summaryrefslogtreecommitdiff
path: root/sysdeps/x86_64
diff options
context:
space:
mode:
Diffstat (limited to 'sysdeps/x86_64')
-rw-r--r--sysdeps/x86_64/cacheinfo.c42
-rw-r--r--sysdeps/x86_64/dl-trampoline.S244
-rw-r--r--sysdeps/x86_64/dl-trampoline.h269
-rw-r--r--sysdeps/x86_64/multiarch/Makefile2
-rw-r--r--sysdeps/x86_64/multiarch/rawmemchr.S1
-rw-r--r--sysdeps/x86_64/multiarch/strcmp-ssse3.S3
-rw-r--r--sysdeps/x86_64/multiarch/strcmp.S12
-rw-r--r--sysdeps/x86_64/multiarch/strcspn-c.c6
-rw-r--r--sysdeps/x86_64/multiarch/strlen.S1
-rw-r--r--sysdeps/x86_64/multiarch/strncmp-ssse3.S4
-rw-r--r--sysdeps/x86_64/multiarch/strspn-c.c4
-rw-r--r--sysdeps/x86_64/strcmp.S211
12 files changed, 507 insertions, 292 deletions
diff --git a/sysdeps/x86_64/cacheinfo.c b/sysdeps/x86_64/cacheinfo.c
index f252fc2c6c..5b66c62eb3 100644
--- a/sysdeps/x86_64/cacheinfo.c
+++ b/sysdeps/x86_64/cacheinfo.c
@@ -516,13 +516,15 @@ init_cacheinfo (void)
shared = handle_intel (_SC_LEVEL2_CACHE_SIZE, max_cpuid);
}
+ unsigned int ebx_1;
+
#ifdef USE_MULTIARCH
eax = __cpu_features.cpuid[COMMON_CPUID_INDEX_1].eax;
- ebx = __cpu_features.cpuid[COMMON_CPUID_INDEX_1].ebx;
+ ebx_1 = __cpu_features.cpuid[COMMON_CPUID_INDEX_1].ebx;
ecx = __cpu_features.cpuid[COMMON_CPUID_INDEX_1].ecx;
edx = __cpu_features.cpuid[COMMON_CPUID_INDEX_1].edx;
#else
- __cpuid (1, eax, ebx, ecx, edx);
+ __cpuid (1, eax, ebx_1, ecx, edx);
#endif
#ifndef DISABLE_PREFERRED_MEMORY_INSTRUCTION
@@ -554,14 +556,46 @@ init_cacheinfo (void)
}
while (((eax >> 5) & 0x7) != level);
- threads = ((eax >> 14) & 0x3ff) + 1;
+ threads = (eax >> 14) & 0x3ff;
+
+ /* If max_cpuid >= 11, THREADS is the maximum number of
+ addressable IDs for logical processors sharing the
+ cache, instead of the maximum number of threads
+ sharing the cache. */
+ if (threads && max_cpuid >= 11)
+ {
+ /* Find the number of logical processors shipped in
+ one core and apply count mask. */
+ i = 0;
+ while (1)
+ {
+ __cpuid_count (11, i++, eax, ebx, ecx, edx);
+
+ int shipped = ebx & 0xff;
+ int type = ecx & 0xff0;
+ if (shipped == 0 || type == 0)
+ break;
+ else if (type == 0x200)
+ {
+ int count_mask;
+
+ /* Compute count mask. */
+ asm ("bsr %1, %0"
+ : "=r" (count_mask) : "g" (threads));
+ count_mask = ~(-1 << (count_mask + 1));
+ threads = (shipped - 1) & count_mask;
+ break;
+ }
+ }
+ }
+ threads += 1;
}
else
{
intel_bug_no_cache_info:
/* Assume that all logical threads share the highest cache level. */
- threads = (ebx >> 16) & 0xff;
+ threads = (ebx_1 >> 16) & 0xff;
}
/* Cap usage of highest cache level to the number of supported
diff --git a/sysdeps/x86_64/dl-trampoline.S b/sysdeps/x86_64/dl-trampoline.S
index 20da6956f1..f9c60ad5cf 100644
--- a/sysdeps/x86_64/dl-trampoline.S
+++ b/sysdeps/x86_64/dl-trampoline.S
@@ -146,247 +146,17 @@ L(have_avx):
2: movl %eax, L(have_avx)(%rip)
cmpl $0, %eax
-1: js L(no_avx1)
+1: js L(no_avx)
- /* This is to support AVX audit modules. */
- vmovdqu %ymm0, (LR_VECTOR_OFFSET)(%rsp)
- vmovdqu %ymm1, (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp)
- vmovdqu %ymm2, (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp)
- vmovdqu %ymm3, (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp)
- vmovdqu %ymm4, (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp)
- vmovdqu %ymm5, (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp)
- vmovdqu %ymm6, (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp)
- vmovdqu %ymm7, (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp)
+# define RESTORE_AVX
+# include "dl-trampoline.h"
- /* Save xmm0-xmm7 registers to detect if any of them are
- changed by audit module. */
- vmovdqa %xmm0, (LR_SIZE)(%rsp)
- vmovdqa %xmm1, (LR_SIZE + XMM_SIZE)(%rsp)
- vmovdqa %xmm2, (LR_SIZE + XMM_SIZE*2)(%rsp)
- vmovdqa %xmm3, (LR_SIZE + XMM_SIZE*3)(%rsp)
- vmovdqa %xmm4, (LR_SIZE + XMM_SIZE*4)(%rsp)
- vmovdqa %xmm5, (LR_SIZE + XMM_SIZE*5)(%rsp)
- vmovdqa %xmm6, (LR_SIZE + XMM_SIZE*6)(%rsp)
- vmovdqa %xmm7, (LR_SIZE + XMM_SIZE*7)(%rsp)
-
-L(no_avx1):
-# endif
-
- movq %rsp, %rcx # La_x86_64_regs pointer to %rcx.
