1 | /* Copyright (C) 2001-2019 Free Software Foundation, Inc. |
2 | This file is part of the GNU C Library. |
3 | |
4 | The GNU C Library is free software; you can redistribute it and/or |
5 | modify it under the terms of the GNU Lesser General Public |
6 | License as published by the Free Software Foundation; either |
7 | version 2.1 of the License, or (at your option) any later version. |
8 | |
9 | The GNU C Library is distributed in the hope that it will be useful, |
10 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
12 | Lesser General Public License for more details. |
13 | |
14 | You should have received a copy of the GNU Lesser General Public |
15 | License along with the GNU C Library; if not, see |
16 | <http://www.gnu.org/licenses/>. */ |
17 | |
18 | #ifndef _LINUX_X86_64_SYSDEP_H |
19 | #define _LINUX_X86_64_SYSDEP_H 1 |
20 | |
21 | /* Always enable vsyscalls on x86_64 */ |
22 | #define ALWAYS_USE_VSYSCALL 1 |
23 | |
24 | /* There is some commonality. */ |
25 | #include <sysdeps/unix/sysv/linux/sysdep.h> |
26 | #include <sysdeps/unix/x86_64/sysdep.h> |
27 | #include <tls.h> |
28 | |
29 | /* Defines RTLD_PRIVATE_ERRNO. */ |
30 | #include <dl-sysdep.h> |
31 | |
32 | /* For Linux we can use the system call table in the header file |
33 | /usr/include/asm/unistd.h |
34 | of the kernel. But these symbols do not follow the SYS_* syntax |
35 | so we have to redefine the `SYS_ify' macro here. */ |
36 | #undef SYS_ify |
37 | #define SYS_ify(syscall_name) __NR_##syscall_name |
38 | |
39 | /* This is a kludge to make syscalls.list find these under the names |
40 | pread and pwrite, since some kernel headers define those names |
41 | and some define the *64 names for the same system calls. */ |
42 | #if !defined __NR_pread && defined __NR_pread64 |
43 | # define __NR_pread __NR_pread64 |
44 | #endif |
45 | #if !defined __NR_pwrite && defined __NR_pwrite64 |
46 | # define __NR_pwrite __NR_pwrite64 |
47 | #endif |
48 | |
49 | /* This is to help the old kernel headers where __NR_semtimedop is not |
50 | available. */ |
51 | #ifndef __NR_semtimedop |
52 | # define __NR_semtimedop 220 |
53 | #endif |
54 | |
55 | |
56 | #ifdef __ASSEMBLER__ |
57 | |
58 | /* Linux uses a negative return value to indicate syscall errors, |
59 | unlike most Unices, which use the condition codes' carry flag. |
60 | |
61 | Since version 2.1 the return value of a system call might be |
62 | negative even if the call succeeded. E.g., the `lseek' system call |
63 | might return a large offset. Therefore we must not anymore test |
64 | for < 0, but test for a real error by making sure the value in %eax |
65 | is a real error number. Linus said he will make sure the no syscall |
66 | returns a value in -1 .. -4095 as a valid result so we can savely |
67 | test with -4095. */ |
68 | |
69 | /* We don't want the label for the error handle to be global when we define |
70 | it here. */ |
71 | # ifdef PIC |
72 | # define SYSCALL_ERROR_LABEL 0f |
73 | # else |
74 | # define SYSCALL_ERROR_LABEL syscall_error |
75 | # endif |
76 | |
77 | # undef PSEUDO |
78 | # define PSEUDO(name, syscall_name, args) \ |
79 | .text; \ |
80 | ENTRY (name) \ |
81 | DO_CALL (syscall_name, args); \ |
82 | cmpq $-4095, %rax; \ |
