From: "H.J. Lu" <hjl.tools@gmail.com>
To: Noah Goldstein <goldstein.w.n@gmail.com>
Cc: libc-alpha@sourceware.org, andrey.kolesov@intel.com,
carlos@systemhalted.org
Subject: Re: [PATCH v4 1/3] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h
Date: Fri, 18 Nov 2022 15:25:27 -0800 [thread overview]
Message-ID: <CAMe9rOojkFq4r_uWDkhFbPFzwuUxzJ4AeVWUf4+56XTUCV728Q@mail.gmail.com> (raw)
In-Reply-To: <20221118212255.1305681-1-goldstein.w.n@gmail.com>
On Fri, Nov 18, 2022 at 1:23 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> Just reformat with the style convention used in other x86 assembler
> files. This doesn't change libm.so or libmvec.so.
> ---
> sysdeps/x86_64/fpu/svml_d_wrapper_impl.h | 474 ++++++++++----------
> sysdeps/x86_64/fpu/svml_s_wrapper_impl.h | 546 +++++++++++------------
> 2 files changed, 510 insertions(+), 510 deletions(-)
>
> diff --git a/sysdeps/x86_64/fpu/svml_d_wrapper_impl.h b/sysdeps/x86_64/fpu/svml_d_wrapper_impl.h
> index 2334713015..b03a2122b9 100644
> --- a/sysdeps/x86_64/fpu/svml_d_wrapper_impl.h
> +++ b/sysdeps/x86_64/fpu/svml_d_wrapper_impl.h
> @@ -18,273 +18,273 @@
>
> /* SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2 callee
> - subq $40, %rsp
> - cfi_adjust_cfa_offset(40)
> - movaps %xmm0, (%rsp)
> - call JUMPTARGET(\callee)
> - movsd %xmm0, 16(%rsp)
> - movsd 8(%rsp), %xmm0
> - call JUMPTARGET(\callee)
> - movsd 16(%rsp), %xmm1
> - movsd %xmm0, 24(%rsp)
> - unpcklpd %xmm0, %xmm1
> - movaps %xmm1, %xmm0
> - addq $40, %rsp
> - cfi_adjust_cfa_offset(-40)
> - ret
> + subq $40, %rsp
> + cfi_adjust_cfa_offset (40)
> + movaps %xmm0, (%rsp)
> + call JUMPTARGET(\callee)
> + movsd %xmm0, 16(%rsp)
> + movsd 8(%rsp), %xmm0
> + call JUMPTARGET(\callee)
> + movsd 16(%rsp), %xmm1
> + movsd %xmm0, 24(%rsp)
> + unpcklpd %xmm0, %xmm1
> + movaps %xmm1, %xmm0
> + addq $40, %rsp
> + cfi_adjust_cfa_offset (-40)
> + ret
> .endm
>
> /* 2 argument SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2_ff callee
> - subq $56, %rsp
> - cfi_adjust_cfa_offset(56)
> - movaps %xmm0, (%rsp)
> - movaps %xmm1, 16(%rsp)
> - call JUMPTARGET(\callee)
> - movsd %xmm0, 32(%rsp)
> - movsd 8(%rsp), %xmm0
> - movsd 24(%rsp), %xmm1
> - call JUMPTARGET(\callee)
> - movsd 32(%rsp), %xmm1
> - movsd %xmm0, 40(%rsp)
> - unpcklpd %xmm0, %xmm1
> - movaps %xmm1, %xmm0
> - addq $56, %rsp
> - cfi_adjust_cfa_offset(-56)
> - ret
> + subq $56, %rsp
> + cfi_adjust_cfa_offset (56)
> + movaps %xmm0, (%rsp)
> + movaps %xmm1, 16(%rsp)
> + call JUMPTARGET(\callee)
> + movsd %xmm0, 32(%rsp)
> + movsd 8(%rsp), %xmm0
> + movsd 24(%rsp), %xmm1
> + call JUMPTARGET(\callee)
> + movsd 32(%rsp), %xmm1
> + movsd %xmm0, 40(%rsp)
> + unpcklpd %xmm0, %xmm1
> + movaps %xmm1, %xmm0
> + addq $56, %rsp
> + cfi_adjust_cfa_offset (-56)
> + ret
> .endm
>
> /* 3 argument SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - pushq %rbx
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbx, 0)
> - movq %rdi, %rbp
> - movq %rsi, %rbx
> - subq $40, %rsp
> - cfi_adjust_cfa_offset(40)
> - leaq 16(%rsp), %rsi
> - leaq 24(%rsp), %rdi
> - movaps %xmm0, (%rsp)
> - call JUMPTARGET(\callee)
> - leaq 16(%rsp), %rsi
> - leaq 24(%rsp), %rdi
> - movsd 24(%rsp), %xmm0
> - movapd (%rsp), %xmm1
> - movsd %xmm0, 0(%rbp)
> - unpckhpd %xmm1, %xmm1
> - movsd 16(%rsp), %xmm0
> - movsd %xmm0, (%rbx)
> - movapd %xmm1, %xmm0
> - call JUMPTARGET(\callee)
> - movsd 24(%rsp), %xmm0
> - movsd %xmm0, 8(%rbp)
> - movsd 16(%rsp), %xmm0
> - movsd %xmm0, 8(%rbx)
> - addq $40, %rsp
