From: "H.J. Lu" <hjl.tools@gmail.com>
To: Noah Goldstein <goldstein.w.n@gmail.com>
Cc: GNU C Library <libc-alpha@sourceware.org>,
"Carlos O'Donell" <carlos@systemhalted.org>
Subject: Re: [PATCH v9 1/6] x86: Update VEC macros to complete API for evex/evex512 impls
Date: Fri, 14 Oct 2022 19:45:22 -0700 [thread overview]
Message-ID: <CAMe9rOp9H+o=b9eTn3OgTn8d5EYdeykWuEh05hgUFdjH8Jyzxg@mail.gmail.com> (raw)
In-Reply-To: <20221015002100.129511-1-goldstein.w.n@gmail.com>
On Fri, Oct 14, 2022 at 5:21 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> 1) Copy so that backport will be easier.
> 2) Make section only define if there is not a previous definition
> 3) Add `VEC_lo` definition for proper reg-width but in the
> ymm/zmm0-15 range.
> 4) Add macros for accessing GPRs based on VEC_SIZE
> This is to make it easier to do think like:
> ```
> vpcmpb %VEC(0), %VEC(1), %k0
> kmov{d|q} %k0, %{eax|rax}
> test %{eax|rax}
> ```
> It adds macro s.t any GPR can get the proper width with:
> `V{upcase_GPR_name}`
>
> and any mask insn can get the proper width with:
> `{upcase_mask_insn_without_postfix}`
>
> This commit does not change libc.so
>
> Tested build on x86-64
> ---
> sysdeps/x86_64/multiarch/reg-macros.h | 168 ++++++++++++++++++
> .../multiarch/scripts/gen-reg-macros.py | 133 ++++++++++++++
> sysdeps/x86_64/multiarch/x86-avx-rtm-vecs.h | 35 ++++
> sysdeps/x86_64/multiarch/x86-avx-vecs.h | 47 +++++
> .../x86_64/multiarch/x86-evex-vecs-common.h | 39 ++++
> sysdeps/x86_64/multiarch/x86-evex256-vecs.h | 38 ++++
> sysdeps/x86_64/multiarch/x86-evex512-vecs.h | 38 ++++
> sysdeps/x86_64/multiarch/x86-sse2-vecs.h | 47 +++++
> sysdeps/x86_64/multiarch/x86-vec-macros.h | 90 ++++++++++
> 9 files changed, 635 insertions(+)
> create mode 100644 sysdeps/x86_64/multiarch/reg-macros.h
> create mode 100644 sysdeps/x86_64/multiarch/scripts/gen-reg-macros.py
> create mode 100644 sysdeps/x86_64/multiarch/x86-avx-rtm-vecs.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-avx-vecs.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-evex-vecs-common.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-evex256-vecs.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-evex512-vecs.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-sse2-vecs.h
> create mode 100644 sysdeps/x86_64/multiarch/x86-vec-macros.h
>
> diff --git a/sysdeps/x86_64/multiarch/reg-macros.h b/sysdeps/x86_64/multiarch/reg-macros.h
> new file mode 100644
> index 0000000000..c8ea330256
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/reg-macros.h
> @@ -0,0 +1,168 @@
> +/* This file was generated by: gen-reg-macros.py.
