From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <SRS0=NS/1=3S=gmail.com=hjl.tools@sourceware.org>
Received: from mail-oa1-x2b.google.com (mail-oa1-x2b.google.com [IPv6:2001:4860:4864:20::2b])
	by sourceware.org (Postfix) with ESMTPS id 89749384D9AB
	for <libc-alpha@sourceware.org>; Fri, 18 Nov 2022 23:26:04 +0000 (GMT)
DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 89749384D9AB
Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=gmail.com
Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=gmail.com
Received: by mail-oa1-x2b.google.com with SMTP id 586e51a60fabf-13be3ef361dso7682980fac.12
        for <libc-alpha@sourceware.org>; Fri, 18 Nov 2022 15:26:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20210112;
        h=cc:to:subject:message-id:date:from:in-reply-to:references
         :mime-version:from:to:cc:subject:date:message-id:reply-to;
        bh=OGw2fy2DmwkimUlYEvZyyuoQY5GJckjcn093XH13/IU=;
        b=kthyvR3lp7X6gxNjEOMt9SUVq/i7j4pRjJdM4s1stC23JxR2axGuxeInFYwdHDtE0n
         4FqQ7k8drnL/ZhTBYcwvKVO3Kmt6CmC+oCyCzNM4dV/h5GPjurhtEzyvNSpa4rLJQOaC
         vTxTYr7ics/rbGslsqRSg5rSv+T+BqQuS46vXItFj31HLlVLGZ3caaOVQ9QEfpJi0W+9
         sqHg36D0UfS2JvzGe3w0FpP/dwA5LLpFB5VMvpyu1Ng2jIAi0CW3gWAZVyv8Ek6PTkxY
         bXViR+cnjjDg+fSDPg5lwjLzT2pvpNLpEPOEHtJf7yd7OflYR+eBRviNUHehnRuaqe/W
         +dMQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=1e100.net; s=20210112;
        h=cc:to:subject:message-id:date:from:in-reply-to:references
         :mime-version:x-gm-message-state:from:to:cc:subject:date:message-id
         :reply-to;
        bh=OGw2fy2DmwkimUlYEvZyyuoQY5GJckjcn093XH13/IU=;
        b=fA0v/I6LFWxxkgMYOwQxK8WIOduFHx4/PxYQZGgSEPd9eCcKy+3/rxPo2oYp6DzAU7
         hA2LR7z2FezrDEMLUV2txfHQRrDQ5EUSAs2E+al9Dlg+hifxzWF6cEd5aSOQCiZKGSDQ
         MQneoffrLcwWVSywdyYlhFZGEQRWuddTdq7Ztn74DJhOXQ3HBwnjpRtHpnUGx9WQkAmm
         r+itmlhK2tI5RRRMaFNgMOuQLoPm/eVoXYFqLpb4e8d9Lu3BuQnGCMZYg7XF4FAMJNph
         XVdSbi4zywKupKlr8EoHVdsXVuDdtY1nrm25nAHWR5xOuVIuER+h8tL9JeFkNrX+n3y2
         icKw==
X-Gm-Message-State: ANoB5pmGE3eNW8qobpmWraiGPWQp9HoN+InTAzbJiITJjnkBn1Zr6qMI
	gGaKqjhCNXq3qd8E/qKnVF+lm8wm2Nu0GvoOJX0=
X-Google-Smtp-Source: AA0mqf67jumgD7CpZvT4NtkbH8XFW4VWzFBPQkfWKB16l0yGdimp2oj5RaMjoeOzB4H1Vyg0Z4KuPH2eJ4p8lK3rbxM=
X-Received: by 2002:a05:6870:cb45:b0:13b:8bb8:5c5b with SMTP id
 oz5-20020a056870cb4500b0013b8bb85c5bmr5594088oab.298.1668813963397; Fri, 18
 Nov 2022 15:26:03 -0800 (PST)
MIME-Version: 1.0
References: <20221118190835.1033248-1-goldstein.w.n@gmail.com> <20221118212255.1305681-1-goldstein.w.n@gmail.com>
In-Reply-To: <20221118212255.1305681-1-goldstein.w.n@gmail.com>
From: "H.J. Lu" <hjl.tools@gmail.com>
Date: Fri, 18 Nov 2022 15:25:27 -0800
Message-ID: <CAMe9rOojkFq4r_uWDkhFbPFzwuUxzJ4AeVWUf4+56XTUCV728Q@mail.gmail.com>
Subject: Re: [PATCH v4 1/3] x86/fpu: Reformat svml_{s|d}_wrapper_impl.h
To: Noah Goldstein <goldstein.w.n@gmail.com>
Cc: libc-alpha@sourceware.org, andrey.kolesov@intel.com, 
	carlos@systemhalted.org
Content-Type: text/plain; charset="UTF-8"
X-Spam-Status: No, score=-3023.3 required=5.0 tests=BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,FREEMAIL_FROM,GIT_PATCH_0,RCVD_IN_DNSWL_NONE,SPF_HELO_NONE,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org
List-Id: <libc-alpha.sourceware.org>

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.