83 | jae SYSCALL_ERROR_LABEL |
84 | |
85 | # undef PSEUDO_END |
86 | # define PSEUDO_END(name) \ |
87 | SYSCALL_ERROR_HANDLER \ |
88 | END (name) |
89 | |
90 | # undef PSEUDO_NOERRNO |
91 | # define PSEUDO_NOERRNO(name, syscall_name, args) \ |
92 | .text; \ |
93 | ENTRY (name) \ |
94 | DO_CALL (syscall_name, args) |
95 | |
96 | # undef PSEUDO_END_NOERRNO |
97 | # define PSEUDO_END_NOERRNO(name) \ |
98 | END (name) |
99 | |
100 | # define ret_NOERRNO ret |
101 | |
102 | # undef PSEUDO_ERRVAL |
103 | # define PSEUDO_ERRVAL(name, syscall_name, args) \ |
104 | .text; \ |
105 | ENTRY (name) \ |
106 | DO_CALL (syscall_name, args); \ |
107 | negq %rax |
108 | |
109 | # undef PSEUDO_END_ERRVAL |
110 | # define PSEUDO_END_ERRVAL(name) \ |
111 | END (name) |
112 | |
113 | # define ret_ERRVAL ret |
114 | |
115 | # if defined PIC && RTLD_PRIVATE_ERRNO |
116 | # define SYSCALL_SET_ERRNO \ |
117 | lea rtld_errno(%rip), %RCX_LP; \ |
118 | neg %eax; \ |
119 | movl %eax, (%rcx) |
120 | # else |
121 | # if IS_IN (libc) |
122 | # define SYSCALL_ERROR_ERRNO __libc_errno |
123 | # else |
124 | # define SYSCALL_ERROR_ERRNO errno |
125 | # endif |
126 | # define SYSCALL_SET_ERRNO \ |
127 | movq SYSCALL_ERROR_ERRNO@GOTTPOFF(%rip), %rcx;\ |
128 | neg %eax; \ |
129 | movl %eax, %fs:(%rcx); |
130 | # endif |
131 | |
132 | # ifndef PIC |
133 | # define SYSCALL_ERROR_HANDLER /* Nothing here; code in sysdep.S is used. */ |
134 | # else |
135 | # define SYSCALL_ERROR_HANDLER \ |
136 | 0: \ |
137 | SYSCALL_SET_ERRNO; \ |
138 | or $-1, %RAX_LP; \ |
139 | ret; |
140 | # endif /* PIC */ |
141 | |
142 | /* The Linux/x86-64 kernel expects the system call parameters in |
143 | registers according to the following table: |
144 | |
145 | syscall number rax |
146 | arg 1 rdi |
147 | arg 2 rsi |
148 | arg 3 rdx |
149 | arg 4 r10 |
150 | arg 5 r8 |
151 | arg 6 r9 |
152 | |
153 | The Linux kernel uses and destroys internally these registers: |
154 | return address from |
155 | syscall rcx |
156 | eflags from syscall r11 |
157 | |
158 | Normal function call, including calls to the system call stub |
159 | functions in the libc, get the first six parameters passed in |
160 | registers and the seventh parameter and later on the stack. The |
161 | register use is as follows: |
162 | |
163 | system call number in the DO_CALL macro |
164 | arg 1 rdi |
165 | arg 2 rsi |
166 | arg 3 rdx |
167 | arg 4 rcx |
168 | arg 5 r8 |
169 | arg 6 r9 |
170 | |
171 | We have to take care that the stack is aligned to 16 bytes. When |
172 | called the stack is not aligned since the return address has just |
173 | been pushed. |
174 | |
175 | |
176 | Syscalls of more than 6 arguments are not supported. */ |
177 | |
178 | # undef DO_CALL |
179 | # define DO_CALL(syscall_name, args) \ |
180 | DOARGS_##args \ |
181 | movl $SYS_ify (syscall_name), %eax; \ |
182 | syscall; |
183 | |
184 | # define DOARGS_0 /* nothing */ |
185 | # define DOARGS_1 /* nothing */ |
186 | # define DOARGS_2 /* nothing */ |
187 | # define DOARGS_3 /* nothing */ |
188 | # define DOARGS_4 movq %rcx, %r10; |
189 | # define DOARGS_5 DOARGS_4 |
190 | # define DOARGS_6 DOARGS_5 |
191 | |
192 | #else /* !__ASSEMBLER__ */ |
