]> git.ipfire.org Git - thirdparty/man-pages.git/blob - man7/vdso.7
pldd.1, bpf.2, chdir.2, clone.2, fanotify_init.2, fanotify_mark.2, intro.2, ipc.2...
[thirdparty/man-pages.git] / man7 / vdso.7
1 .\" Written by Mike Frysinger <vapier@gentoo.org>
2 .\"
3 .\" %%%LICENSE_START(PUBLIC_DOMAIN)
4 .\" This page is in the public domain.
5 .\" %%%LICENSE_END
6 .\"
7 .\" Useful background:
8 .\" http://articles.manugarg.com/systemcallinlinux2_6.html
9 .\" https://lwn.net/Articles/446528/
10 .\" http://www.linuxjournal.com/content/creating-vdso-colonels-other-chicken
11 .\" http://www.trilithium.com/johan/2005/08/linux-gate/
12 .\"
13 .TH VDSO 7 2019-08-02 "Linux" "Linux Programmer's Manual"
14 .SH NAME
15 vdso \- overview of the virtual ELF dynamic shared object
16 .SH SYNOPSIS
17 .B #include <sys/auxv.h>
18 .PP
19 .B void *vdso = (uintptr_t) getauxval(AT_SYSINFO_EHDR);
20 .SH DESCRIPTION
21 The "vDSO" (virtual dynamic shared object) is a small shared library that
22 the kernel automatically maps into the
23 address space of all user-space applications.
24 Applications usually do not need to concern themselves with these details
25 as the vDSO is most commonly called by the C library.
26 This way you can code in the normal way using standard functions
27 and the C library will take care
28 of using any functionality that is available via the vDSO.
29 .PP
30 Why does the vDSO exist at all?
31 There are some system calls the kernel provides that
32 user-space code ends up using frequently,
33 to the point that such calls can dominate overall performance.
34 This is due both to the frequency of the call as well as the
35 context-switch overhead that results
36 from exiting user space and entering the kernel.
37 .PP
38 The rest of this documentation is geared toward the curious and/or
39 C library writers rather than general developers.
40 If you're trying to call the vDSO in your own application rather than using
41 the C library, you're most likely doing it wrong.
42 .SS Example background
43 Making system calls can be slow.
44 In x86 32-bit systems, you can trigger a software interrupt
45 .RI ( "int $0x80" )
46 to tell the kernel you wish to make a system call.
47 However, this instruction is expensive: it goes through
48 the full interrupt-handling paths
49 in the processor's microcode as well as in the kernel.
50 Newer processors have faster (but backward incompatible) instructions to
51 initiate system calls.
52 Rather than require the C library to figure out if this functionality is
53 available at run time,
54 the C library can use functions provided by the kernel in
55 the vDSO.
56 .PP
57 Note that the terminology can be confusing.
58 On x86 systems, the vDSO function
59 used to determine the preferred method of making a system call is
60 named "__kernel_vsyscall", but on x86-64,
61 the term "vsyscall" also refers to an obsolete way to ask the kernel
62 what time it is or what CPU the caller is on.
63 .PP
64 One frequently used system call is
65 .BR gettimeofday (2).
66 This system call is called both directly by user-space applications
67 as well as indirectly by
68 the C library.
69 Think timestamps or timing loops or polling\(emall of these
70 frequently need to know what time it is right now.
71 This information is also not secret\(emany application in any
72 privilege mode (root or any unprivileged user) will get the same answer.
73 Thus the kernel arranges for the information required to answer
74 this question to be placed in memory the process can access.
75 Now a call to
76 .BR gettimeofday (2)
77 changes from a system call to a normal function
78 call and a few memory accesses.
79 .SS Finding the vDSO
80 The base address of the vDSO (if one exists) is passed by the kernel to
81 each program in the initial auxiliary vector (see
82 .BR getauxval (3)),
83 via the
84 .B AT_SYSINFO_EHDR
85 tag.
86 .PP
87 You must not assume the vDSO is mapped at any particular location in the
88 user's memory map.
89 The base address will usually be randomized at run time every time a new
90 process image is created (at
91 .BR execve (2)
92 time).
93 This is done for security reasons,
94 to prevent "return-to-libc" attacks.
95 .PP
96 For some architectures, there is also an
97 .B AT_SYSINFO
98 tag.
99 This is used only for locating the vsyscall entry point and is frequently
100 omitted or set to 0 (meaning it's not available).
101 This tag is a throwback to the initial vDSO work (see
102 .IR History
103 below) and its use should be avoided.