> - cfi_adjust_cfa_offset(-40)
> - popq %rbx
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbx)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + pushq %rbx
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbx, 0)
> + movq %rdi, %rbp
> + movq %rsi, %rbx
> + subq $40, %rsp
> + cfi_adjust_cfa_offset (40)
> + leaq 16(%rsp), %rsi
> + leaq 24(%rsp), %rdi
> + movaps %xmm0, (%rsp)
> + call JUMPTARGET(\callee)
> + leaq 16(%rsp), %rsi
> + leaq 24(%rsp), %rdi
> + movsd 24(%rsp), %xmm0
> + movapd (%rsp), %xmm1
> + movsd %xmm0, 0(%rbp)
> + unpckhpd %xmm1, %xmm1
> + movsd 16(%rsp), %xmm0
> + movsd %xmm0, (%rbx)
> + movapd %xmm1, %xmm0
> + call JUMPTARGET(\callee)
> + movsd 24(%rsp), %xmm0
> + movsd %xmm0, 8(%rbp)
> + movsd 16(%rsp), %xmm0
> + movsd %xmm0, 8(%rbx)
> + addq $40, %rsp
> + cfi_adjust_cfa_offset (-40)
> + popq %rbx
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbx)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - subq $32, %rsp
> - vextractf128 $1, %ymm0, (%rsp)
> - vzeroupper
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovapd %xmm0, 16(%rsp)
> - vmovaps (%rsp), %xmm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovapd %xmm0, %xmm1
> - vmovapd 16(%rsp), %xmm0
> - vinsertf128 $1, %xmm1, %ymm0, %ymm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + subq $32, %rsp
> + vextractf128 $1, %ymm0, (%rsp)
> + vzeroupper
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovapd %xmm0, 16(%rsp)
> + vmovaps (%rsp), %xmm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovapd %xmm0, %xmm1
> + vmovapd 16(%rsp), %xmm0
> + vinsertf128 $1, %xmm1, %ymm0, %ymm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX_ff callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - subq $64, %rsp
> - vextractf128 $1, %ymm0, 16(%rsp)
> - vextractf128 $1, %ymm1, (%rsp)
> - vzeroupper
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, 32(%rsp)
> - vmovaps 16(%rsp), %xmm0
> - vmovaps (%rsp), %xmm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, %xmm1
> - vmovaps 32(%rsp), %xmm0
> - vinsertf128 $1, %xmm1, %ymm0, %ymm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + subq $64, %rsp
> + vextractf128 $1, %ymm0, 16(%rsp)
> + vextractf128 $1, %ymm1, (%rsp)
> + vzeroupper
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, 32(%rsp)
> + vmovaps 16(%rsp), %xmm0
> + vmovaps (%rsp), %xmm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, %xmm1
> + vmovaps 32(%rsp), %xmm0
> + vinsertf128 $1, %xmm1, %ymm0, %ymm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - pushq %r13
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r13, 0)
> - pushq %r14
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r14, 0)
> - subq $48, %rsp
> - movq %rsi, %r14
> - movq %rdi, %r13
> - vextractf128 $1, %ymm0, 32(%rsp)
> - vzeroupper
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps 32(%rsp), %xmm0
> - lea (%rsp), %rdi
> - lea 16(%rsp), %rsi
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovapd (%rsp), %xmm0
> - vmovapd 16(%rsp), %xmm1
> - vmovapd %xmm0, 16(%r13)
> - vmovapd %xmm1, 16(%r14)
> - addq $48, %rsp
> - popq %r14
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r14)
> - popq %r13
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r13)
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + pushq %r13
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r13, 0)
> + pushq %r14
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r14, 0)
> + subq $48, %rsp
> + movq %rsi, %r14
> + movq %rdi, %r13
