AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-12-04 10:50:07 +00:00
Code Issues Packages Projects Releases Wiki Activity
5e1122827a
glibc/sysdeps/x86_64/dl-trampoline.h

648 lines
20 KiB
C
Raw Normal View History

Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
/* PLT trampolines. x86-64 version.
Update copyright dates with scripts/update-copyrights.
2017-01-01 00:14:16 +00:00
Copyright (C) 2009-2017 Free Software Foundation, Inc.
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
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
Replace FSF snail mail address with URLs.
2012-02-09 23:18:22 +00:00
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
#undef REGISTER_SAVE_AREA_RAW
#ifdef __ILP32__
/* X32 saves RCX, RDX, RSI, RDI, R8 and R9 plus RAX as well as VEC0 to
VEC7. */
# define REGISTER_SAVE_AREA_RAW (8 * 7 + VEC_SIZE * 8)
#else
/* X86-64 saves RCX, RDX, RSI, RDI, R8 and R9 plus RAX as well as
BND0, BND1, BND2, BND3 and VEC0 to VEC7. */
# define REGISTER_SAVE_AREA_RAW (8 * 7 + 16 * 4 + VEC_SIZE * 8)
#endif
#undef REGISTER_SAVE_AREA
#undef LOCAL_STORAGE_AREA
#undef BASE
[x86_64] Set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 Due to GCC bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58066 __tls_get_addr may be called with 8-byte stack alignment. Although this bug has been fixed in GCC 4.9.4, 5.3 and 6, we can't assume that stack will be always aligned at 16 bytes. Since SSE optimized memory/string functions with aligned SSE register load and store are used in the dynamic linker, we must set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 so that _dl_runtime_resolve_sse will align the stack before calling _dl_fixup: Dump of assembler code for function _dl_runtime_resolve_sse: 0x00007ffff7deea90 <+0>: push %rbx 0x00007ffff7deea91 <+1>: mov %rsp,%rbx 0x00007ffff7deea94 <+4>: and $0xfffffffffffffff0,%rsp ^^^^^^^^^^^ Align stack to 16 bytes 0x00007ffff7deea98 <+8>: sub $0x100,%rsp 0x00007ffff7deea9f <+15>: mov %rax,0xc0(%rsp) 0x00007ffff7deeaa7 <+23>: mov %rcx,0xc8(%rsp) 0x00007ffff7deeaaf <+31>: mov %rdx,0xd0(%rsp) 0x00007ffff7deeab7 <+39>: mov %rsi,0xd8(%rsp) 0x00007ffff7deeabf <+47>: mov %rdi,0xe0(%rsp) 0x00007ffff7deeac7 <+55>: mov %r8,0xe8(%rsp) 0x00007ffff7deeacf <+63>: mov %r9,0xf0(%rsp) 0x00007ffff7deead7 <+71>: movaps %xmm0,(%rsp) 0x00007ffff7deeadb <+75>: movaps %xmm1,0x10(%rsp) 0x00007ffff7deeae0 <+80>: movaps %xmm2,0x20(%rsp) 0x00007ffff7deeae5 <+85>: movaps %xmm3,0x30(%rsp) 0x00007ffff7deeaea <+90>: movaps %xmm4,0x40(%rsp) 0x00007ffff7deeaef <+95>: movaps %xmm5,0x50(%rsp) 0x00007ffff7deeaf4 <+100>: movaps %xmm6,0x60(%rsp) 0x00007ffff7deeaf9 <+105>: movaps %xmm7,0x70(%rsp) [BZ #19679] * sysdeps/x86_64/dl-trampoline.S (DL_RUNIME_UNALIGNED_VEC_SIZE): Renamed to ... (DL_RUNTIME_UNALIGNED_VEC_SIZE): This. Set to 8. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. Updated. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. * sysdeps/x86_64/dl-trampoline.h (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This.
2016-02-19 23:43:45 +00:00
#if DL_RUNTIME_RESOLVE_REALIGN_STACK
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# define REGISTER_SAVE_AREA (REGISTER_SAVE_AREA_RAW + 8)
/* Local stack area before jumping to function address: RBX. */
# define LOCAL_STORAGE_AREA 8
# define BASE rbx
# if (REGISTER_SAVE_AREA % VEC_SIZE) != 0
# error REGISTER_SAVE_AREA must be multples of VEC_SIZE
# endif
#else
# define REGISTER_SAVE_AREA REGISTER_SAVE_AREA_RAW
/* Local stack area before jumping to function address: All saved
registers. */
# define LOCAL_STORAGE_AREA REGISTER_SAVE_AREA
# define BASE rsp
# if (REGISTER_SAVE_AREA % 16) != 8
# error REGISTER_SAVE_AREA must be odd multples of 8
# endif
#endif
.text
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
#ifdef _dl_runtime_resolve_opt
/* Use the smallest vector registers to preserve the full YMM/ZMM
registers to avoid SSE transition penalty. */
# if VEC_SIZE == 32
/* Check if the upper 128 bits in %ymm0 - %ymm7 registers are non-zero
and preserve %xmm0 - %xmm7 registers with the zero upper bits. Since
there is no SSE transition penalty on AVX512 processors which don't
support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't
provided. */
.globl _dl_runtime_resolve_avx_slow
.hidden _dl_runtime_resolve_avx_slow
.type _dl_runtime_resolve_avx_slow, @function
.align 16
_dl_runtime_resolve_avx_slow:
cfi_startproc
cfi_adjust_cfa_offset(16) # Incorporate PLT
vorpd %ymm0, %ymm1, %ymm8
vorpd %ymm2, %ymm3, %ymm9
vorpd %ymm4, %ymm5, %ymm10
vorpd %ymm6, %ymm7, %ymm11
vorpd %ymm8, %ymm9, %ymm9
vorpd %ymm10, %ymm11, %ymm10
vpcmpeqd %xmm8, %xmm8, %xmm8
vorpd %ymm9, %ymm10, %ymm10
vptest %ymm10, %ymm8
# Preserve %ymm0 - %ymm7 registers if the upper 128 bits of any
# %ymm0 - %ymm7 registers aren't zero.
PRESERVE_BND_REGS_PREFIX
jnc _dl_runtime_resolve_avx
# Use vzeroupper to avoid SSE transition penalty.
vzeroupper
# Preserve %xmm0 - %xmm7 registers with the zero upper 128 bits
# when the upper 128 bits of %ymm0 - %ymm7 registers are zero.
PRESERVE_BND_REGS_PREFIX
jmp _dl_runtime_resolve_sse_vex
cfi_adjust_cfa_offset(-16) # Restore PLT adjustment
cfi_endproc
.size _dl_runtime_resolve_avx_slow, .-_dl_runtime_resolve_avx_slow
# endif
/* Use XGETBV with ECX == 1 to check which bits in vector registers are
non-zero and only preserve the non-zero lower bits with zero upper
bits. */
.globl _dl_runtime_resolve_opt
.hidden _dl_runtime_resolve_opt
.type _dl_runtime_resolve_opt, @function
.align 16
_dl_runtime_resolve_opt:
cfi_startproc
cfi_adjust_cfa_offset(16) # Incorporate PLT
pushq %rax
cfi_adjust_cfa_offset(8)
cfi_rel_offset(%rax, 0)
pushq %rcx
cfi_adjust_cfa_offset(8)
cfi_rel_offset(%rcx, 0)
pushq %rdx
cfi_adjust_cfa_offset(8)
cfi_rel_offset(%rdx, 0)
movl $1, %ecx
xgetbv
movl %eax, %r11d
popq %rdx
cfi_adjust_cfa_offset(-8)
cfi_restore (%rdx)
popq %rcx
cfi_adjust_cfa_offset(-8)
cfi_restore (%rcx)
popq %rax
cfi_adjust_cfa_offset(-8)
cfi_restore (%rax)
# if VEC_SIZE == 32
# For YMM registers, check if YMM state is in use.
andl $bit_YMM_state, %r11d
# Preserve %xmm0 - %xmm7 registers with the zero upper 128 bits if
# YMM state isn't in use.
PRESERVE_BND_REGS_PREFIX
jz _dl_runtime_resolve_sse_vex
x86-64: Improve branch predication in _dl_runtime_resolve_avx512_opt [BZ #21258] On Skylake server, _dl_runtime_resolve_avx512_opt is used to preserve the first 8 vector registers. The code layout is if only %xmm0 - %xmm7 registers are used preserve %xmm0 - %xmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %zmm0 - %zmm7 registers Branch predication always executes the fallthrough code path to preserve %zmm0 - %zmm7 registers speculatively, even though only %xmm0 - %xmm7 registers are used. This leads to lower CPU frequency on Skylake server. This patch changes the fallthrough code path to preserve %xmm0 - %xmm7 registers instead: if whole %zmm0 - %zmm7 registers are used preserve %zmm0 - %zmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %xmm0 - %xmm7 registers Tested on Skylake server. [BZ #21258] * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve_opt): Define only if _dl_runtime_resolve is defined to _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_opt): Fallthrough to _dl_runtime_resolve_sse_vex.
2017-03-21 17:59:31 +00:00
# elif VEC_SIZE == 16
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
# For ZMM registers, check if YMM state and ZMM state are in
# use.
andl $(bit_YMM_state | bit_ZMM0_15_state), %r11d
cmpl $bit_YMM_state, %r11d
x86-64: Improve branch predication in _dl_runtime_resolve_avx512_opt [BZ #21258] On Skylake server, _dl_runtime_resolve_avx512_opt is used to preserve the first 8 vector registers. The code layout is if only %xmm0 - %xmm7 registers are used preserve %xmm0 - %xmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %zmm0 - %zmm7 registers Branch predication always executes the fallthrough code path to preserve %zmm0 - %zmm7 registers speculatively, even though only %xmm0 - %xmm7 registers are used. This leads to lower CPU frequency on Skylake server. This patch changes the fallthrough code path to preserve %xmm0 - %xmm7 registers instead: if whole %zmm0 - %zmm7 registers are used preserve %zmm0 - %zmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %xmm0 - %xmm7 registers Tested on Skylake server. [BZ #21258] * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve_opt): Define only if _dl_runtime_resolve is defined to _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_opt): Fallthrough to _dl_runtime_resolve_sse_vex.
2017-03-21 17:59:31 +00:00
# Preserve %zmm0 - %zmm7 registers if ZMM state is in use.
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
PRESERVE_BND_REGS_PREFIX
x86-64: Improve branch predication in _dl_runtime_resolve_avx512_opt [BZ #21258] On Skylake server, _dl_runtime_resolve_avx512_opt is used to preserve the first 8 vector registers. The code layout is if only %xmm0 - %xmm7 registers are used preserve %xmm0 - %xmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %zmm0 - %zmm7 registers Branch predication always executes the fallthrough code path to preserve %zmm0 - %zmm7 registers speculatively, even though only %xmm0 - %xmm7 registers are used. This leads to lower CPU frequency on Skylake server. This patch changes the fallthrough code path to preserve %xmm0 - %xmm7 registers instead: if whole %zmm0 - %zmm7 registers are used preserve %zmm0 - %zmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %xmm0 - %xmm7 registers Tested on Skylake server. [BZ #21258] * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve_opt): Define only if _dl_runtime_resolve is defined to _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_opt): Fallthrough to _dl_runtime_resolve_sse_vex.
2017-03-21 17:59:31 +00:00
jg _dl_runtime_resolve_avx512
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
# Preserve %ymm0 - %ymm7 registers with the zero upper 256 bits if
# ZMM state isn't in use.
PRESERVE_BND_REGS_PREFIX
je _dl_runtime_resolve_avx
x86-64: Improve branch predication in _dl_runtime_resolve_avx512_opt [BZ #21258] On Skylake server, _dl_runtime_resolve_avx512_opt is used to preserve the first 8 vector registers. The code layout is if only %xmm0 - %xmm7 registers are used preserve %xmm0 - %xmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %zmm0 - %zmm7 registers Branch predication always executes the fallthrough code path to preserve %zmm0 - %zmm7 registers speculatively, even though only %xmm0 - %xmm7 registers are used. This leads to lower CPU frequency on Skylake server. This patch changes the fallthrough code path to preserve %xmm0 - %xmm7 registers instead: if whole %zmm0 - %zmm7 registers are used preserve %zmm0 - %zmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %xmm0 - %xmm7 registers Tested on Skylake server. [BZ #21258] * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve_opt): Define only if _dl_runtime_resolve is defined to _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_opt): Fallthrough to _dl_runtime_resolve_sse_vex.