104 .SS File format
105 Since the vDSO is a fully formed ELF image, you can do symbol lookups on it.
106 This allows new symbols to be added with newer kernel releases,
107 and allows the C library to detect available functionality at
108 run time when running under different kernel versions.
109 Oftentimes the C library will do detection with the first call and then
110 cache the result for subsequent calls.
111 .PP
112 All symbols are also versioned (using the GNU version format).
113 This allows the kernel to update the function signature without breaking
114 backward compatibility.
115 This means changing the arguments that the function accepts as well as the
116 return value.
117 Thus, when looking up a symbol in the vDSO,
118 you must always include the version
119 to match the ABI you expect.
120 .PP
121 Typically the vDSO follows the naming convention of prefixing
122 all symbols with "__vdso_" or "__kernel_"
123 so as to distinguish them from other standard symbols.
124 For example, the "gettimeofday" function is named "__vdso_gettimeofday".
125 .PP
126 You use the standard C calling conventions when calling
127 any of these functions.
128 No need to worry about weird register or stack behavior.
129 .SH NOTES
130 .SS Source
131 When you compile the kernel,
132 it will automatically compile and link the vDSO code for you.
133 You will frequently find it under the architecture-specific directory:
134 .PP
135 find arch/$ARCH/ \-name \(aq*vdso*.so*\(aq \-o \-name \(aq*gate*.so*\(aq
136 .\"
137 .SS vDSO names
138 The name of the vDSO varies across architectures.
139 It will often show up in things like glibc's
140 .BR ldd (1)
141 output.
142 The exact name should not matter to any code, so do not hardcode it.
143 .if t \{\
144 .ft CW
145 \}
146 .TS
147 l l.
148 user ABI vDSO name
149 _
150 aarch64 linux\-vdso.so.1
151 arm linux\-vdso.so.1
152 ia64 linux\-gate.so.1
153 mips linux\-vdso.so.1
154 ppc/32 linux\-vdso32.so.1
155 ppc/64 linux\-vdso64.so.1
156 riscv linux\-vdso.so.1
157 s390 linux\-vdso32.so.1
158 s390x linux\-vdso64.so.1
159 sh linux\-gate.so.1
160 i386 linux\-gate.so.1
161 x86-64 linux\-vdso.so.1
162 x86/x32 linux\-vdso.so.1
163 .TE
164 .if t \{\
165 .in
166 .ft P
167 \}
168 .SS strace(1), seccomp(2), and the vDSO
169 When tracing systems calls with
170 .BR strace (1),
171 symbols (system calls) that are exported by the vDSO will
172 .I not
173 appear in the trace output.
174 Those system calls will likewise not be visible to
175 .BR seccomp (2)
176 filters.
177 .SH ARCHITECTURE-SPECIFIC NOTES
178 The subsections below provide architecture-specific notes
179 on the vDSO.
180 .PP
181 Note that the vDSO that is used is based on the ABI of your user-space code
182 and not the ABI of the kernel.
183 Thus, for example,
184 when you run an i386 32-bit ELF binary,
185 you'll get the same vDSO regardless of whether you run it under
186 an i386 32-bit kernel or under an x86-64 64-bit kernel.
187 Therefore, the name of the user-space ABI should be used to determine
188 which of the sections below is relevant.
189 .SS ARM functions
190 .\" See linux/arch/arm/vdso/vdso.lds.S
191 .\" Commit: 8512287a8165592466cb9cb347ba94892e9c56a5
192 The table below lists the symbols exported by the vDSO.
193 .if t \{\
194 .ft CW
195 \}
196 .TS
197 l l.
198 symbol version
199 _
200 __vdso_gettimeofday LINUX_2.6 (exported since Linux 4.1)
201 __vdso_clock_gettime LINUX_2.6 (exported since Linux 4.1)
202 .TE
203 .if t \{\
204 .in
205 .ft P
206 \}
207 .PP
208 .\" See linux/arch/arm/kernel/entry-armv.S
209 .\" See linux/Documentation/arm/kernel_user_helpers.txt
210 Additionally, the ARM port has a code page full of utility functions.
211 Since it's just a raw page of code, there is no ELF information for doing
212 symbol lookups or versioning.
213 It does provide support for different versions though.
214 .PP
215 For information on this code page,
216 it's best to refer to the kernel documentation
217 as it's extremely detailed and covers everything you need to know:
218 .IR Documentation/arm/kernel_user_helpers.txt .
219 .SS aarch64 functions
220 .\" See linux/arch/arm64/kernel/vdso/vdso.lds.S
221 The table below lists the symbols exported by the vDSO.
222 .if t \{\
223 .ft CW
224 \}
225 .TS
226 l l.
