glibc/sysdeps/x86_64/multiarch/memset-avx512-no-vzeroupper.S
H.J. Lu f43cb35c9b Require binutils 2.24 to build x86-64 glibc [BZ #20139]
If assembler doesn't support AVX512DQ, _dl_runtime_resolve_avx is used
to save the first 8 vector registers, which only saves the lower 256
bits of vector register, for lazy binding.  When it is called on AVX512
platform, the upper 256 bits of ZMM registers are clobbered.  Parameters
passed in ZMM registers will be wrong when the function is called the
first time.  This patch requires binutils 2.24, whose assembler can store
and load ZMM registers, to build x86-64 glibc.  Since mathvec library
needs assembler support for AVX512DQ,  we disable mathvec if assembler
doesn't support AVX512DQ.

	[BZ #20139]
	* config.h.in (HAVE_AVX512_ASM_SUPPORT): Renamed to ...
	(HAVE_AVX512DQ_ASM_SUPPORT): This.
	* sysdeps/x86_64/configure.ac: Require assembler from binutils
	2.24 or above.
	(HAVE_AVX512_ASM_SUPPORT): Removed.
	(HAVE_AVX512DQ_ASM_SUPPORT): New.
	* sysdeps/x86_64/configure: Regenerated.
	* sysdeps/x86_64/dl-trampoline.S: Make HAVE_AVX512_ASM_SUPPORT
	check unconditional.
	* sysdeps/x86_64/multiarch/ifunc-impl-list.c: Likewise.
	* sysdeps/x86_64/multiarch/memcpy.S: Likewise.
	* sysdeps/x86_64/multiarch/memcpy_chk.S: Likewise.
	* sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S:
	Likewise.
	* sysdeps/x86_64/multiarch/memmove-avx512-unaligned-erms.S:
	Likewise.
	* sysdeps/x86_64/multiarch/memmove.S: Likewise.
	* sysdeps/x86_64/multiarch/memmove_chk.S: Likewise.
	* sysdeps/x86_64/multiarch/mempcpy.S: Likewise.
	* sysdeps/x86_64/multiarch/mempcpy_chk.S: Likewise.
	* sysdeps/x86_64/multiarch/memset-avx512-no-vzeroupper.S:
	Likewise.
	* sysdeps/x86_64/multiarch/memset-avx512-unaligned-erms.S:
	Likewise.
	* sysdeps/x86_64/multiarch/memset.S: Likewise.
	* sysdeps/x86_64/multiarch/memset_chk.S: Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S: Check
	HAVE_AVX512DQ_ASM_SUPPORT instead of HAVE_AVX512_ASM_SUPPORT.
	* sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_log8_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_pow8_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sin8_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_expf16_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_logf16_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_powf16_core_avx512.S:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx51:
	Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf16_core_avx512.S:
	Likewise.
2016-07-01 06:03:05 -07:00