2017-03-21 17:59:31 +00:00
# Preserve %xmm0 - %xmm7 registers with the zero upper 384 bits if
# neither YMM state nor ZMM state are in use.
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
# else
# error Unsupported VEC_SIZE!
# endif
cfi_adjust_cfa_offset(-16) # Restore PLT adjustment
cfi_endproc
.size _dl_runtime_resolve_opt, .-_dl_runtime_resolve_opt
#endif
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
.globl _dl_runtime_resolve
.hidden _dl_runtime_resolve
.type _dl_runtime_resolve, @function
.align 16
cfi_startproc
_dl_runtime_resolve:
cfi_adjust_cfa_offset(16) # Incorporate PLT
[x86_64] Set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 Due to GCC bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58066 __tls_get_addr may be called with 8-byte stack alignment. Although this bug has been fixed in GCC 4.9.4, 5.3 and 6, we can't assume that stack will be always aligned at 16 bytes. Since SSE optimized memory/string functions with aligned SSE register load and store are used in the dynamic linker, we must set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 so that _dl_runtime_resolve_sse will align the stack before calling _dl_fixup: Dump of assembler code for function _dl_runtime_resolve_sse: 0x00007ffff7deea90 <+0>: push %rbx 0x00007ffff7deea91 <+1>: mov %rsp,%rbx 0x00007ffff7deea94 <+4>: and $0xfffffffffffffff0,%rsp ^^^^^^^^^^^ Align stack to 16 bytes 0x00007ffff7deea98 <+8>: sub $0x100,%rsp 0x00007ffff7deea9f <+15>: mov %rax,0xc0(%rsp) 0x00007ffff7deeaa7 <+23>: mov %rcx,0xc8(%rsp) 0x00007ffff7deeaaf <+31>: mov %rdx,0xd0(%rsp) 0x00007ffff7deeab7 <+39>: mov %rsi,0xd8(%rsp) 0x00007ffff7deeabf <+47>: mov %rdi,0xe0(%rsp) 0x00007ffff7deeac7 <+55>: mov %r8,0xe8(%rsp) 0x00007ffff7deeacf <+63>: mov %r9,0xf0(%rsp) 0x00007ffff7deead7 <+71>: movaps %xmm0,(%rsp) 0x00007ffff7deeadb <+75>: movaps %xmm1,0x10(%rsp) 0x00007ffff7deeae0 <+80>: movaps %xmm2,0x20(%rsp) 0x00007ffff7deeae5 <+85>: movaps %xmm3,0x30(%rsp) 0x00007ffff7deeaea <+90>: movaps %xmm4,0x40(%rsp) 0x00007ffff7deeaef <+95>: movaps %xmm5,0x50(%rsp) 0x00007ffff7deeaf4 <+100>: movaps %xmm6,0x60(%rsp) 0x00007ffff7deeaf9 <+105>: movaps %xmm7,0x70(%rsp) [BZ #19679] * sysdeps/x86_64/dl-trampoline.S (DL_RUNIME_UNALIGNED_VEC_SIZE): Renamed to ... (DL_RUNTIME_UNALIGNED_VEC_SIZE): This. Set to 8. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. Updated. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. * sysdeps/x86_64/dl-trampoline.h (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This.
2016-02-19 23:43:45 +00:00
#if DL_RUNTIME_RESOLVE_REALIGN_STACK
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# if LOCAL_STORAGE_AREA != 8
# error LOCAL_STORAGE_AREA must be 8
# endif
pushq %rbx # push subtracts stack by 8.
cfi_adjust_cfa_offset(8)
cfi_rel_offset(%rbx, 0)
mov %RSP_LP, %RBX_LP
cfi_def_cfa_register(%rbx)
and $-VEC_SIZE, %RSP_LP
#endif
sub $REGISTER_SAVE_AREA, %RSP_LP
X86-64: Correct CFA in _dl_runtime_resolve When stack is re-aligned in _dl_runtime_resolve, there is no need to adjust CFA when allocating register save area on stack. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve): Don't adjust CFA when allocating register save area on re-aligned stack.
2016-08-26 15:57:42 +00:00
#if !DL_RUNTIME_RESOLVE_REALIGN_STACK
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
cfi_adjust_cfa_offset(REGISTER_SAVE_AREA)
X86-64: Correct CFA in _dl_runtime_resolve When stack is re-aligned in _dl_runtime_resolve, there is no need to adjust CFA when allocating register save area on stack. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve): Don't adjust CFA when allocating register save area on re-aligned stack.
2016-08-26 15:57:42 +00:00
#endif
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# Preserve registers otherwise clobbered.
movq %rax, REGISTER_SAVE_RAX(%rsp)
movq %rcx, REGISTER_SAVE_RCX(%rsp)
movq %rdx, REGISTER_SAVE_RDX(%rsp)
movq %rsi, REGISTER_SAVE_RSI(%rsp)
movq %rdi, REGISTER_SAVE_RDI(%rsp)
movq %r8, REGISTER_SAVE_R8(%rsp)
movq %r9, REGISTER_SAVE_R9(%rsp)
VMOV %VEC(0), (REGISTER_SAVE_VEC_OFF)(%rsp)
VMOV %VEC(1), (REGISTER_SAVE_VEC_OFF + VEC_SIZE)(%rsp)
VMOV %VEC(2), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 2)(%rsp)
VMOV %VEC(3), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 3)(%rsp)
VMOV %VEC(4), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 4)(%rsp)
VMOV %VEC(5), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 5)(%rsp)
VMOV %VEC(6), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 6)(%rsp)
VMOV %VEC(7), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 7)(%rsp)
#ifndef __ILP32__
# We also have to preserve bound registers. These are nops if
# Intel MPX isn't available or disabled.
# ifdef HAVE_MPX_SUPPORT
bndmov %bnd0, REGISTER_SAVE_BND0(%rsp)
bndmov %bnd1, REGISTER_SAVE_BND1(%rsp)
bndmov %bnd2, REGISTER_SAVE_BND2(%rsp)
bndmov %bnd3, REGISTER_SAVE_BND3(%rsp)
# else
# if REGISTER_SAVE_BND0 == 0
.byte 0x66,0x0f,0x1b,0x04,0x24
# else
.byte 0x66,0x0f,0x1b,0x44,0x24,REGISTER_SAVE_BND0
# endif
.byte 0x66,0x0f,0x1b,0x4c,0x24,REGISTER_SAVE_BND1
.byte 0x66,0x0f,0x1b,0x54,0x24,REGISTER_SAVE_BND2
.byte 0x66,0x0f,0x1b,0x5c,0x24,REGISTER_SAVE_BND3
# endif
#endif
# Copy args pushed by PLT in register.
# %rdi: link_map, %rsi: reloc_index
mov (LOCAL_STORAGE_AREA + 8)(%BASE), %RSI_LP
mov LOCAL_STORAGE_AREA(%BASE), %RDI_LP
call _dl_fixup # Call resolver.