> + vextractf128 $1, %ymm0, 32(%rsp)
> + vzeroupper
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps 32(%rsp), %xmm0
> + lea (%rsp), %rdi
> + lea 16(%rsp), %rsi
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovapd (%rsp), %xmm0
> + vmovapd 16(%rsp), %xmm1
> + vmovapd %xmm0, 16(%r13)
> + vmovapd %xmm1, 16(%r14)
> + addq $48, %rsp
> + popq %r14
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r14)
> + popq %r13
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r13)
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512 callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - subq $128, %rsp
> - vmovups %zmm0, (%rsp)
> - vmovupd (%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 64(%rsp)
> - vmovupd 32(%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 96(%rsp)
> - vmovups 64(%rsp), %zmm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + subq $128, %rsp
> + vmovups %zmm0, (%rsp)
> + vmovupd (%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 64(%rsp)
> + vmovupd 32(%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 96(%rsp)
> + vmovups 64(%rsp), %zmm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512_ff callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - subq $192, %rsp
> - vmovups %zmm0, (%rsp)
> - vmovups %zmm1, 64(%rsp)
> - vmovupd (%rsp), %ymm0
> - vmovupd 64(%rsp), %ymm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 128(%rsp)
> - vmovupd 32(%rsp), %ymm0
> - vmovupd 96(%rsp), %ymm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 160(%rsp)
> - vmovups 128(%rsp), %zmm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + subq $192, %rsp
> + vmovups %zmm0, (%rsp)
> + vmovups %zmm1, 64(%rsp)
> + vmovupd (%rsp), %ymm0
> + vmovupd 64(%rsp), %ymm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 128(%rsp)
> + vmovupd 32(%rsp), %ymm0
> + vmovupd 96(%rsp), %ymm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 160(%rsp)
> + vmovups 128(%rsp), %zmm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - pushq %r12
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r12, 0)
> - pushq %r13
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r13, 0)
> - subq $176, %rsp
> - movq %rsi, %r13
> - vmovups %zmm0, (%rsp)
> - movq %rdi, %r12
> - vmovupd (%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd 32(%rsp), %ymm0
> - lea 64(%rsp), %rdi
> - lea 96(%rsp), %rsi
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd 64(%rsp), %ymm0
> - vmovupd 96(%rsp), %ymm1
> - vmovupd %ymm0, 32(%r12)
> - vmovupd %ymm1, 32(%r13)
> - vzeroupper
> - addq $176, %rsp
> - popq %r13
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r13)
> - popq %r12
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r12)
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + pushq %r12
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r12, 0)
> + pushq %r13
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r13, 0)
> + subq $176, %rsp
> + movq %rsi, %r13
> + vmovups %zmm0, (%rsp)
> + movq %rdi, %r12
> + vmovupd (%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd 32(%rsp), %ymm0
> + lea 64(%rsp), %rdi
> + lea 96(%rsp), %rsi
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd 64(%rsp), %ymm0
> + vmovupd 96(%rsp), %ymm1
> + vmovupd %ymm0, 32(%r12)
> + vmovupd %ymm1, 32(%r13)
> + vzeroupper
> + addq $176, %rsp
> + popq %r13
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r13)
> + popq %r12
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r12)
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