> +
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _REG_MACROS_H
> +#define _REG_MACROS_H 1
> +
> +#define rax_8 al
> +#define rax_16 ax
> +#define rax_32 eax
> +#define rax_64 rax
> +#define rbx_8 bl
> +#define rbx_16 bx
> +#define rbx_32 ebx
> +#define rbx_64 rbx
> +#define rcx_8 cl
> +#define rcx_16 cx
> +#define rcx_32 ecx
> +#define rcx_64 rcx
> +#define rdx_8 dl
> +#define rdx_16 dx
> +#define rdx_32 edx
> +#define rdx_64 rdx
> +#define rbp_8 bpl
> +#define rbp_16 bp
> +#define rbp_32 ebp
> +#define rbp_64 rbp
> +#define rsp_8 spl
> +#define rsp_16 sp
> +#define rsp_32 esp
> +#define rsp_64 rsp
> +#define rsi_8 sil
> +#define rsi_16 si
> +#define rsi_32 esi
> +#define rsi_64 rsi
> +#define rdi_8 dil
> +#define rdi_16 di
> +#define rdi_32 edi
> +#define rdi_64 rdi
> +#define r8_8 r8b
> +#define r8_16 r8w
> +#define r8_32 r8d
> +#define r8_64 r8
> +#define r9_8 r9b
> +#define r9_16 r9w
> +#define r9_32 r9d
> +#define r9_64 r9
> +#define r10_8 r10b
> +#define r10_16 r10w
> +#define r10_32 r10d
> +#define r10_64 r10
> +#define r11_8 r11b
> +#define r11_16 r11w
> +#define r11_32 r11d
> +#define r11_64 r11
> +#define r12_8 r12b
> +#define r12_16 r12w
> +#define r12_32 r12d
> +#define r12_64 r12
> +#define r13_8 r13b
> +#define r13_16 r13w
> +#define r13_32 r13d
> +#define r13_64 r13
> +#define r14_8 r14b
> +#define r14_16 r14w
> +#define r14_32 r14d
> +#define r14_64 r14
> +#define r15_8 r15b
> +#define r15_16 r15w
> +#define r15_32 r15d
> +#define r15_64 r15
> +
> +#define kmov_8 kmovb
> +#define kmov_16 kmovw
> +#define kmov_32 kmovd
> +#define kmov_64 kmovq
> +#define kortest_8 kortestb
> +#define kortest_16 kortestw
> +#define kortest_32 kortestd
> +#define kortest_64 kortestq
> +#define kor_8 korb
> +#define kor_16 korw
> +#define kor_32 kord
> +#define kor_64 korq
> +#define ktest_8 ktestb
> +#define ktest_16 ktestw
> +#define ktest_32 ktestd
> +#define ktest_64 ktestq
> +#define kand_8 kandb
> +#define kand_16 kandw
> +#define kand_32 kandd
> +#define kand_64 kandq
> +#define kxor_8 kxorb
> +#define kxor_16 kxorw
> +#define kxor_32 kxord
> +#define kxor_64 kxorq
> +#define knot_8 knotb
> +#define knot_16 knotw
> +#define knot_32 knotd
> +#define knot_64 knotq
> +#define kxnor_8 kxnorb
> +#define kxnor_16 kxnorw
> +#define kxnor_32 kxnord
> +#define kxnor_64 kxnorq
> +#define kunpack_8 kunpackbw
> +#define kunpack_16 kunpackwd
> +#define kunpack_32 kunpackdq
> +
> +/* Common API for accessing proper width GPR is V{upcase_GPR_name}. */
> +#define VRAX VGPR(rax)
> +#define VRBX VGPR(rbx)
> +#define VRCX VGPR(rcx)
> +#define VRDX VGPR(rdx)
> +#define VRBP VGPR(rbp)
> +#define VRSP VGPR(rsp)
> +#define VRSI VGPR(rsi)
> +#define VRDI VGPR(rdi)
> +#define VR8 VGPR(r8)
> +#define VR9 VGPR(r9)
> +#define VR10 VGPR(r10)
> +#define VR11 VGPR(r11)
> +#define VR12 VGPR(r12)
> +#define VR13 VGPR(r13)
> +#define VR14 VGPR(r14)
> +#define VR15 VGPR(r15)
> +
> +/* Common API for accessing proper width mask insn is {upcase_mask_insn}. */
> +#define KMOV VKINSN(kmov)
> +#define KORTEST VKINSN(kortest)
> +#define KOR VKINSN(kor)
> +#define KTEST VKINSN(ktest)
> +#define KAND VKINSN(kand)
> +#define KXOR VKINSN(kxor)
> +#define KNOT VKINSN(knot)
> +#define KXNOR VKINSN(kxnor)
> +#define KUNPACK VKINSN(kunpack)
> +
> +#ifdef USE_WIDE_CHAR
> +# define REG_WIDTH 32
> +#else
> +# define REG_WIDTH VEC_SIZE
> +#endif
> +
> +#define VPASTER(x, y) x##_##y
> +#define VEVALUATOR(x, y) VPASTER(x, y)
> +
> +#define VGPR_SZ(reg_name, reg_size) VEVALUATOR(reg_name, reg_size)
> +#define VKINSN_SZ(insn, reg_size) VEVALUATOR(insn, reg_size)
> +
> +#define VGPR(reg_name) VGPR_SZ(reg_name, REG_WIDTH)
> +#define VKINSN(mask_insn) VKINSN_SZ(mask_insn, REG_WIDTH)
> +
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/scripts/gen-reg-macros.py b/sysdeps/x86_64/multiarch/scripts/gen-reg-macros.py
> new file mode 100644
> index 0000000000..9fb6903212
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/scripts/gen-reg-macros.py
> @@ -0,0 +1,133 @@
> +#!/usr/bin/python3
> +# Copyright (C) 2022 Free Software Foundation, Inc.