227 symbol version
228 _
229 __kernel_rt_sigreturn LINUX_2.6.39
230 __kernel_gettimeofday LINUX_2.6.39
231 __kernel_clock_gettime LINUX_2.6.39
232 __kernel_clock_getres LINUX_2.6.39
233 .TE
234 .if t \{\
235 .in
236 .ft P
237 \}
238 .SS bfin (Blackfin) functions (port removed in Linux 4.17)
239 .\" See linux/arch/blackfin/kernel/fixed_code.S
240 .\" See http://docs.blackfin.uclinux.org/doku.php?id=linux-kernel:fixed-code
241 As this CPU lacks a memory management unit (MMU),
242 it doesn't set up a vDSO in the normal sense.
243 Instead, it maps at boot time a few raw functions into
244 a fixed location in memory.
245 User-space applications then call directly into that region.
246 There is no provision for backward compatibility
247 beyond sniffing raw opcodes,
248 but as this is an embedded CPU, it can get away with things\(emsome of the
249 object formats it runs aren't even ELF based (they're bFLT/FLAT).
250 .PP
251 For information on this code page,
252 it's best to refer to the public documentation:
253 .br
254 http://docs.blackfin.uclinux.org/doku.php?id=linux\-kernel:fixed\-code
255 .SS mips functions
256 .\" See linux/arch/mips/vdso/vdso.ld.S
257 .PP
258 The table below lists the symbols exported by the vDSO.
259 .if t \{\
260 .ft CW
261 \}
262 .TS
263 l l.
264 symbol version
265 _
266 __kernel_gettimeofday LINUX_2.6 (exported since Linux 4.4)
267 __kernel_clock_gettime LINUX_2.6 (exported since Linux 4.4)
268 .TE
269 .if t \{\
270 .in
271 .ft P
272 \}
273 .SS ia64 (Itanium) functions
274 .\" See linux/arch/ia64/kernel/gate.lds.S
275 .\" Also linux/arch/ia64/kernel/fsys.S and linux/Documentation/ia64/fsys.txt
276 The table below lists the symbols exported by the vDSO.
277 .if t \{\
278 .ft CW
279 \}
280 .TS
281 l l.
282 symbol version
283 _
284 __kernel_sigtramp LINUX_2.5
285 __kernel_syscall_via_break LINUX_2.5
286 __kernel_syscall_via_epc LINUX_2.5
287 .TE
288 .if t \{\
289 .in
290 .ft P
291 \}
292 .PP
293 The Itanium port is somewhat tricky.
294 In addition to the vDSO above, it also has "light-weight system calls"
295 (also known as "fast syscalls" or "fsys").
296 You can invoke these via the
297 .I __kernel_syscall_via_epc
298 vDSO helper.
299 The system calls listed here have the same semantics as if you called them
300 directly via
301 .BR syscall (2),
302 so refer to the relevant
303 documentation for each.
304 The table below lists the functions available via this mechanism.
305 .if t \{\
306 .ft CW
307 \}
308 .TS
309 l.
310 function
311 _
312 clock_gettime
313 getcpu
314 getpid
315 getppid
316 gettimeofday
317 set_tid_address
318 .TE
319 .if t \{\
320 .in
321 .ft P
322 \}
323 .SS parisc (hppa) functions
324 .\" See linux/arch/parisc/kernel/syscall.S
325 .\" See linux/Documentation/parisc/registers
326 The parisc port has a code page with utility functions
327 called a gateway page.
328 Rather than use the normal ELF auxiliary vector approach,
329 it passes the address of
330 the page to the process via the SR2 register.
331 The permissions on the page are such that merely executing those addresses
332 automatically executes with kernel privileges and not in user space.
333 This is done to match the way HP-UX works.
334 .PP
335 Since it's just a raw page of code, there is no ELF information for doing
336 symbol lookups or versioning.
337 Simply call into the appropriate offset via the branch instruction,
338 for example:
339 .PP
340 ble <offset>(%sr2, %r0)
341 .if t \{\
342 .ft CW
343 \}
344 .TS
345 l l.
346 offset function
347 _
348 00b0 lws_entry (CAS operations)
349 00e0 set_thread_pointer (used by glibc)
350 0100 linux_gateway_entry (syscall)
351 .TE
352 .if t \{\
353 .in
354 .ft P
355 \}
356 .SS ppc/32 functions
357 .\" See linux/arch/powerpc/kernel/vdso32/vdso32.lds.S
358 The table below lists the symbols exported by the vDSO.
359 The functions marked with a
360 .I *
361 are available only when the kernel is
362 a PowerPC64 (64-bit) kernel.
363 .if t \{\
364 .ft CW
365 \}
366 .TS
367 l l.