mov %RAX_LP, %R11_LP # Save return value
#ifndef __ILP32__
# Restore bound registers. These are nops if Intel MPX isn't
# avaiable or disabled.
# ifdef HAVE_MPX_SUPPORT
bndmov REGISTER_SAVE_BND3(%rsp), %bnd3
bndmov REGISTER_SAVE_BND2(%rsp), %bnd2
bndmov REGISTER_SAVE_BND1(%rsp), %bnd1
bndmov REGISTER_SAVE_BND0(%rsp), %bnd0
# else
.byte 0x66,0x0f,0x1a,0x5c,0x24,REGISTER_SAVE_BND3
.byte 0x66,0x0f,0x1a,0x54,0x24,REGISTER_SAVE_BND2
.byte 0x66,0x0f,0x1a,0x4c,0x24,REGISTER_SAVE_BND1
# if REGISTER_SAVE_BND0 == 0
.byte 0x66,0x0f,0x1a,0x04,0x24
# else
.byte 0x66,0x0f,0x1a,0x44,0x24,REGISTER_SAVE_BND0
# endif
# endif
#endif
# Get register content back.
movq REGISTER_SAVE_R9(%rsp), %r9
movq REGISTER_SAVE_R8(%rsp), %r8
movq REGISTER_SAVE_RDI(%rsp), %rdi
movq REGISTER_SAVE_RSI(%rsp), %rsi
movq REGISTER_SAVE_RDX(%rsp), %rdx
movq REGISTER_SAVE_RCX(%rsp), %rcx
movq REGISTER_SAVE_RAX(%rsp), %rax
VMOV (REGISTER_SAVE_VEC_OFF)(%rsp), %VEC(0)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE)(%rsp), %VEC(1)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 2)(%rsp), %VEC(2)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 3)(%rsp), %VEC(3)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 4)(%rsp), %VEC(4)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 5)(%rsp), %VEC(5)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 6)(%rsp), %VEC(6)
VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 7)(%rsp), %VEC(7)
[x86_64] Set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 Due to GCC bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58066 __tls_get_addr may be called with 8-byte stack alignment. Although this bug has been fixed in GCC 4.9.4, 5.3 and 6, we can't assume that stack will be always aligned at 16 bytes. Since SSE optimized memory/string functions with aligned SSE register load and store are used in the dynamic linker, we must set DL_RUNTIME_UNALIGNED_VEC_SIZE to 8 so that _dl_runtime_resolve_sse will align the stack before calling _dl_fixup: Dump of assembler code for function _dl_runtime_resolve_sse: 0x00007ffff7deea90 <+0>: push %rbx 0x00007ffff7deea91 <+1>: mov %rsp,%rbx 0x00007ffff7deea94 <+4>: and $0xfffffffffffffff0,%rsp ^^^^^^^^^^^ Align stack to 16 bytes 0x00007ffff7deea98 <+8>: sub $0x100,%rsp 0x00007ffff7deea9f <+15>: mov %rax,0xc0(%rsp) 0x00007ffff7deeaa7 <+23>: mov %rcx,0xc8(%rsp) 0x00007ffff7deeaaf <+31>: mov %rdx,0xd0(%rsp) 0x00007ffff7deeab7 <+39>: mov %rsi,0xd8(%rsp) 0x00007ffff7deeabf <+47>: mov %rdi,0xe0(%rsp) 0x00007ffff7deeac7 <+55>: mov %r8,0xe8(%rsp) 0x00007ffff7deeacf <+63>: mov %r9,0xf0(%rsp) 0x00007ffff7deead7 <+71>: movaps %xmm0,(%rsp) 0x00007ffff7deeadb <+75>: movaps %xmm1,0x10(%rsp) 0x00007ffff7deeae0 <+80>: movaps %xmm2,0x20(%rsp) 0x00007ffff7deeae5 <+85>: movaps %xmm3,0x30(%rsp) 0x00007ffff7deeaea <+90>: movaps %xmm4,0x40(%rsp) 0x00007ffff7deeaef <+95>: movaps %xmm5,0x50(%rsp) 0x00007ffff7deeaf4 <+100>: movaps %xmm6,0x60(%rsp) 0x00007ffff7deeaf9 <+105>: movaps %xmm7,0x70(%rsp) [BZ #19679] * sysdeps/x86_64/dl-trampoline.S (DL_RUNIME_UNALIGNED_VEC_SIZE): Renamed to ... (DL_RUNTIME_UNALIGNED_VEC_SIZE): This. Set to 8. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. Updated. (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This. * sysdeps/x86_64/dl-trampoline.h (DL_RUNIME_RESOLVE_REALIGN_STACK): Renamed to ... (DL_RUNTIME_RESOLVE_REALIGN_STACK): This.
2016-02-19 23:43:45 +00:00
#if DL_RUNTIME_RESOLVE_REALIGN_STACK
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
mov %RBX_LP, %RSP_LP
cfi_def_cfa_register(%rsp)
movq (%rsp), %rbx
cfi_restore(%rbx)
#endif
# Adjust stack(PLT did 2 pushes)
add $(LOCAL_STORAGE_AREA + 16), %RSP_LP
cfi_adjust_cfa_offset(-(LOCAL_STORAGE_AREA + 16))
# Preserve bound registers.
PRESERVE_BND_REGS_PREFIX
jmp *%r11 # Jump to function address.