> +# This file is part of the GNU C Library.
> +#
> +# The GNU C Library is free software; you can redistribute it and/or
> +# modify it under the terms of the GNU Lesser General Public
> +# License as published by the Free Software Foundation; either
> +# version 2.1 of the License, or (at your option) any later version.
> +#
> +# The GNU C Library is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> +# Lesser General Public License for more details.
> +#
> +# You should have received a copy of the GNU Lesser General Public
> +# License along with the GNU C Library; if not, see
> +# <https://www.gnu.org/licenses/>.
> +"""Generate macros for getting GPR name of a certain size
> +
> +Inputs: None
> +Output: Prints header fill to stdout
> +
> +API:
> + V{upcase_GPR_name}
> + - Get register name REG_WIDTH component of `upcase_GPR_name`
> + {upcase_mask_insn_without_postfix}
> + - Get proper REG_WIDTH mask insn for `upcase_mask_insn_without_postfix`
> + VGPR(reg_name)
> + - Get register name REG_WIDTH component of `reg_name`
> + VKINSN(mask_insn)
> + - Get proper REG_WIDTH mask insn for `mask_insn`
> + VGPR_SZ(reg_name, reg_size)
> + - Get register name `reg_size` component of `reg_name`
> + VKINSN_SZ(mask_insn, insn_size)
> + - Get proper `insn_size` mask insn for `mask_insn`
> +"""
> +
> +import sys
> +import os
> +from datetime import datetime
> +
> +registers = [["rax", "eax", "ax", "al"], ["rbx", "ebx", "bx", "bl"],
> + ["rcx", "ecx", "cx", "cl"], ["rdx", "edx", "dx", "dl"],
> + ["rbp", "ebp", "bp", "bpl"], ["rsp", "esp", "sp", "spl"],
> + ["rsi", "esi", "si", "sil"], ["rdi", "edi", "di", "dil"],
> + ["r8", "r8d", "r8w", "r8b"], ["r9", "r9d", "r9w", "r9b"],
> + ["r10", "r10d", "r10w", "r10b"], ["r11", "r11d", "r11w", "r11b"],
> + ["r12", "r12d", "r12w", "r12b"], ["r13", "r13d", "r13w", "r13b"],
> + ["r14", "r14d", "r14w", "r14b"], ["r15", "r15d", "r15w", "r15b"]]
> +
> +mask_insns = [
> + "kmov",
> + "kortest",
> + "kor",
> + "ktest",
> + "kand",
> + "kxor",
> + "knot",
> + "kxnor",
> +]
> +mask_insns_ext = ["b", "w", "d", "q"]
> +
> +cr = """
> + Copyright (C) {} Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +"""
> +
> +print("/* This file was generated by: {}.".format(os.path.basename(
> + sys.argv[0])))
> +print(cr.format(datetime.today().year))
> +
> +print("#ifndef _REG_MACROS_H")
> +print("#define _REG_MACROS_H\t1")
> +print("")
> +for reg in registers:
> + for i in range(0, 4):
> + print("#define {}_{}\t{}".format(reg[0], 8 << i, reg[3 - i]))
> +
> +print("")
> +for mask_insn in mask_insns:
> + for i in range(0, 4):
> + print("#define {}_{}\t{}{}".format(mask_insn, 8 << i, mask_insn,
> + mask_insns_ext[i]))
> +for i in range(0, 3):
> + print("#define kunpack_{}\tkunpack{}{}".format(8 << i, mask_insns_ext[i],
> + mask_insns_ext[i + 1]))
> +mask_insns.append("kunpack")
> +
> +print("")
> +print(
> + "/* Common API for accessing proper width GPR is V{upcase_GPR_name}. */")
> +for reg in registers:
> + print("#define V{}\tVGPR({})".format(reg[0].upper(), reg[0]))
> +
> +print("")
> +
> +print(
> + "/* Common API for accessing proper width mask insn is {upcase_mask_insn}. */"
> +)
> +for mask_insn in mask_insns:
> + print("#define {} \tVKINSN({})".format(mask_insn.upper(), mask_insn))
> +print("")
> +
> +print("#ifdef USE_WIDE_CHAR")
> +print("# define REG_WIDTH 32")
> +print("#else")