- movq 48(%rbx), %rdx # Load return address if needed.
- movq 40(%rbx), %rsi # Copy args pushed by PLT in register.
- movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index
- leaq 16(%rbx), %r8
- call _dl_profile_fixup # Call resolver.
-
- movq %rax, %r11 # Save return value.
-
- movq 8(%rbx), %rax # Get back register content.
- movq LR_RDX_OFFSET(%rsp), %rdx
- movq LR_R8_OFFSET(%rsp), %r8
- movq LR_R9_OFFSET(%rsp), %r9
-
- movaps (LR_XMM_OFFSET)(%rsp), %xmm0
- movaps (LR_XMM_OFFSET + XMM_SIZE)(%rsp), %xmm1
- movaps (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp), %xmm2
- movaps (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp), %xmm3
- movaps (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp), %xmm4
- movaps (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp), %xmm5
- movaps (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp), %xmm6
- movaps (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp), %xmm7
-
-# ifdef HAVE_AVX_SUPPORT
- cmpl $0, L(have_avx)(%rip)
- js L(no_avx2)
-
- /* Check if any xmm0-xmm7 registers are changed by audit
- module. */
- vpcmpeqq (LR_SIZE)(%rsp), %xmm0, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET)(%rsp), %ymm0
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp), %ymm1
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*2)(%rsp), %xmm2, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp), %ymm2
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*3)(%rsp), %xmm3, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp), %ymm3
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*4)(%rsp), %xmm4, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp), %ymm4
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*5)(%rsp), %xmm5, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp), %ymm5
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*6)(%rsp), %xmm6, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp), %ymm6
-
-1: vpcmpeqq (LR_SIZE + XMM_SIZE*7)(%rsp), %xmm7, %xmm8
- vpmovmskb %xmm8, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp), %ymm7
-
-L(no_avx2):
-1:
-# endif
- movq 16(%rbx), %r10 # Anything in framesize?
- testq %r10, %r10
- jns 3f
-
- /* There's nothing in the frame size, so there
- will be no call to the _dl_call_pltexit. */
-
- /* Get back registers content. */
- movq LR_RCX_OFFSET(%rsp), %rcx
- movq LR_RSI_OFFSET(%rsp), %rsi
- movq LR_RDI_OFFSET(%rsp), %rdi
-
- movq %rbx, %rsp
- movq (%rsp), %rbx
- cfi_restore(rbx)
- cfi_def_cfa_register(%rsp)
-
- addq $48, %rsp # Adjust the stack to the return value
- # (eats the reloc index and link_map)
- cfi_adjust_cfa_offset(-48)
- jmp *%r11 # Jump to function address.
-
-3:
- cfi_adjust_cfa_offset(48)
- cfi_rel_offset(%rbx, 0)
- cfi_def_cfa_register(%rbx)
-
- /* At this point we need to prepare new stack for the function
- which has to be called. We copy the original stack to a
- temporary buffer of the size specified by the 'framesize'
- returned from _dl_profile_fixup */
-
- leaq LR_RSP_OFFSET(%rbx), %rsi # stack
- addq $8, %r10
- andq $0xfffffffffffffff0, %r10
- movq %r10, %rcx
- subq %r10, %rsp
- movq %rsp, %rdi
- shrq $3, %rcx
- rep
- movsq
-
- movq 24(%rdi), %rcx # Get back register content.