193 | /* Define a macro which expands inline into the wrapper code for a system |
194 | call. */ |
195 | # undef INLINE_SYSCALL |
196 | # define INLINE_SYSCALL(name, nr, args...) \ |
197 | ({ \ |
198 | unsigned long int resultvar = INTERNAL_SYSCALL (name, , nr, args); \ |
199 | if (__glibc_unlikely (INTERNAL_SYSCALL_ERROR_P (resultvar, ))) \ |
200 | { \ |
201 | __set_errno (INTERNAL_SYSCALL_ERRNO (resultvar, )); \ |
202 | resultvar = (unsigned long int) -1; \ |
203 | } \ |
204 | (long int) resultvar; }) |
205 | |
206 | /* Define a macro with explicit types for arguments, which expands inline |
207 | into the wrapper code for a system call. It should be used when size |
208 | of any argument > size of long int. */ |
209 | # undef INLINE_SYSCALL_TYPES |
210 | # define INLINE_SYSCALL_TYPES(name, nr, args...) \ |
211 | ({ \ |
212 | unsigned long int resultvar = INTERNAL_SYSCALL_TYPES (name, , nr, args); \ |
213 | if (__glibc_unlikely (INTERNAL_SYSCALL_ERROR_P (resultvar, ))) \ |
214 | { \ |
215 | __set_errno (INTERNAL_SYSCALL_ERRNO (resultvar, )); \ |
216 | resultvar = (unsigned long int) -1; \ |
217 | } \ |
218 | (long int) resultvar; }) |
219 | |
220 | # undef INTERNAL_SYSCALL_DECL |
221 | # define INTERNAL_SYSCALL_DECL(err) do { } while (0) |
222 | |
223 | /* Registers clobbered by syscall. */ |
224 | # define REGISTERS_CLOBBERED_BY_SYSCALL "cc", "r11", "cx" |
225 | |
226 | /* Create a variable 'name' based on type 'X' to avoid explicit types. |
227 | This is mainly used set use 64-bits arguments in x32. */ |
228 | #define TYPEFY(X, name) __typeof__ ((X) - (X)) name |
229 | /* Explicit cast the argument to avoid integer from pointer warning on |
230 | x32. */ |
231 | #define ARGIFY(X) ((__typeof__ ((X) - (X))) (X)) |
232 | |
233 | #undef INTERNAL_SYSCALL |
234 | #define INTERNAL_SYSCALL(name, err, nr, args...) \ |
235 | internal_syscall##nr (SYS_ify (name), err, args) |
236 | |
237 | #undef INTERNAL_SYSCALL_NCS |
238 | #define INTERNAL_SYSCALL_NCS(number, err, nr, args...) \ |
239 | internal_syscall##nr (number, err, args) |
240 | |
241 | #undef internal_syscall0 |
242 | #define internal_syscall0(number, err, dummy...) \ |
243 | ({ \ |
244 | unsigned long int resultvar; \ |
245 | asm volatile ( \ |
246 | "syscall\n\t" \ |
247 | : "=a" (resultvar) \ |
248 | : "0" (number) \ |
249 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
250 | (long int) resultvar; \ |
251 | }) |
252 | |
253 | #undef internal_syscall1 |
254 | #define internal_syscall1(number, err, arg1) \ |
255 | ({ \ |
256 | unsigned long int resultvar; \ |
257 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
258 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
259 | asm volatile ( \ |
260 | "syscall\n\t" \ |
261 | : "=a" (resultvar) \ |
262 | : "0" (number), "r" (_a1) \ |
263 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
264 | (long int) resultvar; \ |
265 | }) |
266 | |
267 | #undef internal_syscall2 |
268 | #define internal_syscall2(number, err, arg1, arg2) \ |
269 | ({ \ |
270 | unsigned long int resultvar; \ |
271 | TYPEFY (arg2, __arg2) = ARGIFY (arg2); \ |
272 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
273 | register TYPEFY (arg2, _a2) asm ("rsi") = __arg2; \ |
274 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
