsi), %ecx @@ -68,7 +64,7 @@ L(firstbyte): ret # endif - ALIGN (4) + .p2align 4 L(79bytesormore): movdqu (%rsi), %xmm1 movdqu (%rdi), %xmm2 @@ -316,7 +312,7 @@ L(less32bytesin256): add %rdx, %rdi BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) - ALIGN (4) + .p2align 4 L(512bytesormore): # ifdef DATA_CACHE_SIZE_HALF mov $DATA_CACHE_SIZE_HALF, %R8_LP @@ -329,7 +325,7 @@ L(512bytesormore): cmp %r8, %rdx ja L(L2_L3_cache_unaglined) sub $64, %rdx - ALIGN (4) + .p2align 4 L(64bytesormore_loop): movdqu (%rdi), %xmm2 pxor (%rsi), %xmm2 @@ -361,7 +357,7 @@ L(64bytesormore_loop): L(L2_L3_cache_unaglined): sub $64, %rdx - ALIGN (4) + .p2align 4 L(L2_L3_unaligned_128bytes_loop): prefetchnta 0x1c0(%rdi) prefetchnta 0x1c0(%rsi) @@ -396,7 +392,7 @@ L(L2_L3_unaligned_128bytes_loop): /* * This case is for machines which are sensitive for unaligned instructions. */ - ALIGN (4) + .p2align 4 L(2aligned): cmp $128, %rdx ja L(128bytesormorein2aligned) @@ -444,7 +440,7 @@ L(less32bytesin64in2alinged): add %rdx, %rdi BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) - ALIGN (4) + .p2align 4 L(128bytesormorein2aligned): cmp $512, %rdx ja L(512bytesormorein2aligned) @@ -519,7 +515,7 @@ L(less32bytesin128in2aligned): add %rdx, %rdi BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) - ALIGN (4) + .p2align 4 L(256bytesormorein2aligned): sub $256, %rdx @@ -632,7 +628,7 @@ L(less32bytesin256in2alinged): add %rdx, %rdi BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) - ALIGN (4) + .p2align 4 L(512bytesormorein2aligned): # ifdef DATA_CACHE_SIZE_HALF mov $DATA_CACHE_SIZE_HALF, %R8_LP @@ -646,7 +642,7 @@ L(512bytesormorein2aligned): ja L(L2_L3_cache_aglined) sub $64, %rdx - ALIGN (4) + .p2align 4 L(64bytesormore_loopin2aligned): movdqa (%rdi), %xmm2 pxor (%rsi), %xmm2 @@ -678,7 +674,7 @@ L(64bytesormore_loopin2aligned): L(L2_L3_cache_aglined): sub $64, %rdx - ALIGN (4) + .p2align 4 L(L2_L3_aligned_128bytes_loop): prefetchnta 0x1c0(%rdi) prefetchnta 0x1c0(%rsi) @@ -711,7 +707,7 @@ L(L2_L3_aligned_128bytes_loop): BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4) - ALIGN (4) + .p2align 4 L(64bytesormore_loop_end): add $16, %rdi add $16, %rsi @@ -806,7 +802,7 @@ L(8bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(12bytes): mov -12(%rdi), %rax mov -12(%rsi), %rcx @@ -827,7 +823,7 @@ L(0bytes): # ifndef USE_AS_WMEMCMP /* unreal case for wmemcmp */ - ALIGN (4) + .p2align 4 L(65bytes): movdqu -65(%rdi), %xmm1 movdqu -65(%rsi), %xmm2 @@ -864,7 +860,7 @@ L(9bytes): sub %edx, %eax ret - ALIGN (4) + .p2align 4 L(13bytes): mov -13(%rdi), %rax mov -13(%rsi), %rcx @@ -877,7 +873,7 @@ L(13bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(5bytes): mov -5(%rdi), %eax mov -5(%rsi), %ecx @@ -888,7 +884,7 @@ L(5bytes): sub %edx, %eax ret - ALIGN (4) + .p2align 4 L(66bytes): movdqu -66(%rdi), %xmm1 movdqu -66(%rsi), %xmm2 @@ -929,7 +925,7 @@ L(10bytes): sub %ecx, %eax ret - ALIGN (4) + .p2align 4 L(14bytes): mov -14(%rdi), %rax mov -14(%rsi), %rcx @@ -942,7 +938,7 @@ L(14bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(6bytes): mov -6(%rdi), %eax mov -6(%rsi), %ecx @@ -958,7 +954,7 @@ L(2bytes): sub %ecx, %eax ret - ALIGN (4) + .p2align 4 L(67bytes): movdqu -67(%rdi), %xmm2 movdqu -67(%rsi), %xmm1 @@ -997,7 +993,7 @@ L(11bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(15bytes): mov -15(%rdi), %rax mov -15(%rsi), %rcx @@ -1010,7 +1006,7 @@ L(15bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(7bytes): mov -7(%rdi), %eax mov -7(%rsi), %ecx @@ -1023,7 +1019,7 @@ L(7bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(3bytes): movzwl -3(%rdi), %eax movzwl -3(%rsi), %ecx @@ -1036,7 +1032,7 @@ L(1bytes): ret # endif - ALIGN (4) + .p2align 4 L(68bytes): movdqu -68(%rdi), %xmm2 movdqu -68(%rsi), %xmm1 @@ -1079,7 +1075,7 @@ L(20bytes): # ifndef USE_AS_WMEMCMP /* unreal cases for wmemcmp */ - ALIGN (4) + .p2align 4 L(69bytes): movdqu -69(%rsi), %xmm1 movdqu -69(%rdi), %xmm2 @@ -1115,7 +1111,7 @@ L(21bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(70bytes): movdqu -70(%rsi), %xmm1 movdqu -70(%rdi), %xmm2 @@ -1151,7 +1147,7 @@ L(22bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(71bytes): movdqu -71(%rsi), %xmm1 movdqu -71(%rdi), %xmm2 @@ -1188,7 +1184,7 @@ L(23bytes): ret # endif - ALIGN (4) + .p2align 4 L(72bytes): movdqu -72(%rsi), %xmm1 movdqu -72(%rdi), %xmm2 @@ -1227,7 +1223,7 @@ L(24bytes): # ifndef USE_AS_WMEMCMP /* unreal cases for wmemcmp */ - ALIGN (4) + .p2align 4 L(73bytes): movdqu -73(%rsi), %xmm1 movdqu -73(%rdi), %xmm2 @@ -1265,7 +1261,7 @@ L(25bytes): sub %ecx, %eax ret - ALIGN (4) + .p2align 4 L(74bytes): movdqu -74(%rsi), %xmm1 movdqu -74(%rdi), %xmm2 @@ -1302,7 +1298,7 @@ L(26bytes): movzwl -2(%rsi), %ecx jmp L(diffin2bytes) - ALIGN (4) + .p2align 4 L(75bytes): movdqu -75(%rsi), %xmm1 movdqu -75(%rdi), %xmm2 @@ -1342,7 +1338,7 @@ L(27bytes): xor %eax, %eax ret # endif - ALIGN (4) + .p2align 4 L(76bytes): movdqu -76(%rsi), %xmm1 movdqu -76(%rdi), %xmm2 @@ -1388,7 +1384,7 @@ L(28bytes): # ifndef USE_AS_WMEMCMP /* unreal cases for wmemcmp */ - ALIGN (4) + .p2align 4 L(77bytes): movdqu -77(%rsi), %xmm1 movdqu -77(%rdi), %xmm2 @@ -1430,7 +1426,7 @@ L(29bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(78bytes): movdqu -78(%rsi), %xmm1 movdqu -78(%rdi), %xmm2 @@ -1470,7 +1466,7 @@ L(30bytes): xor %eax, %eax ret - ALIGN (4) + .p2align 4 L(79bytes): movdqu -79(%rsi), %xmm1 movdqu -79(%rdi), %xmm2 @@ -1510,7 +1506,7 @@ L(31bytes): xor %eax, %eax ret # endif - ALIGN (4) + .p2align 4 L(64bytes): movdqu -64(%rdi), %xmm2 movdqu -64(%rsi), %xmm1 @@ -1548,7 +1544,7 @@ L(32bytes): /* * Aligned 8 bytes to avoid 2 branch "taken" in one 16 alinged code block. */ - ALIGN (3) + .p2align 3 L(less16bytes): movsbq %dl, %rdx mov (%rsi, %rdx), %rcx @@ -1585,7 +1581,7 @@ L(diffin2bytes): sub %ecx, %eax ret - ALIGN (4) + .p2align 4 L(end): and $0xff, %eax and $0xff, %ecx @@ -1599,7 +1595,7 @@ L(end): neg %eax ret - ALIGN (4) + .p2align 4 L(nequal_bigger): ret @@ -1611,7 +1607,7 @@ L(unreal_case): END (MEMCMP) .section .rodata.sse4.1,"a",@progbits - ALIGN (3) + .p2align 3 # ifndef USE_AS_WMEMCMP L(table_64bytes): .int JMPTBL (L(0bytes), L(table_64bytes)) |