- movq 32(%rdi), %rsi
- movq 40(%rdi), %rdi
-
- call *%r11
-
- mov 24(%rbx), %rsp # Drop the copied stack content
-
- /* Now we have to prepare the La_x86_64_retval structure for the
- _dl_call_pltexit. The La_x86_64_regs is being pointed by rsp now,
- so we just need to allocate the sizeof(La_x86_64_retval) space on
- the stack, since the alignment has already been taken care of. */
-# ifdef HAVE_AVX_SUPPORT
- /* sizeof(La_x86_64_retval). Need extra space for 2 SSE
- registers to detect if xmm0/xmm1 registers are changed
- by audit module. */
- subq $(LRV_SIZE + XMM_SIZE*2), %rsp
-# else
- subq $LRV_SIZE, %rsp # sizeof(La_x86_64_retval)
-# endif
- movq %rsp, %rcx # La_x86_64_retval argument to %rcx.
-
- /* Fill in the La_x86_64_retval structure. */
- movq %rax, LRV_RAX_OFFSET(%rcx)
- movq %rdx, LRV_RDX_OFFSET(%rcx)
-
- movaps %xmm0, LRV_XMM0_OFFSET(%rcx)
- movaps %xmm1, LRV_XMM1_OFFSET(%rcx)
-
-# ifdef HAVE_AVX_SUPPORT
- cmpl $0, L(have_avx)(%rip)
- js L(no_avx3)
-
- /* This is to support AVX audit modules. */
- vmovdqu %ymm0, LRV_VECTOR0_OFFSET(%rcx)
- vmovdqu %ymm1, LRV_VECTOR1_OFFSET(%rcx)
-
- /* Save xmm0/xmm1 registers to detect if they are changed
- by audit module. */
- vmovdqa %xmm0, (LRV_SIZE)(%rcx)
- vmovdqa %xmm1, (LRV_SIZE + XMM_SIZE)(%rcx)
-
-L(no_avx3):
-# endif
-
- fstpt LRV_ST0_OFFSET(%rcx)
- fstpt LRV_ST1_OFFSET(%rcx)
-
- movq 24(%rbx), %rdx # La_x86_64_regs argument to %rdx.
- movq 40(%rbx), %rsi # Copy args pushed by PLT in register.
- movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index
- call _dl_call_pltexit
-
- /* Restore return registers. */
- movq LRV_RAX_OFFSET(%rsp), %rax
- movq LRV_RDX_OFFSET(%rsp), %rdx
-
- movaps LRV_XMM0_OFFSET(%rsp), %xmm0
- movaps LRV_XMM1_OFFSET(%rsp), %xmm1
-
-# ifdef HAVE_AVX_SUPPORT
- cmpl $0, L(have_avx)(%rip)
- js L(no_avx4)
-
- /* Check if xmm0/xmm1 registers are changed by audit module. */
- vpcmpeqq (LRV_SIZE)(%rsp), %xmm0, %xmm2
- vpmovmskb %xmm2, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu LRV_VECTOR0_OFFSET(%rsp), %ymm0
-
-1: vpcmpeqq (LRV_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm2
- vpmovmskb %xmm2, %esi
- cmpl $0xffff, %esi
- jne 1f
- vmovdqu LRV_VECTOR1_OFFSET(%rsp), %ymm1
-
-L(no_avx4):
-1:
+ .align 16
+L(no_avx):
# endif
- fldt LRV_ST1_OFFSET(%rsp)
- fldt LRV_ST0_OFFSET(%rsp)
-
- movq %rbx, %rsp
- movq (%rsp), %rbx
- cfi_restore(rbx)
- cfi_def_cfa_register(%rsp)
-
- addq $48, %rsp # Adjust the stack to the return value
- # (eats the reloc index and link_map)
- cfi_adjust_cfa_offset(-48)
- retq
+# undef RESTORE_AVX
+# include "dl-trampoline.h"
cfi_endproc
.size _dl_runtime_profile, .-_dl_runtime_profile
diff --git a/sysdeps/x86_64/dl-trampoline.h b/sysdeps/x86_64/dl-trampoline.h
new file mode 100644
index 0000000000..5d49ed4408
--- /dev/null
+++ b/sysdeps/x86_64/dl-trampoline.h
@@ -0,0 +1,269 @@
+/* Partial PLT profile trampoline to save and restore x86-64 vector
+ registers.