275 | asm volatile ( \ |
276 | "syscall\n\t" \ |
277 | : "=a" (resultvar) \ |
278 | : "0" (number), "r" (_a1), "r" (_a2) \ |
279 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
280 | (long int) resultvar; \ |
281 | }) |
282 | |
283 | #undef internal_syscall3 |
284 | #define internal_syscall3(number, err, arg1, arg2, arg3) \ |
285 | ({ \ |
286 | unsigned long int resultvar; \ |
287 | TYPEFY (arg3, __arg3) = ARGIFY (arg3); \ |
288 | TYPEFY (arg2, __arg2) = ARGIFY (arg2); \ |
289 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
290 | register TYPEFY (arg3, _a3) asm ("rdx") = __arg3; \ |
291 | register TYPEFY (arg2, _a2) asm ("rsi") = __arg2; \ |
292 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
293 | asm volatile ( \ |
294 | "syscall\n\t" \ |
295 | : "=a" (resultvar) \ |
296 | : "0" (number), "r" (_a1), "r" (_a2), "r" (_a3) \ |
297 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
298 | (long int) resultvar; \ |
299 | }) |
300 | |
301 | #undef internal_syscall4 |
302 | #define internal_syscall4(number, err, arg1, arg2, arg3, arg4) \ |
303 | ({ \ |
304 | unsigned long int resultvar; \ |
305 | TYPEFY (arg4, __arg4) = ARGIFY (arg4); \ |
306 | TYPEFY (arg3, __arg3) = ARGIFY (arg3); \ |
307 | TYPEFY (arg2, __arg2) = ARGIFY (arg2); \ |
308 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
309 | register TYPEFY (arg4, _a4) asm ("r10") = __arg4; \ |
310 | register TYPEFY (arg3, _a3) asm ("rdx") = __arg3; \ |
311 | register TYPEFY (arg2, _a2) asm ("rsi") = __arg2; \ |
312 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
313 | asm volatile ( \ |
314 | "syscall\n\t" \ |
315 | : "=a" (resultvar) \ |
316 | : "0" (number), "r" (_a1), "r" (_a2), "r" (_a3), "r" (_a4) \ |
317 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
318 | (long int) resultvar; \ |
319 | }) |
320 | |
321 | #undef internal_syscall5 |
322 | #define internal_syscall5(number, err, arg1, arg2, arg3, arg4, arg5) \ |
323 | ({ \ |
324 | unsigned long int resultvar; \ |
325 | TYPEFY (arg5, __arg5) = ARGIFY (arg5); \ |
326 | TYPEFY (arg4, __arg4) = ARGIFY (arg4); \ |
327 | TYPEFY (arg3, __arg3) = ARGIFY (arg3); \ |
328 | TYPEFY (arg2, __arg2) = ARGIFY (arg2); \ |
329 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
330 | register TYPEFY (arg5, _a5) asm ("r8") = __arg5; \ |
331 | register TYPEFY (arg4, _a4) asm ("r10") = __arg4; \ |
332 | register TYPEFY (arg3, _a3) asm ("rdx") = __arg3; \ |
333 | register TYPEFY (arg2, _a2) asm ("rsi") = __arg2; \ |
334 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
335 | asm volatile ( \ |
336 | "syscall\n\t" \ |
337 | : "=a" (resultvar) \ |
338 | : "0" (number), "r" (_a1), "r" (_a2), "r" (_a3), "r" (_a4), \ |
339 | "r" (_a5) \ |
340 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
341 | (long int) resultvar; \ |
342 | }) |
343 | |
344 | #undef internal_syscall6 |
345 | #define internal_syscall6(number, err, arg1, arg2, arg3, arg4, arg5, arg6) \ |
346 | ({ \ |
347 | unsigned long int resultvar; \ |
348 | TYPEFY (arg6, __arg6) = ARGIFY (arg6); \ |
349 | TYPEFY (arg5, __arg5) = ARGIFY (arg5); \ |
350 | TYPEFY (arg4, __arg4) = ARGIFY (arg4); \ |
351 | TYPEFY (arg3, __arg3) = ARGIFY (arg3); \ |
