qiurui
/
glibc
mirror of git://sourceware.org/git/glibc.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

x86/fpu: Reformat svml_{s|d}_wrapper_impl.h 

Just reformat with the style convention used in other x86 assembler
files.  This doesn't change libm.so or libmvec.so.
This commit is contained in:
Noah Goldstein 2022-11-18 16:13:30 -08:00
parent 95177b78ff
commit d371be4b11
2 changed files with 510 additions and 510 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

474
sysdeps/x86_64/fpu/svml_d_wrapper_impl.h
View File

@ -18,273 +18,273 @@
/* SSE2 ISA version as wrapper to scalar. */ /* SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2 callee .macro WRAPPER_IMPL_SSE2 callee
subq $40, %rsp subq $40, %rsp
cfi_adjust_cfa_offset(40) cfi_adjust_cfa_offset (40)
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movsd %xmm0, 16(%rsp) movsd %xmm0, 16(%rsp)
movsd 8(%rsp), %xmm0 movsd 8(%rsp), %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movsd 16(%rsp), %xmm1 movsd 16(%rsp), %xmm1
movsd %xmm0, 24(%rsp) movsd %xmm0, 24(%rsp)
unpcklpd %xmm0, %xmm1 unpcklpd %xmm0, %xmm1
movaps %xmm1, %xmm0 movaps %xmm1, %xmm0
addq $40, %rsp addq $40, %rsp
cfi_adjust_cfa_offset(-40) cfi_adjust_cfa_offset (-40)
ret ret
.endm .endm
/* 2 argument SSE2 ISA version as wrapper to scalar. */ /* 2 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_ff callee .macro WRAPPER_IMPL_SSE2_ff callee
subq $56, %rsp subq $56, %rsp
cfi_adjust_cfa_offset(56) cfi_adjust_cfa_offset (56)
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
movaps %xmm1, 16(%rsp) movaps %xmm1, 16(%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movsd %xmm0, 32(%rsp) movsd %xmm0, 32(%rsp)
movsd 8(%rsp), %xmm0 movsd 8(%rsp), %xmm0
movsd 24(%rsp), %xmm1 movsd 24(%rsp), %xmm1
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movsd 32(%rsp), %xmm1 movsd 32(%rsp), %xmm1
movsd %xmm0, 40(%rsp) movsd %xmm0, 40(%rsp)
unpcklpd %xmm0, %xmm1 unpcklpd %xmm0, %xmm1
movaps %xmm1, %xmm0 movaps %xmm1, %xmm0
addq $56, %rsp addq $56, %rsp
cfi_adjust_cfa_offset(-56) cfi_adjust_cfa_offset (-56)
ret ret
.endm .endm
/* 3 argument SSE2 ISA version as wrapper to scalar. */ /* 3 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_fFF callee .macro WRAPPER_IMPL_SSE2_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
pushq %rbx pushq %rbx
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbx, 0) cfi_rel_offset (%rbx, 0)
movq %rdi, %rbp movq %rdi, %rbp
movq %rsi, %rbx movq %rsi, %rbx
subq $40, %rsp subq $40, %rsp
cfi_adjust_cfa_offset(40) cfi_adjust_cfa_offset (40)
leaq 16(%rsp), %rsi leaq 16(%rsp), %rsi
leaq 24(%rsp), %rdi leaq 24(%rsp), %rdi
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
leaq 16(%rsp), %rsi leaq 16(%rsp), %rsi
leaq 24(%rsp), %rdi leaq 24(%rsp), %rdi
movsd 24(%rsp), %xmm0 movsd 24(%rsp), %xmm0
movapd (%rsp), %xmm1 movapd (%rsp), %xmm1
movsd %xmm0, 0(%rbp) movsd %xmm0, 0(%rbp)
unpckhpd %xmm1, %xmm1 unpckhpd %xmm1, %xmm1
movsd 16(%rsp), %xmm0 movsd 16(%rsp), %xmm0
movsd %xmm0, (%rbx) movsd %xmm0, (%rbx)
movapd %xmm1, %xmm0 movapd %xmm1, %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movsd 24(%rsp), %xmm0 movsd 24(%rsp), %xmm0