> +print("# define REG_WIDTH VEC_SIZE")
> +print("#endif")
> +print("")
> +print("#define VPASTER(x, y)\tx##_##y")
> +print("#define VEVALUATOR(x, y)\tVPASTER(x, y)")
> +print("")
> +print("#define VGPR_SZ(reg_name, reg_size)\tVEVALUATOR(reg_name, reg_size)")
> +print("#define VKINSN_SZ(insn, reg_size)\tVEVALUATOR(insn, reg_size)")
> +print("")
> +print("#define VGPR(reg_name)\tVGPR_SZ(reg_name, REG_WIDTH)")
> +print("#define VKINSN(mask_insn)\tVKINSN_SZ(mask_insn, REG_WIDTH)")
> +
> +print("\n#endif")
> diff --git a/sysdeps/x86_64/multiarch/x86-avx-rtm-vecs.h b/sysdeps/x86_64/multiarch/x86-avx-rtm-vecs.h
> new file mode 100644
> index 0000000000..0b326c8a70
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-avx-rtm-vecs.h
> @@ -0,0 +1,35 @@
> +/* Common config for AVX-RTM VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _X86_AVX_RTM_VECS_H
> +#define _X86_AVX_RTM_VECS_H 1
> +
> +#define COND_VZEROUPPER COND_VZEROUPPER_XTEST
> +#define ZERO_UPPER_VEC_REGISTERS_RETURN \
> + ZERO_UPPER_VEC_REGISTERS_RETURN_XTEST
> +
> +#define VZEROUPPER_RETURN jmp L(return_vzeroupper)
> +
> +#define USE_WITH_RTM 1
> +#include "x86-avx-vecs.h"
> +
> +#undef SECTION
> +#define SECTION(p) p##.avx.rtm
> +
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-avx-vecs.h b/sysdeps/x86_64/multiarch/x86-avx-vecs.h
> new file mode 100644
> index 0000000000..dca1089060
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-avx-vecs.h
> @@ -0,0 +1,47 @@
> +/* Common config for AVX VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _X86_AVX_VECS_H
> +#define _X86_AVX_VECS_H 1
> +
> +#ifdef VEC_SIZE
> +# error "Multiple VEC configs included!"
> +#endif
> +
> +#define VEC_SIZE 32
> +#include "x86-vec-macros.h"
> +
> +#define USE_WITH_AVX 1
> +#define SECTION(p) p##.avx
> +
> +/* 4-byte mov instructions with AVX2. */
> +#define MOV_SIZE 4
> +/* 1 (ret) + 3 (vzeroupper). */
> +#define RET_SIZE 4
> +#define VZEROUPPER vzeroupper
> +
> +#define VMOVU vmovdqu
> +#define VMOVA vmovdqa
> +#define VMOVNT vmovntdq
> +
> +/* Often need to access xmm portion. */
> +#define VMM_128 VMM_any_xmm
> +#define VMM VMM_any_ymm
> +
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-evex-vecs-common.h b/sysdeps/x86_64/multiarch/x86-evex-vecs-common.h
> new file mode 100644
> index 0000000000..f331e9d8ec
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-evex-vecs-common.h
> @@ -0,0 +1,39 @@
> +/* Common config for EVEX256 and EVEX512 VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _X86_EVEX_VECS_COMMON_H
> +#define _X86_EVEX_VECS_COMMON_H 1
> +
> +#include "x86-vec-macros.h"
> +
> +/* 6-byte mov instructions with EVEX. */
> +#define MOV_SIZE 6
> +/* No vzeroupper needed. */
> +#define RET_SIZE 1
> +#define VZEROUPPER
> +
> +#define VMOVU vmovdqu64
> +#define VMOVA vmovdqa64
> +#define VMOVNT vmovntdq
> +
> +#define VMM_128 VMM_hi_xmm
> +#define VMM_256 VMM_hi_ymm
> +#define VMM_512 VMM_hi_zmm
> +
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-evex256-vecs.h b/sysdeps/x86_64/multiarch/x86-evex256-vecs.h
> new file mode 100644
> index 0000000000..8337b95504
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-evex256-vecs.h
> @@ -0,0 +1,38 @@
> +/* Common config for EVEX256 VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _EVEX256_VECS_H
> +#define _EVEX256_VECS_H 1
> +
> +#ifdef VEC_SIZE
> +# error "Multiple VEC configs included!"