352 | TYPEFY (arg2, __arg2) = ARGIFY (arg2); \ |
353 | TYPEFY (arg1, __arg1) = ARGIFY (arg1); \ |
354 | register TYPEFY (arg6, _a6) asm ("r9") = __arg6; \ |
355 | register TYPEFY (arg5, _a5) asm ("r8") = __arg5; \ |
356 | register TYPEFY (arg4, _a4) asm ("r10") = __arg4; \ |
357 | register TYPEFY (arg3, _a3) asm ("rdx") = __arg3; \ |
358 | register TYPEFY (arg2, _a2) asm ("rsi") = __arg2; \ |
359 | register TYPEFY (arg1, _a1) asm ("rdi") = __arg1; \ |
360 | asm volatile ( \ |
361 | "syscall\n\t" \ |
362 | : "=a" (resultvar) \ |
363 | : "0" (number), "r" (_a1), "r" (_a2), "r" (_a3), "r" (_a4), \ |
364 | "r" (_a5), "r" (_a6) \ |
365 | : "memory", REGISTERS_CLOBBERED_BY_SYSCALL); \ |
366 | (long int) resultvar; \ |
367 | }) |
368 | |
369 | # undef INTERNAL_SYSCALL_ERROR_P |
370 | # define INTERNAL_SYSCALL_ERROR_P(val, err) \ |
371 | ((unsigned long int) (long int) (val) >= -4095L) |
372 | |
373 | # undef INTERNAL_SYSCALL_ERRNO |
374 | # define INTERNAL_SYSCALL_ERRNO(val, err) (-(val)) |
375 | |
376 | /* List of system calls which are supported as vsyscalls. */ |
377 | # define HAVE_CLOCK_GETTIME_VSYSCALL 1 |
378 | # define HAVE_GETTIMEOFDAY_VSYSCALL 1 |
379 | # define HAVE_GETCPU_VSYSCALL 1 |
380 | |
381 | # define SINGLE_THREAD_BY_GLOBAL 1 |
382 | |
383 | #endif /* __ASSEMBLER__ */ |
384 | |
385 | |
386 | /* Pointer mangling support. */ |
387 | #if IS_IN (rtld) |
388 | /* We cannot use the thread descriptor because in ld.so we use setjmp |
389 | earlier than the descriptor is initialized. */ |
390 | # ifdef __ASSEMBLER__ |
391 | # define PTR_MANGLE(reg) xor __pointer_chk_guard_local(%rip), reg; \ |
392 | rol $2*LP_SIZE+1, reg |
393 | # define PTR_DEMANGLE(reg) ror $2*LP_SIZE+1, reg; \ |
394 | xor __pointer_chk_guard_local(%rip), reg |
395 | # else |
396 | # define PTR_MANGLE(reg) asm ("xor __pointer_chk_guard_local(%%rip), %0\n" \ |
397 | "rol $2*" LP_SIZE "+1, %0" \ |
398 | : "=r" (reg) : "0" (reg)) |
399 | # define PTR_DEMANGLE(reg) asm ("ror $2*" LP_SIZE "+1, %0\n" \ |
400 | "xor __pointer_chk_guard_local(%%rip), %0" \ |
401 | : "=r" (reg) : "0" (reg)) |
402 | # endif |
403 | #else |
404 | # ifdef __ASSEMBLER__ |
405 | # define PTR_MANGLE(reg) xor %fs:POINTER_GUARD, reg; \ |
406 | rol $2*LP_SIZE+1, reg |
407 | # define PTR_DEMANGLE(reg) ror $2*LP_SIZE+1, reg; \ |
408 | xor %fs:POINTER_GUARD, reg |
409 | # else |
410 | # define PTR_MANGLE(var) asm ("xor %%fs:%c2, %0\n" \ |
411 | "rol $2*" LP_SIZE "+1, %0" \ |
412 | : "=r" (var) \ |
413 | : "0" (var), \ |
414 | "i" (offsetof (tcbhead_t, \ |
415 | pointer_guard))) |
416 | # define PTR_DEMANGLE(var) asm ("ror $2*" LP_SIZE "+1, %0\n" \ |
417 | "xor %%fs:%c2, %0" \ |
418 | : "=r" (var) \ |
419 | : "0" (var), \ |
420 | "i" (offsetof (tcbhead_t, \ |
421 | pointer_guard))) |
422 | # endif |
423 | #endif |
424 | |
425 | /* How to pass the off{64}_t argument on p{readv,writev}{64}. */ |
426 | #undef LO_HI_LONG |
427 | #define LO_HI_LONG(val) (val), 0 |
428 | |
429 | /* Each shadow stack slot takes 8 bytes. Assuming that each stack |
430 | frame takes 256 bytes, this is used to compute shadow stack size |
431 | from stack size. */ |
432 | #define STACK_SIZE_TO_SHADOW_STACK_SIZE_SHIFT 5 |
433 | |
434 | #endif /* linux/x86_64/sysdep.h */ |
435 | |