movsd %xmm0, 8(%rbp) movsd %xmm0, 8(%rbp)
movsd 16(%rsp), %xmm0 movsd 16(%rsp), %xmm0
movsd %xmm0, 8(%rbx) movsd %xmm0, 8(%rbx)
addq $40, %rsp addq $40, %rsp
cfi_adjust_cfa_offset(-40) cfi_adjust_cfa_offset (-40)
popq %rbx popq %rbx
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbx) cfi_restore (%rbx)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX callee .macro WRAPPER_IMPL_AVX callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
subq $32, %rsp subq $32, %rsp
vextractf128 $1, %ymm0, (%rsp) vextractf128 $1, %ymm0, (%rsp)
vzeroupper vzeroupper
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovapd %xmm0, 16(%rsp) vmovapd %xmm0, 16(%rsp)
vmovaps (%rsp), %xmm0 vmovaps (%rsp), %xmm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovapd %xmm0, %xmm1 vmovapd %xmm0, %xmm1
vmovapd 16(%rsp), %xmm0 vmovapd 16(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0 vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_ff callee .macro WRAPPER_IMPL_AVX_ff callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
subq $64, %rsp subq $64, %rsp
vextractf128 $1, %ymm0, 16(%rsp) vextractf128 $1, %ymm0, 16(%rsp)
vextractf128 $1, %ymm1, (%rsp) vextractf128 $1, %ymm1, (%rsp)
vzeroupper vzeroupper
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, 32(%rsp) vmovaps %xmm0, 32(%rsp)
vmovaps 16(%rsp), %xmm0 vmovaps 16(%rsp), %xmm0
vmovaps (%rsp), %xmm1 vmovaps (%rsp), %xmm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, %xmm1 vmovaps %xmm0, %xmm1
vmovaps 32(%rsp), %xmm0 vmovaps 32(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0 vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_fFF callee .macro WRAPPER_IMPL_AVX_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
pushq %r13 pushq %r13
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r13, 0) cfi_rel_offset (%r13, 0)
pushq %r14 pushq %r14
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r14, 0) cfi_rel_offset (%r14, 0)
subq $48, %rsp subq $48, %rsp
movq %rsi, %r14 movq %rsi, %r14
movq %rdi, %r13 movq %rdi, %r13
vextractf128 $1, %ymm0, 32(%rsp) vextractf128 $1, %ymm0, 32(%rsp)
vzeroupper vzeroupper
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps 32(%rsp), %xmm0 vmovaps 32(%rsp), %xmm0
lea (%rsp), %rdi lea (%rsp), %rdi
lea 16(%rsp), %rsi lea 16(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovapd (%rsp), %xmm0 vmovapd (%rsp), %xmm0
vmovapd 16(%rsp), %xmm1 vmovapd 16(%rsp), %xmm1
vmovapd %xmm0, 16(%r13) vmovapd %xmm0, 16(%r13)
vmovapd %xmm1, 16(%r14) vmovapd %xmm1, 16(%r14)
addq $48, %rsp addq $48, %rsp
popq %r14 popq %r14
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r14) cfi_restore (%r14)
popq %r13 popq %r13
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r13) cfi_restore (%r13)
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* AVX512 ISA version as wrapper to AVX2 ISA version. */ /* AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512 callee .macro WRAPPER_IMPL_AVX512 callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
subq $128, %rsp subq $128, %rsp
vmovups %zmm0, (%rsp) vmovups %zmm0, (%rsp)