> +#endif
> +
> +#define VEC_SIZE 32
> +#include "x86-evex-vecs-common.h"
> +
> +#define USE_WITH_EVEX256 1
> +
> +#ifndef SECTION
> +# define SECTION(p) p##.evex
> +#endif
> +
> +#define VMM VMM_256
> +#define VMM_lo VMM_any_ymm
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-evex512-vecs.h b/sysdeps/x86_64/multiarch/x86-evex512-vecs.h
> new file mode 100644
> index 0000000000..7dc5c23ad0
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-evex512-vecs.h
> @@ -0,0 +1,38 @@
> +/* Common config for EVEX512 VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _EVEX512_VECS_H
> +#define _EVEX512_VECS_H 1
> +
> +#ifdef VEC_SIZE
> +# error "Multiple VEC configs included!"
> +#endif
> +
> +#define VEC_SIZE 64
> +#include "x86-evex-vecs-common.h"
> +
> +#define USE_WITH_EVEX512 1
> +
> +#ifndef SECTION
> +# define SECTION(p) p##.evex512
> +#endif
> +
> +#define VMM VMM_512
> +#define VMM_lo VMM_any_zmm
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-sse2-vecs.h b/sysdeps/x86_64/multiarch/x86-sse2-vecs.h
> new file mode 100644
> index 0000000000..b8bbd5dc29
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-sse2-vecs.h
> @@ -0,0 +1,47 @@
> +/* Common config for SSE2 VECs
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _X86_SSE2_VECS_H
> +#define _X86_SSE2_VECS_H 1
> +
> +#ifdef VEC_SIZE
> +# error "Multiple VEC configs included!"
> +#endif
> +
> +#define VEC_SIZE 16
> +#include "x86-vec-macros.h"
> +
> +#define USE_WITH_SSE2 1
> +#define SECTION(p) p
> +
> +/* 3-byte mov instructions with SSE2. */
> +#define MOV_SIZE 3
> +/* No vzeroupper needed. */
> +#define RET_SIZE 1
> +#define VZEROUPPER
> +
> +#define VMOVU movups
> +#define VMOVA movaps
> +#define VMOVNT movntdq
> +
> +#define VMM_128 VMM_any_xmm
> +#define VMM VMM_any_xmm
> +
> +
> +#endif
> diff --git a/sysdeps/x86_64/multiarch/x86-vec-macros.h b/sysdeps/x86_64/multiarch/x86-vec-macros.h
> new file mode 100644
> index 0000000000..7d6bb31d55
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/x86-vec-macros.h
> @@ -0,0 +1,90 @@
> +/* Macro helpers for VEC_{type}({vec_num})
> + All versions must be listed in ifunc-impl-list.c.
> + Copyright (C) 2022 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#ifndef _X86_VEC_MACROS_H
> +#define _X86_VEC_MACROS_H 1
> +
> +#ifndef VEC_SIZE
> +# error "Never include this file directly. Always include a vector config."