cfi_endproc
.size _dl_runtime_resolve, .-_dl_runtime_resolve
X86-64: Add _dl_runtime_resolve_avx[512]_{opt|slow} [BZ #20508] There is transition penalty when SSE instructions are mixed with 256-bit AVX or 512-bit AVX512 load instructions. Since _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 save/restore 256-bit YMM/512-bit ZMM registers, there is transition penalty when SSE instructions are used with lazy binding on AVX and AVX512 processors. To avoid SSE transition penalty, if only the lower 128 bits of the first 8 vector registers are non-zero, we can preserve %xmm0 - %xmm7 registers with the zero upper bits. For AVX and AVX512 processors which support XGETBV with ECX == 1, we can use XGETBV with ECX == 1 to check if the upper 128 bits of YMM registers or the upper 256 bits of ZMM registers are zero. We can restore only the non-zero portion of vector registers with AVX/AVX512 load instructions which will zero-extend upper bits of vector registers. This patch adds _dl_runtime_resolve_sse_vex which saves and restores XMM registers with 128-bit AVX store/load instructions. It is used to preserve YMM/ZMM registers when only the lower 128 bits are non-zero. _dl_runtime_resolve_avx_opt and _dl_runtime_resolve_avx512_opt are added and used on AVX/AVX512 processors supporting XGETBV with ECX == 1 so that we store and load only the non-zero portion of vector registers. This avoids SSE transition penalty caused by _dl_runtime_resolve_avx and _dl_runtime_profile_avx512 when only the lower 128 bits of vector registers are used. _dl_runtime_resolve_avx_slow is added and used for AVX processors which don't support XGETBV with ECX == 1. Since there is no SSE transition penalty on AVX512 processors which don't support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't provided. [BZ #20495] [BZ #20508] * sysdeps/x86/cpu-features.c (init_cpu_features): For Intel processors, set Use_dl_runtime_resolve_slow and set Use_dl_runtime_resolve_opt if XGETBV suports ECX == 1. * sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt): New. (bit_arch_Use_dl_runtime_resolve_slow): Likewise. (index_arch_Use_dl_runtime_resolve_opt): Likewise. (index_arch_Use_dl_runtime_resolve_slow): Likewise. * sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Use _dl_runtime_resolve_avx512_opt and _dl_runtime_resolve_avx_opt if Use_dl_runtime_resolve_opt is set. Use _dl_runtime_resolve_slow if Use_dl_runtime_resolve_slow is set. * sysdeps/x86_64/dl-trampoline.S: Include <cpu-features.h>. (_dl_runtime_resolve_opt): New. Defined for AVX and AVX512. (_dl_runtime_resolve): Add one for _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx_slow): New. (_dl_runtime_resolve_opt): Likewise. (_dl_runtime_profile): Define only if _dl_runtime_profile is defined.
2016-09-06 15:50:55 +00:00
/* To preserve %xmm0 - %xmm7 registers, dl-trampoline.h is included
twice, for _dl_runtime_resolve_sse and _dl_runtime_resolve_sse_vex.
But we don't need another _dl_runtime_profile for XMM registers. */
#if !defined PROF && defined _dl_runtime_profile
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# if (LR_VECTOR_OFFSET % VEC_SIZE) != 0
# error LR_VECTOR_OFFSET must be multples of VEC_SIZE
# endif
.globl _dl_runtime_profile
.hidden _dl_runtime_profile
.type _dl_runtime_profile, @function
.align 16
_dl_runtime_profile:
cfi_startproc
cfi_adjust_cfa_offset(16) # Incorporate PLT
/* The La_x86_64_regs data structure pointed to by the
fourth paramater must be VEC_SIZE-byte aligned. This must
be explicitly enforced. We have the set up a dynamically
sized stack frame. %rbx points to the top half which
has a fixed size and preserves the original stack pointer. */
sub $32, %RSP_LP # Allocate the local storage.
cfi_adjust_cfa_offset(32)
movq %rbx, (%rsp)
cfi_rel_offset(%rbx, 0)
/* On the stack:
56(%rbx) parameter #1
48(%rbx) return address
40(%rbx) reloc index
32(%rbx) link_map
24(%rbx) La_x86_64_regs pointer
16(%rbx) framesize
8(%rbx) rax
(%rbx) rbx
*/
movq %rax, 8(%rsp)
mov %RSP_LP, %RBX_LP
cfi_def_cfa_register(%rbx)
/* Actively align the La_x86_64_regs structure. */
and $-VEC_SIZE, %RSP_LP
/* sizeof(La_x86_64_regs). Need extra space for 8 SSE registers
to detect if any xmm0-xmm7 registers are changed by audit
module. */
sub $(LR_SIZE + XMM_SIZE*8), %RSP_LP
movq %rsp, 24(%rbx)
/* Fill the La_x86_64_regs structure. */
movq %rdx, LR_RDX_OFFSET(%rsp)
movq %r8, LR_R8_OFFSET(%rsp)
movq %r9, LR_R9_OFFSET(%rsp)
movq %rcx, LR_RCX_OFFSET(%rsp)
movq %rsi, LR_RSI_OFFSET(%rsp)
movq %rdi, LR_RDI_OFFSET(%rsp)
movq %rbp, LR_RBP_OFFSET(%rsp)
lea 48(%rbx), %RAX_LP
movq %rax, LR_RSP_OFFSET(%rsp)
/* We always store the XMM registers even if AVX is available.
This is to provide backward binary compatibility for existing
audit modules. */
movaps %xmm0, (LR_XMM_OFFSET)(%rsp)
movaps %xmm1, (LR_XMM_OFFSET + XMM_SIZE)(%rsp)
movaps %xmm2, (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp)
movaps %xmm3, (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp)
movaps %xmm4, (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp)
movaps %xmm5, (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp)
movaps %xmm6, (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp)
movaps %xmm7, (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp)
# ifndef __ILP32__
# ifdef HAVE_MPX_SUPPORT
bndmov %bnd0, (LR_BND_OFFSET)(%rsp) # Preserve bound
bndmov %bnd1, (LR_BND_OFFSET + BND_SIZE)(%rsp) # registers. Nops if
bndmov %bnd2, (LR_BND_OFFSET + BND_SIZE*2)(%rsp) # MPX not available
bndmov %bnd3, (LR_BND_OFFSET + BND_SIZE*3)(%rsp) # or disabled.