> diff --git a/sysdeps/x86_64/fpu/svml_s_wrapper_impl.h b/sysdeps/x86_64/fpu/svml_s_wrapper_impl.h
> index c23da7ec83..cecf6c8384 100644
> --- a/sysdeps/x86_64/fpu/svml_s_wrapper_impl.h
> +++ b/sysdeps/x86_64/fpu/svml_s_wrapper_impl.h
> @@ -18,309 +18,309 @@
>
> /* SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2 callee
> - subq $40, %rsp
> - cfi_adjust_cfa_offset(40)
> - movaps %xmm0, (%rsp)
> - call JUMPTARGET(\callee)
> - movss %xmm0, 16(%rsp)
> - movss 4(%rsp), %xmm0
> - call JUMPTARGET(\callee)
> - movss %xmm0, 20(%rsp)
> - movss 8(%rsp), %xmm0
> - call JUMPTARGET(\callee)
> - movss %xmm0, 24(%rsp)
> - movss 12(%rsp), %xmm0
> - call JUMPTARGET(\callee)
> - movss 16(%rsp), %xmm3
> - movss 20(%rsp), %xmm2
> - movss 24(%rsp), %xmm1
> - movss %xmm0, 28(%rsp)
> - unpcklps %xmm1, %xmm3
> - unpcklps %xmm0, %xmm2
> - unpcklps %xmm2, %xmm3
> - movaps %xmm3, %xmm0
> - addq $40, %rsp
> - cfi_adjust_cfa_offset(-40)
> - ret
> + subq $40, %rsp
> + cfi_adjust_cfa_offset (40)
> + movaps %xmm0, (%rsp)
> + call JUMPTARGET(\callee)
> + movss %xmm0, 16(%rsp)
> + movss 4(%rsp), %xmm0
> + call JUMPTARGET(\callee)
> + movss %xmm0, 20(%rsp)
> + movss 8(%rsp), %xmm0
> + call JUMPTARGET(\callee)
> + movss %xmm0, 24(%rsp)
> + movss 12(%rsp), %xmm0
> + call JUMPTARGET(\callee)
> + movss 16(%rsp), %xmm3
> + movss 20(%rsp), %xmm2
> + movss 24(%rsp), %xmm1
> + movss %xmm0, 28(%rsp)
> + unpcklps %xmm1, %xmm3
> + unpcklps %xmm0, %xmm2
> + unpcklps %xmm2, %xmm3
> + movaps %xmm3, %xmm0
> + addq $40, %rsp
> + cfi_adjust_cfa_offset (-40)
> + ret
> .endm
>
> /* 2 argument SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2_ff callee
> - subq $56, %rsp
> - cfi_adjust_cfa_offset(56)
> - movaps %xmm0, (%rsp)
> - movaps %xmm1, 16(%rsp)
> - call JUMPTARGET(\callee)
> - movss %xmm0, 32(%rsp)
> - movss 4(%rsp), %xmm0
> - movss 20(%rsp), %xmm1
> - call JUMPTARGET(\callee)
> - movss %xmm0, 36(%rsp)
> - movss 8(%rsp), %xmm0
> - movss 24(%rsp), %xmm1
> - call JUMPTARGET(\callee)
> - movss %xmm0, 40(%rsp)
> - movss 12(%rsp), %xmm0
> - movss 28(%rsp), %xmm1
> - call JUMPTARGET(\callee)
> - movss 32(%rsp), %xmm3
> - movss 36(%rsp), %xmm2
> - movss 40(%rsp), %xmm1
> - movss %xmm0, 44(%rsp)
> - unpcklps %xmm1, %xmm3
> - unpcklps %xmm0, %xmm2
> - unpcklps %xmm2, %xmm3
> - movaps %xmm3, %xmm0
> - addq $56, %rsp
> - cfi_adjust_cfa_offset(-56)
> - ret
> + subq $56, %rsp
> + cfi_adjust_cfa_offset (56)
> + movaps %xmm0, (%rsp)
> + movaps %xmm1, 16(%rsp)
> + call JUMPTARGET(\callee)
> + movss %xmm0, 32(%rsp)
> + movss 4(%rsp), %xmm0
> + movss 20(%rsp), %xmm1
> + call JUMPTARGET(\callee)
> + movss %xmm0, 36(%rsp)
> + movss 8(%rsp), %xmm0
> + movss 24(%rsp), %xmm1
> + call JUMPTARGET(\callee)
> + movss %xmm0, 40(%rsp)
> + movss 12(%rsp), %xmm0
> + movss 28(%rsp), %xmm1
> + call JUMPTARGET(\callee)
> + movss 32(%rsp), %xmm3
> + movss 36(%rsp), %xmm2
> + movss 40(%rsp), %xmm1
> + movss %xmm0, 44(%rsp)
> + unpcklps %xmm1, %xmm3
> + unpcklps %xmm0, %xmm2
> + unpcklps %xmm2, %xmm3
> + movaps %xmm3, %xmm0
> + addq $56, %rsp
> + cfi_adjust_cfa_offset (-56)
> + ret
> .endm
>
> /* 3 argument SSE2 ISA version as wrapper to scalar. */
> .macro WRAPPER_IMPL_SSE2_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - pushq %rbx
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbx, 0)
> - movq %rdi, %rbp
> - movq %rsi, %rbx
> - subq $40, %rsp
> - cfi_adjust_cfa_offset(40)
> - leaq 24(%rsp), %rsi
> - leaq 28(%rsp), %rdi
> - movaps %xmm0, (%rsp)