> +#endif
> +
> +/* Defines so we can use SSE2 / AVX2 / EVEX / EVEX512 encoding with same
> + VMM(N) values. */
> +#define VMM_hi_xmm0 xmm16
> +#define VMM_hi_xmm1 xmm17
> +#define VMM_hi_xmm2 xmm18
> +#define VMM_hi_xmm3 xmm19
> +#define VMM_hi_xmm4 xmm20
> +#define VMM_hi_xmm5 xmm21
> +#define VMM_hi_xmm6 xmm22
> +#define VMM_hi_xmm7 xmm23
> +#define VMM_hi_xmm8 xmm24
> +#define VMM_hi_xmm9 xmm25
> +#define VMM_hi_xmm10 xmm26
> +#define VMM_hi_xmm11 xmm27
> +#define VMM_hi_xmm12 xmm28
> +#define VMM_hi_xmm13 xmm29
> +#define VMM_hi_xmm14 xmm30
> +#define VMM_hi_xmm15 xmm31
> +
> +#define VMM_hi_ymm0 ymm16
> +#define VMM_hi_ymm1 ymm17
> +#define VMM_hi_ymm2 ymm18
> +#define VMM_hi_ymm3 ymm19
> +#define VMM_hi_ymm4 ymm20
> +#define VMM_hi_ymm5 ymm21
> +#define VMM_hi_ymm6 ymm22
> +#define VMM_hi_ymm7 ymm23
> +#define VMM_hi_ymm8 ymm24
> +#define VMM_hi_ymm9 ymm25
> +#define VMM_hi_ymm10 ymm26
> +#define VMM_hi_ymm11 ymm27
> +#define VMM_hi_ymm12 ymm28
> +#define VMM_hi_ymm13 ymm29
> +#define VMM_hi_ymm14 ymm30
> +#define VMM_hi_ymm15 ymm31
> +
> +#define VMM_hi_zmm0 zmm16
> +#define VMM_hi_zmm1 zmm17
> +#define VMM_hi_zmm2 zmm18
> +#define VMM_hi_zmm3 zmm19
> +#define VMM_hi_zmm4 zmm20
> +#define VMM_hi_zmm5 zmm21
> +#define VMM_hi_zmm6 zmm22
> +#define VMM_hi_zmm7 zmm23
> +#define VMM_hi_zmm8 zmm24
> +#define VMM_hi_zmm9 zmm25
> +#define VMM_hi_zmm10 zmm26
> +#define VMM_hi_zmm11 zmm27
> +#define VMM_hi_zmm12 zmm28
> +#define VMM_hi_zmm13 zmm29
> +#define VMM_hi_zmm14 zmm30
> +#define VMM_hi_zmm15 zmm31
> +
> +#define PRIMITIVE_VMM(vec, num) vec##num
> +
> +#define VMM_any_xmm(i) PRIMITIVE_VMM(xmm, i)
> +#define VMM_any_ymm(i) PRIMITIVE_VMM(ymm, i)
> +#define VMM_any_zmm(i) PRIMITIVE_VMM(zmm, i)
> +
> +#define VMM_hi_xmm(i) PRIMITIVE_VMM(VMM_hi_xmm, i)
> +#define VMM_hi_ymm(i) PRIMITIVE_VMM(VMM_hi_ymm, i)
> +#define VMM_hi_zmm(i) PRIMITIVE_VMM(VMM_hi_zmm, i)
> +
> +#endif
> --
> 2.34.1
>
LGTM.
Thanks.
--
H.J.