+ Copyright (C) 2009 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ 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., 59 Temple Place, Suite 330, Boston, MA
+ 02111-1307 USA. */
+
+#ifdef RESTORE_AVX
+ /* This is to support AVX audit modules. */
+ vmovdqu %ymm0, (LR_VECTOR_OFFSET)(%rsp)
+ vmovdqu %ymm1, (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp)
+ vmovdqu %ymm2, (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp)
+ vmovdqu %ymm3, (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp)
+ vmovdqu %ymm4, (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp)
+ vmovdqu %ymm5, (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp)
+ vmovdqu %ymm6, (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp)
+ vmovdqu %ymm7, (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp)
+
+ /* Save xmm0-xmm7 registers to detect if any of them are
+ changed by audit module. */
+ vmovdqa %xmm0, (LR_SIZE)(%rsp)
+ vmovdqa %xmm1, (LR_SIZE + XMM_SIZE)(%rsp)
+ vmovdqa %xmm2, (LR_SIZE + XMM_SIZE*2)(%rsp)
+ vmovdqa %xmm3, (LR_SIZE + XMM_SIZE*3)(%rsp)
+ vmovdqa %xmm4, (LR_SIZE + XMM_SIZE*4)(%rsp)
+ vmovdqa %xmm5, (LR_SIZE + XMM_SIZE*5)(%rsp)
+ vmovdqa %xmm6, (LR_SIZE + XMM_SIZE*6)(%rsp)
+ vmovdqa %xmm7, (LR_SIZE + XMM_SIZE*7)(%rsp)
+#endif
+
+ movq %rsp, %rcx # La_x86_64_regs pointer to %rcx.
+ movq 48(%rbx), %rdx # Load return address if needed.
+ movq 40(%rbx), %rsi # Copy args pushed by PLT in register.
+ movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index
+ leaq 16(%rbx), %r8
+ call _dl_profile_fixup # Call resolver.
+
+ movq %rax, %r11 # Save return value.
+
+ movq 8(%rbx), %rax # Get back register content.
+ movq LR_RDX_OFFSET(%rsp), %rdx
+ movq LR_R8_OFFSET(%rsp), %r8
+ movq LR_R9_OFFSET(%rsp), %r9
+
+ movaps (LR_XMM_OFFSET)(%rsp), %xmm0
+ movaps (LR_XMM_OFFSET + XMM_SIZE)(%rsp), %xmm1
+ movaps (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp), %xmm2
+ movaps (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp), %xmm3
+ movaps (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp), %xmm4
+ movaps (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp), %xmm5
+ movaps (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp), %xmm6
+ movaps (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp), %xmm7
+
+#ifdef RESTORE_AVX
+ /* Check if any xmm0-xmm7 registers are changed by audit
+ module. */
+ vpcmpeqq (LR_SIZE)(%rsp), %xmm0, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm0, (LR_VECTOR_OFFSET)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET)(%rsp), %ymm0
+ vmovdqa %xmm0, (LR_XMM_OFFSET)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm1, (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp), %ymm1
+ vmovdqa %xmm1, (LR_XMM_OFFSET + XMM_SIZE)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*2)(%rsp), %xmm2, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm2, (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp), %ymm2
+ vmovdqa %xmm2, (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*3)(%rsp), %xmm3, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm3, (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp), %ymm3
+ vmovdqa %xmm3, (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*4)(%rsp), %xmm4, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm4, (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp), %ymm4
+ vmovdqa %xmm4, (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*5)(%rsp), %xmm5, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm5, (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp), %ymm5
+ vmovdqa %xmm5, (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*6)(%rsp), %xmm6, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm6, (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp), %ymm6
+ vmovdqa %xmm6, (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp)
+
+1: vpcmpeqq (LR_SIZE + XMM_SIZE*7)(%rsp), %xmm7, %xmm8
+ vpmovmskb %xmm8, %esi
+ cmpl $0xffff, %esi
+ je 2f
+ vmovdqa %xmm7, (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp)
+ jmp 1f
+2: vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp), %ymm7
+ vmovdqa %xmm7, (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp)
+
+1:
+#endif
+ movq 16(%rbx), %r10 # Anything in framesize?
+ testq %r10, %r10
+ jns 3f
+
+ /* There's nothing in the frame size, so there
+ will be no call to the _dl_call_pltexit. */
+
+ /* Get back registers content. */
+ movq LR_RCX_OFFSET(%rsp), %rcx
+ movq LR_RSI_OFFSET(%rsp), %rsi
+ movq LR_RDI_OFFSET(%rsp), %rdi
+
+ movq %rbx, %rsp
+ movq (%rsp), %rbx
+ cfi_restore(rbx)
+ cfi_def_cfa_register(%rsp)
+
+ addq $48, %rsp # Adjust the stack to the return value
+ # (eats the reloc index and link_map)
+ cfi_adjust_cfa_offset(-48)
+ jmp *%r11 # Jump to function address.