> - call JUMPTARGET(\callee)
> - leaq 24(%rsp), %rsi
> - leaq 28(%rsp), %rdi
> - movss 28(%rsp), %xmm0
> - movss %xmm0, 0(%rbp)
> - movaps (%rsp), %xmm1
> - movss 24(%rsp), %xmm0
> - movss %xmm0, (%rbx)
> - movaps %xmm1, %xmm0
> - shufps $85, %xmm1, %xmm0
> - call JUMPTARGET(\callee)
> - movss 28(%rsp), %xmm0
> - leaq 24(%rsp), %rsi
> - movss %xmm0, 4(%rbp)
> - leaq 28(%rsp), %rdi
> - movaps (%rsp), %xmm1
> - movss 24(%rsp), %xmm0
> - movss %xmm0, 4(%rbx)
> - movaps %xmm1, %xmm0
> - unpckhps %xmm1, %xmm0
> - call JUMPTARGET(\callee)
> - movaps (%rsp), %xmm1
> - leaq 24(%rsp), %rsi
> - leaq 28(%rsp), %rdi
> - movss 28(%rsp), %xmm0
> - shufps $255, %xmm1, %xmm1
> - movss %xmm0, 8(%rbp)
> - movss 24(%rsp), %xmm0
> - movss %xmm0, 8(%rbx)
> - movaps %xmm1, %xmm0
> - call JUMPTARGET(\callee)
> - movss 28(%rsp), %xmm0
> - movss %xmm0, 12(%rbp)
> - movss 24(%rsp), %xmm0
> - movss %xmm0, 12(%rbx)
> - addq $40, %rsp
> - cfi_adjust_cfa_offset(-40)
> - popq %rbx
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbx)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + pushq %rbx
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbx, 0)
> + movq %rdi, %rbp
> + movq %rsi, %rbx
> + subq $40, %rsp
> + cfi_adjust_cfa_offset (40)
> + leaq 24(%rsp), %rsi
> + leaq 28(%rsp), %rdi
> + movaps %xmm0, (%rsp)
> + call JUMPTARGET(\callee)
> + leaq 24(%rsp), %rsi
> + leaq 28(%rsp), %rdi
> + movss 28(%rsp), %xmm0
> + movss %xmm0, 0(%rbp)
> + movaps (%rsp), %xmm1
> + movss 24(%rsp), %xmm0
> + movss %xmm0, (%rbx)
> + movaps %xmm1, %xmm0
> + shufps $85, %xmm1, %xmm0
> + call JUMPTARGET(\callee)
> + movss 28(%rsp), %xmm0
> + leaq 24(%rsp), %rsi
> + movss %xmm0, 4(%rbp)
> + leaq 28(%rsp), %rdi
> + movaps (%rsp), %xmm1
> + movss 24(%rsp), %xmm0
> + movss %xmm0, 4(%rbx)
> + movaps %xmm1, %xmm0
> + unpckhps %xmm1, %xmm0
> + call JUMPTARGET(\callee)
> + movaps (%rsp), %xmm1
> + leaq 24(%rsp), %rsi
> + leaq 28(%rsp), %rdi
> + movss 28(%rsp), %xmm0
> + shufps $255, %xmm1, %xmm1
> + movss %xmm0, 8(%rbp)
> + movss 24(%rsp), %xmm0
> + movss %xmm0, 8(%rbx)
> + movaps %xmm1, %xmm0
> + call JUMPTARGET(\callee)
> + movss 28(%rsp), %xmm0
> + movss %xmm0, 12(%rbp)
> + movss 24(%rsp), %xmm0
> + movss %xmm0, 12(%rbx)
> + addq $40, %rsp
> + cfi_adjust_cfa_offset (-40)
> + popq %rbx
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbx)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - subq $32, %rsp
> - vextractf128 $1, %ymm0, (%rsp)
> - vzeroupper
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, 16(%rsp)
> - vmovaps (%rsp), %xmm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, %xmm1
> - vmovaps 16(%rsp), %xmm0
> - vinsertf128 $1, %xmm1, %ymm0, %ymm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + subq $32, %rsp
> + vextractf128 $1, %ymm0, (%rsp)
> + vzeroupper
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, 16(%rsp)
> + vmovaps (%rsp), %xmm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, %xmm1
> + vmovaps 16(%rsp), %xmm0
> + vinsertf128 $1, %xmm1, %ymm0, %ymm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX_ff callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - subq $64, %rsp
> - vextractf128 $1, %ymm0, 16(%rsp)
> - vextractf128 $1, %ymm1, (%rsp)
> - vzeroupper
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, 32(%rsp)
> - vmovaps 16(%rsp), %xmm0
> - vmovaps (%rsp), %xmm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps %xmm0, %xmm1
> - vmovaps 32(%rsp), %xmm0
> - vinsertf128 $1, %xmm1, %ymm0, %ymm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + subq $64, %rsp