next prev parent reply other threads:[~2022-10-15 2:45 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-14 16:40 [PATCH v1 1/3] x86: Update evex256/512 vec macros Noah Goldstein
2022-10-14 16:40 ` [PATCH v1 2/3] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 18:02 ` H.J. Lu
2022-10-14 18:26 ` Noah Goldstein
2022-10-14 18:35 ` H.J. Lu
2022-10-14 18:38 ` Noah Goldstein
2022-10-14 18:53 ` H.J. Lu
2022-10-14 19:00 ` Noah Goldstein
2022-10-14 19:13 ` H.J. Lu
2022-10-14 19:15 ` Noah Goldstein
2022-10-14 16:40 ` [PATCH v1 3/3] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-14 17:31 ` [PATCH v1 1/3] x86: Update evex256/512 vec macros H.J. Lu
2022-10-14 18:01 ` [PATCH v2 " Noah Goldstein
2022-10-14 18:01 ` [PATCH v2 2/3] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 18:01 ` [PATCH v2 3/3] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-14 18:22 ` [PATCH v3 1/3] x86: Update evex256/512 vec macros Noah Goldstein
2022-10-14 18:22 ` [PATCH v3 2/3] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 18:22 ` [PATCH v3 3/3] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-14 18:41 ` [PATCH v4 1/3] x86: Update evex256/512 vec macros Noah Goldstein
2022-10-14 18:41 ` [PATCH v4 2/3] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 18:41 ` [PATCH v4 3/3] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-14 21:14 ` [PATCH v5 1/3] x86: Update evex256/512 vec macros Noah Goldstein
2022-10-14 21:15 ` [PATCH v5 2/3] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 21:28 ` H.J. Lu
2022-10-14 22:01 ` Noah Goldstein
2022-10-14 22:05 ` H.J. Lu
2022-10-14 22:27 ` Noah Goldstein
2022-10-14 22:41 ` H.J. Lu
2022-10-14 23:15 ` Noah Goldstein
2022-10-14 23:22 ` H.J. Lu
2022-10-14 23:25 ` Noah Goldstein
2022-10-14 21:15 ` [PATCH v5 3/3] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 1/7] x86: Update and move evex256/512 vec macros Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 2/7] x86: Add macros for GPRs / mask insn based on VEC_SIZE Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 3/7] x86: Update memrchr to use new VEC macros Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 4/7] x86: Remove now unused vec header macros Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 5/7] x86: Update memmove to use new VEC macros Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 6/7] x86: Update memset " Noah Goldstein
2022-10-14 22:39 ` [PATCH v6 7/7] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 1/6] x86: Update VEC macros to complete API for evex/evex512 impls Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 2/6] x86: Update memrchr to use new VEC macros Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 3/6] x86: Update memmove " Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 4/6] x86: Update memset " Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 5/6] x86: Remove now unused vec header macros Noah Goldstein
2022-10-15 0:06 ` [PATCH v8 6/6] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-15 0:12 ` [PATCH v8 1/6] x86: Update VEC macros to complete API for evex/evex512 impls H.J. Lu
2022-10-15 0:20 ` Noah Goldstein
2022-10-15 0:20 ` [PATCH v9 " Noah Goldstein
2022-10-15 0:20 ` [PATCH v9 2/6] x86: Update memrchr to use new VEC macros Noah Goldstein
2022-10-15 2:48 ` H.J. Lu
2022-10-15 0:20 ` [PATCH v9 3/6] x86: Update memmove " Noah Goldstein
2022-10-15 2:52 ` H.J. Lu
2022-10-15 2:57 ` Noah Goldstein
2022-10-15 0:20 ` [PATCH v9 4/6] x86: Update memset " Noah Goldstein
2022-10-15 2:53 ` H.J. Lu
2022-10-15 0:20 ` [PATCH v9 5/6] x86: Remove now unused vec header macros Noah Goldstein
2022-10-15 2:56 ` H.J. Lu
2022-10-15 0:21 ` [PATCH v9 6/6] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-15 2:58 ` H.J. Lu
2022-10-15 2:45 ` H.J. Lu [this message]
2022-10-15 3:00 ` [PATCH v10 1/6] x86: Update VEC macros to complete API for evex/evex512 impls Noah Goldstein
2022-10-15 3:00 ` [PATCH v10 2/6] x86: Update memrchr to use new VEC macros Noah Goldstein
2022-10-15 3:44 ` Sunil Pandey
2022-10-15 3:00 ` [PATCH v10 3/6] x86: Update memmove " Noah Goldstein
2022-10-15 3:43 ` Sunil Pandey
2022-10-15 3:00 ` [PATCH v10 4/6] x86: Update memset " Noah Goldstein
2022-10-15 3:42 ` Sunil Pandey
2022-10-15 3:00 ` [PATCH v10 5/6] x86: Remove now unused vec header macros Noah Goldstein
2022-10-15 3:39 ` Sunil Pandey
2022-10-15 3:00 ` [PATCH v10 6/6] x86: Update strlen-evex-base to use new reg/vec macros Noah Goldstein
2022-10-15 3:48 ` Sunil Pandey
2022-10-15 3:37 ` [PATCH v10 1/6] x86: Update VEC macros to complete API for evex/evex512 impls Sunil Pandey
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='CAMe9rOp9H+o=b9eTn3OgTn8d5EYdeykWuEh05hgUFdjH8Jyzxg@mail.gmail.com' \
--to=hjl.tools@gmail.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).