vmovupd (%rsp), %ymm0 vmovupd (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 64(%rsp) vmovupd %ymm0, 64(%rsp)
vmovupd 32(%rsp), %ymm0 vmovupd 32(%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 96(%rsp) vmovupd %ymm0, 96(%rsp)
vmovups 64(%rsp), %zmm0 vmovups 64(%rsp), %zmm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */ /* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_ff callee .macro WRAPPER_IMPL_AVX512_ff callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
subq $192, %rsp subq $192, %rsp
vmovups %zmm0, (%rsp) vmovups %zmm0, (%rsp)
vmovups %zmm1, 64(%rsp) vmovups %zmm1, 64(%rsp)
vmovupd (%rsp), %ymm0 vmovupd (%rsp), %ymm0
vmovupd 64(%rsp), %ymm1 vmovupd 64(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 128(%rsp) vmovupd %ymm0, 128(%rsp)
vmovupd 32(%rsp), %ymm0 vmovupd 32(%rsp), %ymm0
vmovupd 96(%rsp), %ymm1 vmovupd 96(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 160(%rsp) vmovupd %ymm0, 160(%rsp)
vmovups 128(%rsp), %zmm0 vmovups 128(%rsp), %zmm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */ /* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_fFF callee .macro WRAPPER_IMPL_AVX512_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
pushq %r12 pushq %r12
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r12, 0) cfi_rel_offset (%r12, 0)
pushq %r13 pushq %r13
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r13, 0) cfi_rel_offset (%r13, 0)
subq $176, %rsp subq $176, %rsp
movq %rsi, %r13 movq %rsi, %r13
vmovups %zmm0, (%rsp) vmovups %zmm0, (%rsp)
movq %rdi, %r12 movq %rdi, %r12
vmovupd (%rsp), %ymm0 vmovupd (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd 32(%rsp), %ymm0 vmovupd 32(%rsp), %ymm0
lea 64(%rsp), %rdi lea 64(%rsp), %rdi
lea 96(%rsp), %rsi lea 96(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd 64(%rsp), %ymm0 vmovupd 64(%rsp), %ymm0
vmovupd 96(%rsp), %ymm1 vmovupd 96(%rsp), %ymm1
vmovupd %ymm0, 32(%r12) vmovupd %ymm0, 32(%r12)
vmovupd %ymm1, 32(%r13) vmovupd %ymm1, 32(%r13)
vzeroupper vzeroupper
addq $176, %rsp addq $176, %rsp
popq %r13 popq %r13
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r13) cfi_restore (%r13)
popq %r12 popq %r12
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r12) cfi_restore (%r12)
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm

546
sysdeps/x86_64/fpu/svml_s_wrapper_impl.h
View File

@ -18,309 +18,309 @@
/* SSE2 ISA version as wrapper to scalar. */ /* SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2 callee .macro WRAPPER_IMPL_SSE2 callee
subq $40, %rsp subq $40, %rsp
cfi_adjust_cfa_offset(40) cfi_adjust_cfa_offset (40)
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 16(%rsp) movss %xmm0, 16(%rsp)
movss 4(%rsp), %xmm0 movss 4(%rsp), %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 20(%rsp) movss %xmm0, 20(%rsp)
movss 8(%rsp), %xmm0 movss 8(%rsp), %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 24(%rsp) movss %xmm0, 24(%rsp)
movss 12(%rsp), %xmm0 movss 12(%rsp), %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss 16(%rsp), %xmm3 movss 16(%rsp), %xmm3
movss 20(%rsp), %xmm2 movss 20(%rsp), %xmm2
movss 24(%rsp), %xmm1 movss 24(%rsp), %xmm1
movss %xmm0, 28(%rsp) movss %xmm0, 28(%rsp)