# else
.byte 0x66,0x0f,0x1b,0x84,0x24;.long (LR_BND_OFFSET)
.byte 0x66,0x0f,0x1b,0x8c,0x24;.long (LR_BND_OFFSET + BND_SIZE)
.byte 0x66,0x0f,0x1b,0x94,0x24;.long (LR_BND_OFFSET + BND_SIZE*2)
.byte 0x66,0x0f,0x1b,0x9c,0x24;.long (LR_BND_OFFSET + BND_SIZE*3)
# endif
# endif
# ifdef RESTORE_AVX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* This is to support AVX audit modules. */
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
VMOVA %VEC(0), (LR_VECTOR_OFFSET)(%rsp)
VMOVA %VEC(1), (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp)
VMOVA %VEC(2), (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp)
VMOVA %VEC(3), (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp)
VMOVA %VEC(4), (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp)
VMOVA %VEC(5), (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp)
VMOVA %VEC(6), (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp)
VMOVA %VEC(7), (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Save xmm0-xmm7 registers to detect if any of them are
changed by audit module. */
vmovdqa %xmm0, (LR_SIZE)(%rsp)
vmovdqa %xmm1, (LR_SIZE + XMM_SIZE)(%rsp)
vmovdqa %xmm2, (LR_SIZE + XMM_SIZE*2)(%rsp)
vmovdqa %xmm3, (LR_SIZE + XMM_SIZE*3)(%rsp)
vmovdqa %xmm4, (LR_SIZE + XMM_SIZE*4)(%rsp)
vmovdqa %xmm5, (LR_SIZE + XMM_SIZE*5)(%rsp)
vmovdqa %xmm6, (LR_SIZE + XMM_SIZE*6)(%rsp)
vmovdqa %xmm7, (LR_SIZE + XMM_SIZE*7)(%rsp)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
Use R*_LP to call _dl_profile_fixup and frame size
2012-05-15 21:24:00 +00:00
mov %RSP_LP, %RCX_LP # La_x86_64_regs pointer to %rcx.
mov 48(%rbx), %RDX_LP # Load return address if needed.
mov 40(%rbx), %RSI_LP # Copy args pushed by PLT in register.
mov 32(%rbx), %RDI_LP # %rdi: link_map, %rsi: reloc_index
lea 16(%rbx), %R8_LP # Address of framesize
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
call _dl_profile_fixup # Call resolver.
Use R*_LP to call _dl_profile_fixup and frame size
2012-05-15 21:24:00 +00:00
mov %RAX_LP, %R11_LP # Save return value.
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
movq 8(%rbx), %rax # Get back register content.
movq LR_RDX_OFFSET(%rsp), %rdx
movq LR_R8_OFFSET(%rsp), %r8
movq LR_R9_OFFSET(%rsp), %r9
movaps (LR_XMM_OFFSET)(%rsp), %xmm0
movaps (LR_XMM_OFFSET + XMM_SIZE)(%rsp), %xmm1
movaps (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp), %xmm2
movaps (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp), %xmm3
movaps (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp), %xmm4
movaps (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp), %xmm5
movaps (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp), %xmm6
movaps (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp), %xmm7
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifdef RESTORE_AVX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Check if any xmm0-xmm7 registers are changed by audit
module. */
vpcmpeqq (LR_SIZE)(%rsp), %xmm0, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm0, (LR_VECTOR_OFFSET)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET)(%rsp), %VEC(0)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm0, (LR_XMM_OFFSET)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm1, (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp), %VEC(1)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm1, (LR_XMM_OFFSET + XMM_SIZE)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*2)(%rsp), %xmm2, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm2, (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp), %VEC(2)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm2, (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*3)(%rsp), %xmm3, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm3, (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp), %VEC(3)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm3, (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*4)(%rsp), %xmm4, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm4, (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp), %VEC(4)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm4, (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*5)(%rsp), %xmm5, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm5, (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp), %VEC(5)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm5, (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*6)(%rsp), %xmm6, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm6, (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp), %VEC(6)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm6, (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp)
1: vpcmpeqq (LR_SIZE + XMM_SIZE*7)(%rsp), %xmm7, %xmm8
vpmovmskb %xmm8, %esi
cmpl $0xffff, %esi
je 2f
vmovdqa %xmm7, (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp)
jmp 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
2: VMOVA (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp), %VEC(7)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
vmovdqa %xmm7, (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp)
1:
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifndef __ILP32__
# ifdef HAVE_MPX_SUPPORT
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
bndmov (LR_BND_OFFSET)(%rsp), %bnd0 # Restore bound
bndmov (LR_BND_OFFSET + BND_SIZE)(%rsp), %bnd1 # registers.
bndmov (LR_BND_OFFSET + BND_SIZE*2)(%rsp), %bnd2
bndmov (LR_BND_OFFSET + BND_SIZE*3)(%rsp), %bnd3
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# else
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
.byte 0x66,0x0f,0x1a,0x84,0x24;.long (LR_BND_OFFSET)
.byte 0x66,0x0f,0x1a,0x8c,0x24;.long (LR_BND_OFFSET + BND_SIZE)
.byte 0x66,0x0f,0x1a,0x94,0x24;.long (LR_BND_OFFSET + BND_SIZE*2)
.byte 0x66,0x0f,0x1a,0x9c,0x24;.long (LR_BND_OFFSET + BND_SIZE*3)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
# endif
Use R*_LP to call _dl_profile_fixup and frame size
2012-05-15 21:24:00 +00:00
mov 16(%rbx), %R10_LP # Anything in framesize?
test %R10_LP, %R10_LP
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
PRESERVE_BND_REGS_PREFIX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
jns 3f
/* There's nothing in the frame size, so there
will be no call to the _dl_call_pltexit. */
/* Get back registers content. */
movq LR_RCX_OFFSET(%rsp), %rcx
movq LR_RSI_OFFSET(%rsp), %rsi
movq LR_RDI_OFFSET(%rsp), %rdi
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
mov %RBX_LP, %RSP_LP
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
movq (%rsp), %rbx
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
cfi_restore(%rbx)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
cfi_def_cfa_register(%rsp)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
add $48, %RSP_LP # Adjust the stack to the return value
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
# (eats the reloc index and link_map)
cfi_adjust_cfa_offset(-48)
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
PRESERVE_BND_REGS_PREFIX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
jmp *%r11 # Jump to function address.