> + vextractf128 $1, %ymm0, 16(%rsp)
> + vextractf128 $1, %ymm1, (%rsp)
> + vzeroupper
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, 32(%rsp)
> + vmovaps 16(%rsp), %xmm0
> + vmovaps (%rsp), %xmm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps %xmm0, %xmm1
> + vmovaps 32(%rsp), %xmm0
> + vinsertf128 $1, %xmm1, %ymm0, %ymm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
> .macro WRAPPER_IMPL_AVX_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-32, %rsp
> - pushq %r13
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r13, 0)
> - pushq %r14
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%r14, 0)
> - subq $48, %rsp
> - movq %rsi, %r14
> - vmovaps %ymm0, (%rsp)
> - movq %rdi, %r13
> - vmovaps 16(%rsp), %xmm1
> - vmovaps %xmm1, 32(%rsp)
> - vzeroupper
> - vmovaps (%rsp), %xmm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps 32(%rsp), %xmm0
> - lea (%rsp), %rdi
> - lea 16(%rsp), %rsi
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps (%rsp), %xmm0
> - vmovaps 16(%rsp), %xmm1
> - vmovaps %xmm0, 16(%r13)
> - vmovaps %xmm1, 16(%r14)
> - addq $48, %rsp
> - popq %r14
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r14)
> - popq %r13
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%r13)
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-32, %rsp
> + pushq %r13
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r13, 0)
> + pushq %r14
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%r14, 0)
> + subq $48, %rsp
> + movq %rsi, %r14
> + vmovaps %ymm0, (%rsp)
> + movq %rdi, %r13
> + vmovaps 16(%rsp), %xmm1
> + vmovaps %xmm1, 32(%rsp)
> + vzeroupper
> + vmovaps (%rsp), %xmm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps 32(%rsp), %xmm0
> + lea (%rsp), %rdi
> + lea 16(%rsp), %rsi
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps (%rsp), %xmm0
> + vmovaps 16(%rsp), %xmm1
> + vmovaps %xmm0, 16(%r13)
> + vmovaps %xmm1, 16(%r14)
> + addq $48, %rsp
> + popq %r14
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r14)
> + popq %r13
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%r13)
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512 callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - subq $128, %rsp
> - vmovups %zmm0, (%rsp)
> - vmovupd (%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 64(%rsp)
> - vmovupd 32(%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovupd %ymm0, 96(%rsp)
> - vmovups 64(%rsp), %zmm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + subq $128, %rsp
> + vmovups %zmm0, (%rsp)
> + vmovupd (%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 64(%rsp)
> + vmovupd 32(%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovupd %ymm0, 96(%rsp)
> + vmovups 64(%rsp), %zmm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512_ff callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - subq $192, %rsp
> - vmovups %zmm0, (%rsp)
> - vmovups %zmm1, 64(%rsp)
> - vmovups (%rsp), %ymm0
> - vmovups 64(%rsp), %ymm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovups %ymm0, 128(%rsp)
> - vmovups 32(%rsp), %ymm0
> - vmovups 96(%rsp), %ymm1
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovups %ymm0, 160(%rsp)
> - vmovups 128(%rsp), %zmm0
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + subq $192, %rsp
> + vmovups %zmm0, (%rsp)
> + vmovups %zmm1, 64(%rsp)
> + vmovups (%rsp), %ymm0
> + vmovups 64(%rsp), %ymm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovups %ymm0, 128(%rsp)