+
+3:
+ cfi_adjust_cfa_offset(48)
+ cfi_rel_offset(%rbx, 0)
+ cfi_def_cfa_register(%rbx)
+
+ /* At this point we need to prepare new stack for the function
+ which has to be called. We copy the original stack to a
+ temporary buffer of the size specified by the 'framesize'
+ returned from _dl_profile_fixup */
+
+ leaq LR_RSP_OFFSET(%rbx), %rsi # stack
+ addq $8, %r10
+ andq $0xfffffffffffffff0, %r10
+ movq %r10, %rcx
+ subq %r10, %rsp
+ movq %rsp, %rdi
+ shrq $3, %rcx
+ rep
+ movsq
+
+ movq 24(%rdi), %rcx # Get back register content.
+ movq 32(%rdi), %rsi
+ movq 40(%rdi), %rdi
+
+ call *%r11
+
+ mov 24(%rbx), %rsp # Drop the copied stack content
+
+ /* Now we have to prepare the La_x86_64_retval structure for the
+ _dl_call_pltexit. The La_x86_64_regs is being pointed by rsp now,
+ so we just need to allocate the sizeof(La_x86_64_retval) space on
+ the stack, since the alignment has already been taken care of. */
+# ifdef RESTORE_AVX
+ /* sizeof(La_x86_64_retval). Need extra space for 2 SSE
+ registers to detect if xmm0/xmm1 registers are changed
+ by audit module. */
+ subq $(LRV_SIZE + XMM_SIZE*2), %rsp
+# else
+ subq $LRV_SIZE, %rsp # sizeof(La_x86_64_retval)
+# endif
+ movq %rsp, %rcx # La_x86_64_retval argument to %rcx.
+
+ /* Fill in the La_x86_64_retval structure. */
+ movq %rax, LRV_RAX_OFFSET(%rcx)
+ movq %rdx, LRV_RDX_OFFSET(%rcx)
+
+ movaps %xmm0, LRV_XMM0_OFFSET(%rcx)
+ movaps %xmm1, LRV_XMM1_OFFSET(%rcx)
+
+# ifdef RESTORE_AVX
+ /* This is to support AVX audit modules. */
+ vmovdqu %ymm0, LRV_VECTOR0_OFFSET(%rcx)
+ vmovdqu %ymm1, LRV_VECTOR1_OFFSET(%rcx)
+
+ /* Save xmm0/xmm1 registers to detect if they are changed
+ by audit module. */
+ vmovdqa %xmm0, (LRV_SIZE)(%rcx)
+ vmovdqa %xmm1, (LRV_SIZE + XMM_SIZE)(%rcx)
+# endif
+
+ fstpt LRV_ST0_OFFSET(%rcx)
+ fstpt LRV_ST1_OFFSET(%rcx)
+
+ movq 24(%rbx), %rdx # La_x86_64_regs argument to %rdx.
+ movq 40(%rbx), %rsi # Copy args pushed by PLT in register.
+ movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index
+ call _dl_call_pltexit
+
+ /* Restore return registers. */
+ movq LRV_RAX_OFFSET(%rsp), %rax
+ movq LRV_RDX_OFFSET(%rsp), %rdx
+
+ movaps LRV_XMM0_OFFSET(%rsp), %xmm0
+ movaps LRV_XMM1_OFFSET(%rsp), %xmm1
+
+# ifdef RESTORE_AVX
+ /* Check if xmm0/xmm1 registers are changed by audit module. */
+ vpcmpeqq (LRV_SIZE)(%rsp), %xmm0, %xmm2
+ vpmovmskb %xmm2, %esi
+ cmpl $0xffff, %esi
+ jne 1f
+ vmovdqu LRV_VECTOR0_OFFSET(%rsp), %ymm0
+
+1: vpcmpeqq (LRV_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm2
+ vpmovmskb %xmm2, %esi
+ cmpl $0xffff, %esi
+ jne 1f
+ vmovdqu LRV_VECTOR1_OFFSET(%rsp), %ymm1
+
+1:
+# endif
+
+ fldt LRV_ST1_OFFSET(%rsp)
+ fldt LRV_ST0_OFFSET(%rsp)
+
+ movq %rbx, %rsp
+ movq (%rsp), %rbx
+ cfi_restore(rbx)
+ cfi_def_cfa_register(%rsp)
+
+ addq $48, %rsp # Adjust the stack to the return value
+ # (eats the reloc index and link_map)
+ cfi_adjust_cfa_offset(-48)
+ retq
diff --git a/sysdeps/x86_64/multiarch/Makefile b/sysdeps/x86_64/multiarch/Makefile
index b066402204..0ded3b3261 100644
--- a/sysdeps/x86_64/multiarch/Makefile
+++ b/sysdeps/x86_64/multiarch/Makefile
@@ -4,7 +4,7 @@ gen-as-const-headers += ifunc-defines.sym
endif
ifeq ($(subdir),string)
-sysdep_routines += stpncpy-c strncpy-c
+sysdep_routines += stpncpy-c strncpy-c strcmp-ssse3 strncmp-ssse3
ifeq (yes,$(config-cflags-sse4))
sysdep_routines += strcspn-c strpbrk-c strspn-c strstr-c strcasestr-c
CFLAGS-strcspn-c.c += -msse4
diff --git a/sysdeps/x86_64/multiarch/rawmemchr.S b/sysdeps/x86_64/multiarch/rawmemchr.S
index d4f265f430..08fd8769fc 100644
--- a/sysdeps/x86_64/multiarch/rawmemchr.S
+++ b/sysdeps/x86_64/multiarch/rawmemchr.S
@@ -38,6 +38,7 @@ END(rawmemchr)
strong_alias (rawmemchr, __rawmemchr)
+ .section .text.sse4.2,"ax",@progbits
.align 16
.type __rawmemchr_sse42, @function
__rawmemchr_sse42:
diff --git a/sysdeps/x86_64/multiarch/strcmp-ssse3.S b/sysdeps/x86_64/multiarch/strcmp-ssse3.S