unpcklps %xmm1, %xmm3 unpcklps %xmm1, %xmm3
unpcklps %xmm0, %xmm2 unpcklps %xmm0, %xmm2
unpcklps %xmm2, %xmm3 unpcklps %xmm2, %xmm3
movaps %xmm3, %xmm0 movaps %xmm3, %xmm0
addq $40, %rsp addq $40, %rsp
cfi_adjust_cfa_offset(-40) cfi_adjust_cfa_offset (-40)
ret ret
.endm .endm
/* 2 argument SSE2 ISA version as wrapper to scalar. */ /* 2 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_ff callee .macro WRAPPER_IMPL_SSE2_ff callee
subq $56, %rsp subq $56, %rsp
cfi_adjust_cfa_offset(56) cfi_adjust_cfa_offset (56)
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
movaps %xmm1, 16(%rsp) movaps %xmm1, 16(%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 32(%rsp) movss %xmm0, 32(%rsp)
movss 4(%rsp), %xmm0 movss 4(%rsp), %xmm0
movss 20(%rsp), %xmm1 movss 20(%rsp), %xmm1
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 36(%rsp) movss %xmm0, 36(%rsp)
movss 8(%rsp), %xmm0 movss 8(%rsp), %xmm0
movss 24(%rsp), %xmm1 movss 24(%rsp), %xmm1
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss %xmm0, 40(%rsp) movss %xmm0, 40(%rsp)
movss 12(%rsp), %xmm0 movss 12(%rsp), %xmm0
movss 28(%rsp), %xmm1 movss 28(%rsp), %xmm1
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss 32(%rsp), %xmm3 movss 32(%rsp), %xmm3
movss 36(%rsp), %xmm2 movss 36(%rsp), %xmm2
movss 40(%rsp), %xmm1 movss 40(%rsp), %xmm1
movss %xmm0, 44(%rsp) movss %xmm0, 44(%rsp)
unpcklps %xmm1, %xmm3 unpcklps %xmm1, %xmm3
unpcklps %xmm0, %xmm2 unpcklps %xmm0, %xmm2
unpcklps %xmm2, %xmm3 unpcklps %xmm2, %xmm3
movaps %xmm3, %xmm0 movaps %xmm3, %xmm0
addq $56, %rsp addq $56, %rsp
cfi_adjust_cfa_offset(-56) cfi_adjust_cfa_offset (-56)
ret ret
.endm .endm
/* 3 argument SSE2 ISA version as wrapper to scalar. */ /* 3 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_fFF callee .macro WRAPPER_IMPL_SSE2_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
pushq %rbx pushq %rbx
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbx, 0) cfi_rel_offset (%rbx, 0)
movq %rdi, %rbp movq %rdi, %rbp
movq %rsi, %rbx movq %rsi, %rbx
subq $40, %rsp subq $40, %rsp
cfi_adjust_cfa_offset(40) cfi_adjust_cfa_offset (40)
leaq 24(%rsp), %rsi leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi leaq 28(%rsp), %rdi
movaps %xmm0, (%rsp) movaps %xmm0, (%rsp)
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
leaq 24(%rsp), %rsi leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi leaq 28(%rsp), %rdi
movss 28(%rsp), %xmm0 movss 28(%rsp), %xmm0
movss %xmm0, 0(%rbp) movss %xmm0, 0(%rbp)
movaps (%rsp), %xmm1 movaps (%rsp), %xmm1
movss 24(%rsp), %xmm0 movss 24(%rsp), %xmm0
movss %xmm0, (%rbx) movss %xmm0, (%rbx)
movaps %xmm1, %xmm0 movaps %xmm1, %xmm0
shufps $85, %xmm1, %xmm0 shufps $85, %xmm1, %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss 28(%rsp), %xmm0 movss 28(%rsp), %xmm0
leaq 24(%rsp), %rsi leaq 24(%rsp), %rsi
movss %xmm0, 4(%rbp) movss %xmm0, 4(%rbp)
leaq 28(%rsp), %rdi leaq 28(%rsp), %rdi
movaps (%rsp), %xmm1 movaps (%rsp), %xmm1
movss 24(%rsp), %xmm0 movss 24(%rsp), %xmm0
movss %xmm0, 4(%rbx) movss %xmm0, 4(%rbx)
movaps %xmm1, %xmm0 movaps %xmm1, %xmm0
