1/* PLT trampolines. x86-64 version.
2 Copyright (C) 2004-2016 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
9
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
14
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library; if not, see
17 <http://www.gnu.org/licenses/>. */
18
19#include <config.h>
20#include <sysdep.h>
21#include <link-defines.h>
22
23#ifndef DL_STACK_ALIGNMENT
24/* Due to GCC bug:
25
26 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58066
27
28 __tls_get_addr may be called with 8-byte stack alignment. Although
29 this bug has been fixed in GCC 4.9.4, 5.3 and 6, we can't assume
30 that stack will be always aligned at 16 bytes. We use unaligned
31 16-byte move to load and store SSE registers, which has no penalty
32 on modern processors if stack is 16-byte aligned. */
33# define DL_STACK_ALIGNMENT 8
34#endif
35
36#ifndef DL_RUNTIME_UNALIGNED_VEC_SIZE
37/* The maximum size in bytes of unaligned vector load and store in the
38 dynamic linker. Since SSE optimized memory/string functions with
39 aligned SSE register load and store are used in the dynamic linker,
40 we must set this to 8 so that _dl_runtime_resolve_sse will align the
41 stack before calling _dl_fixup. */
42# define DL_RUNTIME_UNALIGNED_VEC_SIZE 8
43#endif
44
45/* True if _dl_runtime_resolve should align stack to VEC_SIZE bytes. */
46#define DL_RUNTIME_RESOLVE_REALIGN_STACK \
47 (VEC_SIZE > DL_STACK_ALIGNMENT \
48 && VEC_SIZE > DL_RUNTIME_UNALIGNED_VEC_SIZE)
49
50/* Align vector register save area to 16 bytes. */
51#define REGISTER_SAVE_VEC_OFF 0
52
53/* Area on stack to save and restore registers used for parameter
54 passing when calling _dl_fixup. */
55#ifdef __ILP32__
56# define REGISTER_SAVE_RAX (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 8)
57# define PRESERVE_BND_REGS_PREFIX
58#else
59/* Align bound register save area to 16 bytes. */
60# define REGISTER_SAVE_BND0 (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 8)
61# define REGISTER_SAVE_BND1 (REGISTER_SAVE_BND0 + 16)
62# define REGISTER_SAVE_BND2 (REGISTER_SAVE_BND1 + 16)
63# define REGISTER_SAVE_BND3 (REGISTER_SAVE_BND2 + 16)
64# define REGISTER_SAVE_RAX (REGISTER_SAVE_BND3 + 16)
65# ifdef HAVE_MPX_SUPPORT
66# define PRESERVE_BND_REGS_PREFIX bnd
67# else
68# define PRESERVE_BND_REGS_PREFIX .byte 0xf2
69# endif
70#endif
71#define REGISTER_SAVE_RCX (REGISTER_SAVE_RAX + 8)
72#define REGISTER_SAVE_RDX (REGISTER_SAVE_RCX + 8)
73#define REGISTER_SAVE_RSI (REGISTER_SAVE_RDX + 8)
74#define REGISTER_SAVE_RDI (REGISTER_SAVE_RSI + 8)
75#define REGISTER_SAVE_R8 (REGISTER_SAVE_RDI + 8)
76#define REGISTER_SAVE_R9 (REGISTER_SAVE_R8 + 8)
77
78#define RESTORE_AVX
79
80#define VEC_SIZE 64
81#define VMOVA vmovdqa64
82#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
83# define VMOV vmovdqa64
84#else
85# define VMOV vmovdqu64
86#endif
87#define VEC(i) zmm##i
88#define _dl_runtime_resolve _dl_runtime_resolve_avx512
89#define _dl_runtime_profile _dl_runtime_profile_avx512
90#include "dl-trampoline.h"
91#undef _dl_runtime_resolve
92#undef _dl_runtime_profile
93#undef VEC
94#undef VMOV
95#undef VMOVA
96#undef VEC_SIZE
97
98#define VEC_SIZE 32
99#define VMOVA vmovdqa
100#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
101# define VMOV vmovdqa
102#else
103# define VMOV vmovdqu
104#endif
105#define VEC(i) ymm##i
106#define _dl_runtime_resolve _dl_runtime_resolve_avx
107#define _dl_runtime_profile _dl_runtime_profile_avx
108#include "dl-trampoline.h"
109#undef _dl_runtime_resolve
110#undef _dl_runtime_profile
111#undef VEC
112#undef VMOV
113#undef VMOVA
114#undef VEC_SIZE
115
116/* movaps/movups is 1-byte shorter. */
117#define VEC_SIZE 16
118#define VMOVA movaps
119#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
120# define VMOV movaps
121#else
122# define VMOV movups
123#endif
124#define VEC(i) xmm##i
125#define _dl_runtime_resolve _dl_runtime_resolve_sse
126#define _dl_runtime_profile _dl_runtime_profile_sse
127#undef RESTORE_AVX
128#include "dl-trampoline.h"
129