new file mode 100644
index 0000000000..98cecb8942
--- /dev/null
+++ b/sysdeps/x86_64/multiarch/strcmp-ssse3.S
@@ -0,0 +1,3 @@
+#define USE_SSSE3 1
+#define STRCMP __strcmp_ssse3
+#include "../strcmp.S"
diff --git a/sysdeps/x86_64/multiarch/strcmp.S b/sysdeps/x86_64/multiarch/strcmp.S
index 1a315737af..05adf1e2e6 100644
--- a/sysdeps/x86_64/multiarch/strcmp.S
+++ b/sysdeps/x86_64/multiarch/strcmp.S
@@ -34,6 +34,7 @@
mov %r9, %r11
#define STRCMP_SSE42 __strncmp_sse42
+#define STRCMP_SSSE3 __strncmp_ssse3
#define STRCMP_SSE2 __strncmp_sse2
#define __GI_STRCMP __GI_strncmp
#else
@@ -41,6 +42,7 @@
#ifndef STRCMP
#define STRCMP strcmp
#define STRCMP_SSE42 __strcmp_sse42
+#define STRCMP_SSSE3 __strcmp_ssse3
#define STRCMP_SSE2 __strcmp_sse2
#define __GI_STRCMP __GI_strcmp
#endif
@@ -60,10 +62,14 @@ ENTRY(STRCMP)
cmpl $0, __cpu_features+KIND_OFFSET(%rip)
jne 1f
call __init_cpu_features
-1: leaq STRCMP_SSE2(%rip), %rax
- testl $(1<<20), __cpu_features+CPUID_OFFSET+COMMON_CPUID_INDEX_1*CPUID_SIZE+CPUID_ECX_OFFSET(%rip)
- jz 2f
+1:
leaq STRCMP_SSE42(%rip), %rax
+ testl $(1<<20), __cpu_features+CPUID_OFFSET+COMMON_CPUID_INDEX_1*CPUID_SIZE+CPUID_ECX_OFFSET(%rip)
+ jnz 2f
+ leaq STRCMP_SSSE3(%rip), %rax
+ testl $(1<<9), __cpu_features+CPUID_OFFSET+COMMON_CPUID_INDEX_1*CPUID_SIZE+CPUID_ECX_OFFSET(%rip)
+ jnz 2f
+ leaq STRCMP_SSE2(%rip), %rax
2: ret
END(STRCMP)
diff --git a/sysdeps/x86_64/multiarch/strcspn-c.c b/sysdeps/x86_64/multiarch/strcspn-c.c
index 4512267d3f..daeebe1bf5 100644
--- a/sysdeps/x86_64/multiarch/strcspn-c.c
+++ b/sysdeps/x86_64/multiarch/strcspn-c.c
@@ -86,11 +86,13 @@ STRCSPN_SSE42 (const char *s, const char *a)
const char *aligned;
__m128i mask;
+ /* Fake initialization. gcc otherwise will warn. */
+ asm ("" : "=xm" (mask));
int offset = (int) ((size_t) a & 15);
if (offset != 0)
{
/* Load masks. */
- aligned = (const char *) ((size_t) a & 0xfffffffffffffff0L);
+ aligned = (const char *) ((size_t) a & -16L);
__m128i mask0 = _mm_load_si128 ((__m128i *) aligned);
switch (offset)
@@ -229,7 +231,7 @@ STRCSPN_SSE42 (const char *s, const char *a)
if (offset != 0)
{
/* Check partial string. */
- aligned = (const char *) ((size_t) s & 0xfffffffffffffff0L);
+ aligned = (const char *) ((size_t) s & -16L);
__m128i value = _mm_load_si128 ((__m128i *) aligned);
switch (offset)
diff --git a/sysdeps/x86_64/multiarch/strlen.S b/sysdeps/x86_64/multiarch/strlen.S
index 82b03ccc28..4342c6cdab 100644
--- a/sysdeps/x86_64/multiarch/strlen.S
+++ b/sysdeps/x86_64/multiarch/strlen.S
@@ -40,6 +40,7 @@ ENTRY(strlen)
END(strlen)
+ .section .text.sse4.2,"ax",@progbits
.align 16
.type __strlen_sse42, @function
__strlen_sse42:
diff --git a/sysdeps/x86_64/multiarch/strncmp-ssse3.S b/sysdeps/x86_64/multiarch/strncmp-ssse3.S
new file mode 100644
index 0000000000..a320a3e949
--- /dev/null
+++ b/sysdeps/x86_64/multiarch/strncmp-ssse3.S
@@ -0,0 +1,4 @@
+#define USE_SSSE3 1
+#define STRCMP __strncmp_ssse3
+#define USE_AS_STRNCMP
+#include "../strcmp.S"
diff --git a/sysdeps/x86_64/multiarch/strspn-c.c b/sysdeps/x86_64/multiarch/strspn-c.c
index 5b99f0d383..be9e8ac0a8 100644
--- a/sysdeps/x86_64/multiarch/strspn-c.c
+++ b/sysdeps/x86_64/multiarch/strspn-c.c
@@ -68,7 +68,7 @@ __strspn_sse42 (const char *s, const char *a)
if (offset != 0)
{
/* Load masks. */
- aligned = (const char *) ((size_t) a & 0xfffffffffffffff0L);
+ aligned = (const char *) ((size_t) a & -16L);
__m128i mask0 = _mm_load_si128 ((__m128i *) aligned);
switch (offset)
@@ -207,7 +207,7 @@ __strspn_sse42 (const char *s, const char *a)
if (offset != 0)
{
/* Check partial string. */
- aligned = (const char *) ((size_t) s & 0xfffffffffffffff0L);
+ aligned = (const char *) ((size_t) s & -16L);
__m128i value = _mm_load_si128 ((__m128i *) aligned);
switch (offset)
diff --git a/sysdeps/x86_64/strcmp.S b/sysdeps/x86_64/strcmp.S
index 340a64ba35..650ec173b6 100644
--- a/sysdeps/x86_64/strcmp.S
+++ b/sysdeps/x86_64/strcmp.S