195 lines
4.2 KiB
ArmAsm

/* memset optimized with AVX512 for KNL hardware.
Copyright (C) 2015-2016 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
<http://www.gnu.org/licenses/>. */
#include <sysdep.h>
#if IS_IN (libc)
#include "asm-syntax.h"
#ifndef MEMSET
# define MEMSET __memset_avx512_no_vzeroupper
# define MEMSET_CHK __memset_chk_avx512_no_vzeroupper
#endif
.section .text.avx512,"ax",@progbits
#if defined PIC
ENTRY (MEMSET_CHK)
cmpq %rdx, %rcx
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMSET_CHK)
#endif
ENTRY (MEMSET)
vpxor %xmm0, %xmm0, %xmm0
vmovd %esi, %xmm1
lea (%rdi, %rdx), %rsi
mov %rdi, %rax
vpshufb %xmm0, %xmm1, %xmm0
cmp $16, %rdx
jb L(less_16bytes)
cmp $512, %rdx
vbroadcastss %xmm0, %zmm2
ja L(512bytesormore)
cmp $256, %rdx
jb L(less_256bytes)
vmovups %zmm2, (%rdi)
vmovups %zmm2, 0x40(%rdi)
vmovups %zmm2, 0x80(%rdi)
vmovups %zmm2, 0xC0(%rdi)
vmovups %zmm2, -0x100(%rsi)
vmovups %zmm2, -0xC0(%rsi)
vmovups %zmm2, -0x80(%rsi)
vmovups %zmm2, -0x40(%rsi)
ret
L(less_256bytes):
cmp $128, %dl
jb L(less_128bytes)
vmovups %zmm2, (%rdi)
vmovups %zmm2, 0x40(%rdi)
vmovups %zmm2, -0x80(%rsi)
vmovups %zmm2, -0x40(%rsi)
ret
L(less_128bytes):
cmp $64, %dl
jb L(less_64bytes)
vmovups %zmm2, (%rdi)
vmovups %zmm2, -0x40(%rsi)
ret
L(less_64bytes):
cmp $32, %dl
jb L(less_32bytes)
vmovdqu %ymm2, (%rdi)
vmovdqu %ymm2, -0x20(%rsi)
ret
L(less_32bytes):
vmovdqu %xmm0, (%rdi)
vmovdqu %xmm0, -0x10(%rsi)
ret
L(less_16bytes):
cmp $8, %dl
jb L(less_8bytes)
vmovq %xmm0, (%rdi)
vmovq %xmm0, -0x08(%rsi)
ret
L(less_8bytes):
vmovd %xmm0, %ecx
cmp $4, %dl
jb L(less_4bytes)
mov %ecx, (%rdi)
mov %ecx, -0x04(%rsi)
ret
L(less_4bytes):
cmp $2, %dl
jb L(less_2bytes)
mov %cx, (%rdi)
mov %cx, -0x02(%rsi)
ret
L(less_2bytes):
cmp $1, %dl
jb L(less_1bytes)
mov %cl, (%rdi)
L(less_1bytes):
ret
L(512bytesormore):
mov __x86_shared_cache_size_half(%rip), %rcx
cmp %rcx, %rdx
ja L(preloop_large)
cmp $1024, %rdx
ja L(1024bytesormore)
vmovups %zmm2, (%rdi)
vmovups %zmm2, 0x40(%rdi)
vmovups %zmm2, 0x80(%rdi)
vmovups %zmm2, 0xC0(%rdi)
vmovups %zmm2, 0x100(%rdi)
vmovups %zmm2, 0x140(%rdi)
vmovups %zmm2, 0x180(%rdi)
vmovups %zmm2, 0x1C0(%rdi)
vmovups %zmm2, -0x200(%rsi)
vmovups %zmm2, -0x1C0(%rsi)
vmovups %zmm2, -0x180(%rsi)
vmovups %zmm2, -0x140(%rsi)
vmovups %zmm2, -0x100(%rsi)
vmovups %zmm2, -0xC0(%rsi)
vmovups %zmm2, -0x80(%rsi)
vmovups %zmm2, -0x40(%rsi)
ret
/* Align on 64 and loop with aligned stores. */
L(1024bytesormore):
sub $0x100, %rsi
vmovups %zmm2, (%rax)
and $-0x40, %rdi
add $0x40, %rdi
L(gobble_256bytes_loop):
vmovaps %zmm2, (%rdi)
vmovaps %zmm2, 0x40(%rdi)
vmovaps %zmm2, 0x80(%rdi)
vmovaps %zmm2, 0xC0(%rdi)
add $0x100, %rdi
cmp %rsi, %rdi
jb L(gobble_256bytes_loop)
vmovups %zmm2, (%rsi)
vmovups %zmm2, 0x40(%rsi)
vmovups %zmm2, 0x80(%rsi)
vmovups %zmm2, 0xC0(%rsi)
ret
/* Align on 128 and loop with non-temporal stores. */
L(preloop_large):
and $-0x80, %rdi
add $0x80, %rdi
vmovups %zmm2, (%rax)
vmovups %zmm2, 0x40(%rax)
sub $0x200, %rsi
L(gobble_512bytes_nt_loop):
vmovntdq %zmm2, (%rdi)
vmovntdq %zmm2, 0x40(%rdi)
vmovntdq %zmm2, 0x80(%rdi)
vmovntdq %zmm2, 0xC0(%rdi)
vmovntdq %zmm2, 0x100(%rdi)
vmovntdq %zmm2, 0x140(%rdi)
vmovntdq %zmm2, 0x180(%rdi)
vmovntdq %zmm2, 0x1C0(%rdi)
add $0x200, %rdi
cmp %rsi, %rdi
jb L(gobble_512bytes_nt_loop)
sfence
vmovups %zmm2, (%rsi)
vmovups %zmm2, 0x40(%rsi)
vmovups %zmm2, 0x80(%rsi)
vmovups %zmm2, 0xC0(%rsi)
vmovups %zmm2, 0x100(%rsi)
vmovups %zmm2, 0x140(%rsi)
vmovups %zmm2, 0x180(%rsi)
vmovups %zmm2, 0x1C0(%rsi)
ret
END (MEMSET)
#endif