unpckhps %xmm1, %xmm0 unpckhps %xmm1, %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movaps (%rsp), %xmm1 movaps (%rsp), %xmm1
leaq 24(%rsp), %rsi leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi leaq 28(%rsp), %rdi
movss 28(%rsp), %xmm0 movss 28(%rsp), %xmm0
shufps $255, %xmm1, %xmm1 shufps $255, %xmm1, %xmm1
movss %xmm0, 8(%rbp) movss %xmm0, 8(%rbp)
movss 24(%rsp), %xmm0 movss 24(%rsp), %xmm0
movss %xmm0, 8(%rbx) movss %xmm0, 8(%rbx)
movaps %xmm1, %xmm0 movaps %xmm1, %xmm0
call JUMPTARGET(\callee) call JUMPTARGET(\callee)
movss 28(%rsp), %xmm0 movss 28(%rsp), %xmm0
movss %xmm0, 12(%rbp) movss %xmm0, 12(%rbp)
movss 24(%rsp), %xmm0 movss 24(%rsp), %xmm0
movss %xmm0, 12(%rbx) movss %xmm0, 12(%rbx)
addq $40, %rsp addq $40, %rsp
cfi_adjust_cfa_offset(-40) cfi_adjust_cfa_offset (-40)
popq %rbx popq %rbx
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbx) cfi_restore (%rbx)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX callee .macro WRAPPER_IMPL_AVX callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
subq $32, %rsp subq $32, %rsp
vextractf128 $1, %ymm0, (%rsp) vextractf128 $1, %ymm0, (%rsp)
vzeroupper vzeroupper
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, 16(%rsp) vmovaps %xmm0, 16(%rsp)
vmovaps (%rsp), %xmm0 vmovaps (%rsp), %xmm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, %xmm1 vmovaps %xmm0, %xmm1
vmovaps 16(%rsp), %xmm0 vmovaps 16(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0 vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_ff callee .macro WRAPPER_IMPL_AVX_ff callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
subq $64, %rsp subq $64, %rsp
vextractf128 $1, %ymm0, 16(%rsp) vextractf128 $1, %ymm0, 16(%rsp)
vextractf128 $1, %ymm1, (%rsp) vextractf128 $1, %ymm1, (%rsp)
vzeroupper vzeroupper
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, 32(%rsp) vmovaps %xmm0, 32(%rsp)
vmovaps 16(%rsp), %xmm0 vmovaps 16(%rsp), %xmm0
vmovaps (%rsp), %xmm1 vmovaps (%rsp), %xmm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, %xmm1 vmovaps %xmm0, %xmm1
vmovaps 32(%rsp), %xmm0 vmovaps 32(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0 vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */ /* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_fFF callee .macro WRAPPER_IMPL_AVX_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-32, %rsp andq $-32, %rsp
pushq %r13 pushq %r13
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r13, 0) cfi_rel_offset (%r13, 0)
pushq %r14 pushq %r14
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r14, 0) cfi_rel_offset (%r14, 0)
subq $48, %rsp subq $48, %rsp
movq %rsi, %r14 movq %rsi, %r14
vmovaps %ymm0, (%rsp) vmovaps %ymm0, (%rsp)
movq %rdi, %r13 movq %rdi, %r13
vmovaps 16(%rsp), %xmm1 vmovaps 16(%rsp), %xmm1
vmovaps %xmm1, 32(%rsp) vmovaps %xmm1, 32(%rsp)
vzeroupper vzeroupper
vmovaps (%rsp), %xmm0 vmovaps (%rsp), %xmm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps 32(%rsp), %xmm0 vmovaps 32(%rsp), %xmm0
lea (%rsp), %rdi lea (%rsp), %rdi
lea 16(%rsp), %rsi lea 16(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps (%rsp), %xmm0 vmovaps (%rsp), %xmm0
vmovaps 16(%rsp), %xmm1 vmovaps 16(%rsp), %xmm1