@@ -51,7 +51,12 @@
# endif
#endif
+#ifndef USE_SSSE3
.text
+#else
+ .section .text.ssse3,"ax",@progbits
+#endif
+
ENTRY (BP_SYM (STRCMP))
#ifdef NOT_IN_libc
/* Simple version since we can't use SSE registers in ld.so. */
@@ -244,9 +249,13 @@ LABEL(gobble_ashr_1):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4 /* store for next cycle */
+#ifndef USE_SSSE3
psrldq $1, %xmm3
pslldq $15, %xmm2
por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $1, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -269,9 +278,13 @@ LABEL(gobble_ashr_1):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4 /* store for next cycle */
+#ifndef USE_SSSE3
psrldq $1, %xmm3
- pslldq $15, %xmm2
+ pslldq $15, %xmm2
por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $1, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -363,9 +376,13 @@ LABEL(gobble_ashr_2):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $2, %xmm3
pslldq $14, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $2, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -389,9 +406,13 @@ LABEL(gobble_ashr_2):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $2, %xmm3
- pslldq $14, %xmm2
- por %xmm3, %xmm2
+ pslldq $14, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $2, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -477,9 +498,13 @@ LABEL(gobble_ashr_3):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $3, %xmm3
pslldq $13, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $3, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -503,9 +528,13 @@ LABEL(gobble_ashr_3):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $3, %xmm3
- pslldq $13, %xmm2
- por %xmm3, %xmm2
+ pslldq $13, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $3, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -591,9 +620,13 @@ LABEL(gobble_ashr_4):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $4, %xmm3
pslldq $12, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $4, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -617,9 +650,13 @@ LABEL(gobble_ashr_4):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $4, %xmm3
- pslldq $12, %xmm2
- por %xmm3, %xmm2
+ pslldq $12, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $4, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -705,9 +742,13 @@ LABEL(gobble_ashr_5):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $5, %xmm3
pslldq $11, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $5, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -731,9 +772,13 @@ LABEL(gobble_ashr_5):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $5, %xmm3
- pslldq $11, %xmm2
- por %xmm3, %xmm2
+ pslldq $11, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $5, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -819,9 +864,13 @@ LABEL(gobble_ashr_6):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $6, %xmm3
pslldq $10, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $6, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -845,9 +894,13 @@ LABEL(gobble_ashr_6):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $6, %xmm3
- pslldq $10, %xmm2
- por %xmm3, %xmm2
+ pslldq $10, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $6, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -933,9 +986,13 @@ LABEL(gobble_ashr_7):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $7, %xmm3
pslldq $9, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $7, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -959,9 +1016,13 @@ LABEL(gobble_ashr_7):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $7, %xmm3
- pslldq $9, %xmm2
- por %xmm3, %xmm2
+ pslldq $9, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $7, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1047,9 +1108,13 @@ LABEL(gobble_ashr_8):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $8, %xmm3