> + vmovups 32(%rsp), %ymm0
> + vmovups 96(%rsp), %ymm1
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovups %ymm0, 160(%rsp)
> + vmovups 128(%rsp), %zmm0
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
>
> /* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
> .macro WRAPPER_IMPL_AVX512_fFF callee
> - pushq %rbp
> - cfi_adjust_cfa_offset (8)
> - cfi_rel_offset (%rbp, 0)
> - movq %rsp, %rbp
> - cfi_def_cfa_register (%rbp)
> - andq $-64, %rsp
> - pushq %r12
> - pushq %r13
> - subq $176, %rsp
> - movq %rsi, %r13
> - vmovaps %zmm0, (%rsp)
> - movq %rdi, %r12
> - vmovaps (%rsp), %ymm0
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps 32(%rsp), %ymm0
> - lea 64(%rsp), %rdi
> - lea 96(%rsp), %rsi
> - call HIDDEN_JUMPTARGET(\callee)
> - vmovaps 64(%rsp), %ymm0
> - vmovaps 96(%rsp), %ymm1
> - vmovaps %ymm0, 32(%r12)
> - vmovaps %ymm1, 32(%r13)
> - addq $176, %rsp
> - popq %r13
> - popq %r12
> - movq %rbp, %rsp
> - cfi_def_cfa_register (%rsp)
> - popq %rbp
> - cfi_adjust_cfa_offset (-8)
> - cfi_restore (%rbp)
> - ret
> + pushq %rbp
> + cfi_adjust_cfa_offset (8)
> + cfi_rel_offset (%rbp, 0)
> + movq %rsp, %rbp
> + cfi_def_cfa_register (%rbp)
> + andq $-64, %rsp
> + pushq %r12
> + pushq %r13
> + subq $176, %rsp
> + movq %rsi, %r13
> + vmovaps %zmm0, (%rsp)
> + movq %rdi, %r12
> + vmovaps (%rsp), %ymm0
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps 32(%rsp), %ymm0
> + lea 64(%rsp), %rdi
> + lea 96(%rsp), %rsi
> + call HIDDEN_JUMPTARGET(\callee)
> + vmovaps 64(%rsp), %ymm0
> + vmovaps 96(%rsp), %ymm1
> + vmovaps %ymm0, 32(%r12)
> + vmovaps %ymm1, 32(%r13)
> + addq $176, %rsp
> + popq %r13
> + popq %r12
> + movq %rbp, %rsp
> + cfi_def_cfa_register (%rsp)
> + popq %rbp
> + cfi_adjust_cfa_offset (-8)
> + cfi_restore (%rbp)
> + ret
> .endm
> --
> 2.34.1
>
LGTM.
Thanks.
--
H.J.
next prev parent reply other threads:[~2022-11-18 23:26 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-18 19:08 [PATCH v2 1/2] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h.S Noah Goldstein
2022-11-18 19:08 ` [PATCH v2 2/2] x86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h.S Noah Goldstein
2022-11-18 19:56 ` H.J. Lu
2022-11-18 20:36 ` Noah Goldstein
2022-11-18 20:35 ` [PATCH v3 1/2] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-18 20:35 ` [PATCH v3 2/2] x86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-18 20:49 ` H.J. Lu
2022-11-18 21:23 ` Noah Goldstein
2022-11-18 21:22 ` [PATCH v4 1/3] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-18 21:22 ` [PATCH v4 2/3] x86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-19 0:06 ` H.J. Lu
2022-11-19 0:13 ` Noah Goldstein
2022-11-18 21:22 ` [PATCH v4 3/3] x86/fpu: Factor out shared avx2/avx512 " Noah Goldstein
2022-11-18 21:27 ` H.J. Lu
2022-11-18 21:59 ` Noah Goldstein
2022-11-19 0:07 ` H.J. Lu
2022-11-18 23:25 ` H.J. Lu [this message]
2022-11-19 0:13 ` [PATCH v5 1/3] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-19 0:13 ` [PATCH v5 2/3] x86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h Noah Goldstein
2022-11-19 0:36 ` H.J. Lu
2022-11-19 0:13 ` [PATCH v5 3/3] x86/fpu: Factor out shared avx2/avx512 " Noah Goldstein
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAMe9rOojkFq4r_uWDkhFbPFzwuUxzJ4AeVWUf4+56XTUCV728Q@mail.gmail.com \
--to=hjl.tools@gmail.com \
--cc=andrey.kolesov@intel.com \
--cc=carlos@systemhalted.org \
--cc=goldstein.w.n@gmail.com \
--cc=libc-alpha@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).