vmovaps %xmm0, 16(%r13) vmovaps %xmm0, 16(%r13)
vmovaps %xmm1, 16(%r14) vmovaps %xmm1, 16(%r14)
addq $48, %rsp addq $48, %rsp
popq %r14 popq %r14
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r14) cfi_restore (%r14)
popq %r13 popq %r13
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%r13) cfi_restore (%r13)
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* AVX512 ISA version as wrapper to AVX2 ISA version. */ /* AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512 callee .macro WRAPPER_IMPL_AVX512 callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
subq $128, %rsp subq $128, %rsp
vmovups %zmm0, (%rsp) vmovups %zmm0, (%rsp)
vmovupd (%rsp), %ymm0 vmovupd (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 64(%rsp) vmovupd %ymm0, 64(%rsp)
vmovupd 32(%rsp), %ymm0 vmovupd 32(%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 96(%rsp) vmovupd %ymm0, 96(%rsp)
vmovups 64(%rsp), %zmm0 vmovups 64(%rsp), %zmm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */ /* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_ff callee .macro WRAPPER_IMPL_AVX512_ff callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
subq $192, %rsp subq $192, %rsp
vmovups %zmm0, (%rsp) vmovups %zmm0, (%rsp)
vmovups %zmm1, 64(%rsp) vmovups %zmm1, 64(%rsp)
vmovups (%rsp), %ymm0 vmovups (%rsp), %ymm0
vmovups 64(%rsp), %ymm1 vmovups 64(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovups %ymm0, 128(%rsp) vmovups %ymm0, 128(%rsp)
vmovups 32(%rsp), %ymm0 vmovups 32(%rsp), %ymm0
vmovups 96(%rsp), %ymm1 vmovups 96(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovups %ymm0, 160(%rsp) vmovups %ymm0, 160(%rsp)
vmovups 128(%rsp), %zmm0 vmovups 128(%rsp), %zmm0
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm
/* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */ /* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_fFF callee .macro WRAPPER_IMPL_AVX512_fFF callee
pushq %rbp pushq %rbp
cfi_adjust_cfa_offset (8) cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0) cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp movq %rsp, %rbp
cfi_def_cfa_register (%rbp) cfi_def_cfa_register (%rbp)
andq $-64, %rsp andq $-64, %rsp
pushq %r12 pushq %r12
pushq %r13 pushq %r13
subq $176, %rsp subq $176, %rsp
movq %rsi, %r13 movq %rsi, %r13
vmovaps %zmm0, (%rsp) vmovaps %zmm0, (%rsp)
movq %rdi, %r12 movq %rdi, %r12
vmovaps (%rsp), %ymm0 vmovaps (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps 32(%rsp), %ymm0 vmovaps 32(%rsp), %ymm0
lea 64(%rsp), %rdi lea 64(%rsp), %rdi
lea 96(%rsp), %rsi lea 96(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee) call HIDDEN_JUMPTARGET(\callee)
vmovaps 64(%rsp), %ymm0 vmovaps 64(%rsp), %ymm0
vmovaps 96(%rsp), %ymm1 vmovaps 96(%rsp), %ymm1
vmovaps %ymm0, 32(%r12) vmovaps %ymm0, 32(%r12)
vmovaps %ymm1, 32(%r13) vmovaps %ymm1, 32(%r13)
addq $176, %rsp addq $176, %rsp
popq %r13 popq %r13
popq %r12 popq %r12
movq %rbp, %rsp movq %rbp, %rsp
cfi_def_cfa_register (%rsp) cfi_def_cfa_register (%rsp)
popq %rbp popq %rbp
cfi_adjust_cfa_offset (-8) cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp) cfi_restore (%rbp)
ret ret
.endm .endm