pslldq $8, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $8, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1073,9 +1138,13 @@ LABEL(gobble_ashr_8):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $8, %xmm3
- pslldq $8, %xmm2
- por %xmm3, %xmm2
+ pslldq $8, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $8, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1161,9 +1230,13 @@ LABEL(gobble_ashr_9):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $9, %xmm3
pslldq $7, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $9, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1187,9 +1260,13 @@ LABEL(gobble_ashr_9):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $9, %xmm3
- pslldq $7, %xmm2
- por %xmm3, %xmm2
+ pslldq $7, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $9, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1275,9 +1352,13 @@ LABEL(gobble_ashr_10):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $10, %xmm3
pslldq $6, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $10, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1301,9 +1382,13 @@ LABEL(gobble_ashr_10):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $10, %xmm3
- pslldq $6, %xmm2
- por %xmm3, %xmm2
+ pslldq $6, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $10, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1389,9 +1474,13 @@ LABEL(gobble_ashr_11):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $11, %xmm3
pslldq $5, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $11, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1415,9 +1504,13 @@ LABEL(gobble_ashr_11):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $11, %xmm3
- pslldq $5, %xmm2
- por %xmm3, %xmm2
+ pslldq $5, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $11, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1503,9 +1596,13 @@ LABEL(gobble_ashr_12):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $12, %xmm3
pslldq $4, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $12, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1529,9 +1626,13 @@ LABEL(gobble_ashr_12):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $12, %xmm3
- pslldq $4, %xmm2
- por %xmm3, %xmm2
+ pslldq $4, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $12, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1617,9 +1718,13 @@ LABEL(gobble_ashr_13):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $13, %xmm3
pslldq $3, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $13, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1643,9 +1748,13 @@ LABEL(gobble_ashr_13):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $13, %xmm3
- pslldq $3, %xmm2
- por %xmm3, %xmm2
+ pslldq $3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $13, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1731,9 +1840,13 @@ LABEL(gobble_ashr_14):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $14, %xmm3
pslldq $2, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $14, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1757,9 +1870,13 @@ LABEL(gobble_ashr_14):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $14, %xmm3
- pslldq $2, %xmm2
- por %xmm3, %xmm2
+ pslldq $2, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $14, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1847,9 +1964,13 @@ LABEL(gobble_ashr_15):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $15, %xmm3
pslldq $1, %xmm2
- por %xmm3, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $15, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
@@ -1873,9 +1994,13 @@ LABEL(gobble_ashr_15):
movdqa (%rdi, %rcx), %xmm2
movdqa %xmm2, %xmm4
+#ifndef USE_SSSE3
psrldq $15, %xmm3
- pslldq $1, %xmm2
- por %xmm3, %xmm2
+ pslldq $1, %xmm2
+ por %xmm3, %xmm2 /* merge into one 16byte value */
+#else
+ palignr $15, %xmm3, %xmm2 /* merge into one 16byte value */
+#endif
pcmpeqb %xmm1, %xmm0
pcmpeqb %xmm2, %xmm1
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-24 19:04:03 +0000