3:
cfi_adjust_cfa_offset(48)
cfi_rel_offset(%rbx, 0)
cfi_def_cfa_register(%rbx)
/* At this point we need to prepare new stack for the function
which has to be called. We copy the original stack to a
temporary buffer of the size specified by the 'framesize'
returned from _dl_profile_fixup */
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
lea LR_RSP_OFFSET(%rbx), %RSI_LP # stack
add $8, %R10_LP
and $-16, %R10_LP
mov %R10_LP, %RCX_LP
sub %R10_LP, %RSP_LP
mov %RSP_LP, %RDI_LP
shr $3, %RCX_LP
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
rep
movsq
movq 24(%rdi), %rcx # Get back register content.
movq 32(%rdi), %rsi
movq 40(%rdi), %rdi
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
PRESERVE_BND_REGS_PREFIX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
call *%r11
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
mov 24(%rbx), %RSP_LP # Drop the copied stack content
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Now we have to prepare the La_x86_64_retval structure for the
_dl_call_pltexit. The La_x86_64_regs is being pointed by rsp now,
so we just need to allocate the sizeof(La_x86_64_retval) space on
the stack, since the alignment has already been taken care of. */
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifdef RESTORE_AVX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* sizeof(La_x86_64_retval). Need extra space for 2 SSE
registers to detect if xmm0/xmm1 registers are changed
by audit module. */
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
sub $(LRV_SIZE + XMM_SIZE*2), %RSP_LP
# else
sub $LRV_SIZE, %RSP_LP # sizeof(La_x86_64_retval)
# endif
mov %RSP_LP, %RCX_LP # La_x86_64_retval argument to %rcx.
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Fill in the La_x86_64_retval structure. */
movq %rax, LRV_RAX_OFFSET(%rcx)
movq %rdx, LRV_RDX_OFFSET(%rcx)
movaps %xmm0, LRV_XMM0_OFFSET(%rcx)
movaps %xmm1, LRV_XMM1_OFFSET(%rcx)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifdef RESTORE_AVX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* This is to support AVX audit modules. */
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
VMOVA %VEC(0), LRV_VECTOR0_OFFSET(%rcx)
VMOVA %VEC(1), LRV_VECTOR1_OFFSET(%rcx)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Save xmm0/xmm1 registers to detect if they are changed
by audit module. */
vmovdqa %xmm0, (LRV_SIZE)(%rcx)
vmovdqa %xmm1, (LRV_SIZE + XMM_SIZE)(%rcx)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifndef __ILP32__
# ifdef HAVE_MPX_SUPPORT
Save/restore bound registers for _dl_runtime_profile This patch saves and restores bound registers in x86-64 PLT for ld.so profile and LD_AUDIT: * sysdeps/x86_64/bits/link.h (La_x86_64_regs): Add lr_bnd. (La_x86_64_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Save Intel MPX bound registers before _dl_profile_fixup. * sysdeps/x86_64/dl-trampoline.h: Restore Intel MPX bound registers after _dl_profile_fixup. Save and restore bound registers bnd0/bnd1 when calling _dl_call_pltexit. * sysdeps/x86_64/link-defines.sym (BND_SIZE): New. (LR_BND_OFFSET): Likewise. (LRV_BND0_OFFSET): Likewise. (LRV_BND1_OFFSET): Likewise.
2014-04-16 21:43:16 +00:00
bndmov %bnd0, LRV_BND0_OFFSET(%rcx) # Preserve returned bounds.
bndmov %bnd1, LRV_BND1_OFFSET(%rcx)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# else
Save/restore bound registers for _dl_runtime_profile This patch saves and restores bound registers in x86-64 PLT for ld.so profile and LD_AUDIT: * sysdeps/x86_64/bits/link.h (La_x86_64_regs): Add lr_bnd. (La_x86_64_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Save Intel MPX bound registers before _dl_profile_fixup. * sysdeps/x86_64/dl-trampoline.h: Restore Intel MPX bound registers after _dl_profile_fixup. Save and restore bound registers bnd0/bnd1 when calling _dl_call_pltexit. * sysdeps/x86_64/link-defines.sym (BND_SIZE): New. (LR_BND_OFFSET): Likewise. (LRV_BND0_OFFSET): Likewise. (LRV_BND1_OFFSET): Likewise.
2014-04-16 21:43:16 +00:00
.byte 0x66,0x0f,0x1b,0x81;.long (LRV_BND0_OFFSET)
.byte 0x66,0x0f,0x1b,0x89;.long (LRV_BND1_OFFSET)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Save/restore bound registers for _dl_runtime_profile This patch saves and restores bound registers in x86-64 PLT for ld.so profile and LD_AUDIT: * sysdeps/x86_64/bits/link.h (La_x86_64_regs): Add lr_bnd. (La_x86_64_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Save Intel MPX bound registers before _dl_profile_fixup. * sysdeps/x86_64/dl-trampoline.h: Restore Intel MPX bound registers after _dl_profile_fixup. Save and restore bound registers bnd0/bnd1 when calling _dl_call_pltexit. * sysdeps/x86_64/link-defines.sym (BND_SIZE): New. (LR_BND_OFFSET): Likewise. (LRV_BND0_OFFSET): Likewise. (LRV_BND1_OFFSET): Likewise.
2014-04-16 21:43:16 +00:00
# endif
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
fstpt LRV_ST0_OFFSET(%rcx)
fstpt LRV_ST1_OFFSET(%rcx)
movq 24(%rbx), %rdx # La_x86_64_regs argument to %rdx.
movq 40(%rbx), %rsi # Copy args pushed by PLT in register.
Fix CFI info in x86-64 trampolines for non-AVX code
2011-08-20 12:56:30 +00:00
movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
call _dl_call_pltexit
/* Restore return registers. */
movq LRV_RAX_OFFSET(%rsp), %rax
movq LRV_RDX_OFFSET(%rsp), %rdx
movaps LRV_XMM0_OFFSET(%rsp), %xmm0
movaps LRV_XMM1_OFFSET(%rsp), %xmm1
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifdef RESTORE_AVX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
/* Check if xmm0/xmm1 registers are changed by audit module. */
vpcmpeqq (LRV_SIZE)(%rsp), %xmm0, %xmm2
vpmovmskb %xmm2, %esi
cmpl $0xffff, %esi
jne 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
VMOVA LRV_VECTOR0_OFFSET(%rsp), %VEC(0)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
1: vpcmpeqq (LRV_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm2
vpmovmskb %xmm2, %esi
cmpl $0xffff, %esi
jne 1f
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
VMOVA LRV_VECTOR1_OFFSET(%rsp), %VEC(1)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
1:
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Save/restore bound registers for _dl_runtime_profile This patch saves and restores bound registers in x86-64 PLT for ld.so profile and LD_AUDIT: * sysdeps/x86_64/bits/link.h (La_x86_64_regs): Add lr_bnd. (La_x86_64_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Save Intel MPX bound registers before _dl_profile_fixup. * sysdeps/x86_64/dl-trampoline.h: Restore Intel MPX bound registers after _dl_profile_fixup. Save and restore bound registers bnd0/bnd1 when calling _dl_call_pltexit. * sysdeps/x86_64/link-defines.sym (BND_SIZE): New. (LR_BND_OFFSET): Likewise. (LRV_BND0_OFFSET): Likewise. (LRV_BND1_OFFSET): Likewise.
2014-04-16 21:43:16 +00:00
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# ifndef __ILP32__
# ifdef HAVE_MPX_SUPPORT
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
bndmov LRV_BND0_OFFSET(%rsp), %bnd0 # Restore bound registers.
bndmov LRV_BND1_OFFSET(%rsp), %bnd1
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# else
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
.byte 0x66,0x0f,0x1a,0x84,0x24;.long (LRV_BND0_OFFSET)
.byte 0x66,0x0f,0x1a,0x8c,0x24;.long (LRV_BND1_OFFSET)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
# endif
Save/restore bound registers for _dl_runtime_profile This patch saves and restores bound registers in x86-64 PLT for ld.so profile and LD_AUDIT: * sysdeps/x86_64/bits/link.h (La_x86_64_regs): Add lr_bnd. (La_x86_64_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Save Intel MPX bound registers before _dl_profile_fixup. * sysdeps/x86_64/dl-trampoline.h: Restore Intel MPX bound registers after _dl_profile_fixup. Save and restore bound registers bnd0/bnd1 when calling _dl_call_pltexit. * sysdeps/x86_64/link-defines.sym (BND_SIZE): New. (LR_BND_OFFSET): Likewise. (LRV_BND0_OFFSET): Likewise. (LRV_BND1_OFFSET): Likewise.
2014-04-16 21:43:16 +00:00
# endif
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
fldt LRV_ST1_OFFSET(%rsp)
fldt LRV_ST0_OFFSET(%rsp)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
mov %RBX_LP, %RSP_LP
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
movq (%rsp), %rbx
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
cfi_restore(%rbx)
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
cfi_def_cfa_register(%rsp)
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
add $48, %RSP_LP # Adjust the stack to the return value
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
# (eats the reloc index and link_map)
cfi_adjust_cfa_offset(-48)
Preserve bound registers for pointer pass/return We need to save/restore bound registers and add a BND prefix before branches in _dl_runtime_profile so that bound registers for pointer pass and return are preserved when LD_AUDIT is used. [BZ #18134] * sysdeps/i386/configure.ac: Set HAVE_MPX_SUPPORT. * sysdeps/i386/configure: Regenerated. * sysdeps/i386/dl-trampoline.S (PRESERVE_BND_REGS_PREFIX): New. (_dl_runtime_profile): Save and restore Intel MPX return bound registers when calling _dl_call_pltexit. Add PRESERVE_BND_REGS_PREFIX before return. * sysdeps/i386/link-defines.sym (LRV_BND0_OFFSET): New. (LRV_BND1_OFFSET): Likewise. * sysdeps/x86/bits/link.h (La_i86_retval): Add lrv_bnd0 and lrv_bnd1. * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix typo in bndmov encoding. * sysdeps/x86_64/dl-trampoline.h: Properly save and restore Intel MPX bound registers. Add PRESERVE_BND_REGS_PREFIX before branch instructions to preserve bounds.
2015-07-09 13:50:12 +00:00
PRESERVE_BND_REGS_PREFIX
Support mixed SSE/AVX audit and check AVX only once. This patch fixes mixed SSE/AVX audit and checks AVX only once in _dl_runtime_profile. When an AVX or SSE register value in pltenter is modified, we have to make sure that the SSE part value is the same in both lr_xmm and lr_vector fields so that pltexit will get the correct value from either lr_xmm or lr_vector fields. AVX-enabled pltenter should update both lr_xmm and lr_vector fields to support stacked AVX/SSE pltenter functions.
2009-08-08 17:54:42 +00:00
retq
Fix CFI info in x86-64 trampolines for non-AVX code
2011-08-20 12:56:30 +00:00
Save and restore vector registers in x86-64 ld.so This patch adds SSE, AVX and AVX512 versions of _dl_runtime_resolve and _dl_runtime_profile, which save and restore the first 8 vector registers used for parameter passing. elf_machine_runtime_setup selects the proper _dl_runtime_resolve or _dl_runtime_profile based on _dl_x86_cpu_features. It avoids race condition caused by FOREIGN_CALL macros, which are only used for x86-64. Performance impact of saving and restoring 8 vector registers are negligible on Nehalem, Sandy Bridge, Ivy Bridge and Haswell when ld.so is optimized with SSE2. [BZ #15128] * sysdeps/x86_64/Makefile [$(subdir) == elf] (tests): Add ifuncmain8. (modules-names): Add ifuncmod8. ($(objpfx)ifuncmain8): New rule. * sysdeps/x86_64/dl-machine.h: Include <dl-procinfo.h> and <cpuid.h>. (elf_machine_runtime_setup): Use _dl_runtime_resolve_sse, _dl_runtime_resolve_avx, or _dl_runtime_resolve_avx512, _dl_runtime_profile_sse, _dl_runtime_profile_avx, or _dl_runtime_profile_avx512, based on HAS_ARCH_FEATURE. * sysdeps/x86_64/dl-trampoline.S: Rewrite. * sysdeps/x86_64/dl-trampoline.h: Likewise. * sysdeps/x86_64/ifuncmain8.c: New file. * sysdeps/x86_64/ifuncmod8.c: Likewise. * sysdeps/x86_64/nptl/tcb-offsets.sym (RTLD_SAVESPACE_SSE): Removed. * sysdeps/x86_64/nptl/tls.h (__128bits): Removed. (tcbhead_t): Change rtld_must_xmm_save to __glibc_unused1. Change rtld_savespace_sse to __glibc_unused2. (RTLD_CHECK_FOREIGN_CALL): Removed. (RTLD_ENABLE_FOREIGN_CALL): Likewise. (RTLD_PREPARE_FOREIGN_CALL): Likewise. (RTLD_FINALIZE_FOREIGN_CALL): Likewise.
2015-08-25 11:33:54 +00:00
cfi_endproc
.size _dl_runtime_profile, .-_dl_runtime_profile
Fix CFI info in x86-64 trampolines for non-AVX code
2011-08-20 12:56:30 +00:00
#endif
Reference in New Issue Copy Permalink