368 symbol version
369 _
370 __kernel_clock_getres LINUX_2.6.15
371 __kernel_clock_gettime LINUX_2.6.15
372 __kernel_datapage_offset LINUX_2.6.15
373 __kernel_get_syscall_map LINUX_2.6.15
374 __kernel_get_tbfreq LINUX_2.6.15
375 __kernel_getcpu \fI*\fR LINUX_2.6.15
376 __kernel_gettimeofday LINUX_2.6.15
377 __kernel_sigtramp_rt32 LINUX_2.6.15
378 __kernel_sigtramp32 LINUX_2.6.15
379 __kernel_sync_dicache LINUX_2.6.15
380 __kernel_sync_dicache_p5 LINUX_2.6.15
381 .TE
382 .if t \{\
383 .in
384 .ft P
385 \}
386 .PP
387 The
388 .B CLOCK_REALTIME_COARSE
389 and
390 .B CLOCK_MONOTONIC_COARSE
391 clocks are
392 .I not
393 supported by the
394 .I __kernel_clock_getres
395 and
396 .I __kernel_clock_gettime
397 interfaces;
398 the kernel falls back to the real system call.
399 .SS ppc/64 functions
400 .\" See linux/arch/powerpc/kernel/vdso64/vdso64.lds.S
401 The table below lists the symbols exported by the vDSO.
402 .if t \{\
403 .ft CW
404 \}
405 .TS
406 l l.
407 symbol version
408 _
409 __kernel_clock_getres LINUX_2.6.15
410 __kernel_clock_gettime LINUX_2.6.15
411 __kernel_datapage_offset LINUX_2.6.15
412 __kernel_get_syscall_map LINUX_2.6.15
413 __kernel_get_tbfreq LINUX_2.6.15
414 __kernel_getcpu LINUX_2.6.15
415 __kernel_gettimeofday LINUX_2.6.15
416 __kernel_sigtramp_rt64 LINUX_2.6.15
417 __kernel_sync_dicache LINUX_2.6.15
418 __kernel_sync_dicache_p5 LINUX_2.6.15
419 .TE
420 .if t \{\
421 .in
422 .ft P
423 \}
424 .PP
425 The
426 .B CLOCK_REALTIME_COARSE
427 and
428 .B CLOCK_MONOTONIC_COARSE
429 clocks are
430 .I not
431 supported by the
432 .I __kernel_clock_getres
433 and
434 .I __kernel_clock_gettime
435 interfaces;
436 the kernel falls back to the real system call.
437 .SS riscv functions
438 .\" See linux/arch/riscv/kernel/vdso/vdso.lds.S
439 The table below lists the symbols exported by the vDSO.
440 .if t \{\
441 .ft CW
442 \}
443 .TS
444 l l.
445 symbol version
446 _
447 __kernel_rt_sigreturn LINUX_4.15
448 __kernel_gettimeofday LINUX_4.15
449 __kernel_clock_gettime LINUX_4.15
450 __kernel_clock_getres LINUX_4.15
451 __kernel_getcpu LINUX_4.15
452 __kernel_flush_icache LINUX_4.15
453 .TE
454 .if t \{\
455 .in
456 .ft P
457 \}
458 .SS s390 functions
459 .\" See linux/arch/s390/kernel/vdso32/vdso32.lds.S
460 The table below lists the symbols exported by the vDSO.
461 .if t \{\
462 .ft CW
463 \}
464 .TS
465 l l.
466 symbol version
467 _
468 __kernel_clock_getres LINUX_2.6.29
469 __kernel_clock_gettime LINUX_2.6.29
470 __kernel_gettimeofday LINUX_2.6.29
471 .TE
472 .if t \{\
473 .in
474 .ft P
475 \}
476 .SS s390x functions
477 .\" See linux/arch/s390/kernel/vdso64/vdso64.lds.S
478 The table below lists the symbols exported by the vDSO.
479 .if t \{\
480 .ft CW
481 \}
482 .TS
483 l l.
484 symbol version
485 _
486 __kernel_clock_getres LINUX_2.6.29
487 __kernel_clock_gettime LINUX_2.6.29
488 __kernel_gettimeofday LINUX_2.6.29
489 .TE
490 .if t \{\
491 .in
492 .ft P
493 \}
494 .SS sh (SuperH) functions
495 .\" See linux/arch/sh/kernel/vsyscall/vsyscall.lds.S
496 The table below lists the symbols exported by the vDSO.
497 .if t \{\
498 .ft CW
499 \}
500 .TS
501 l l.
502 symbol version
503 _
504 __kernel_rt_sigreturn LINUX_2.6
505 __kernel_sigreturn LINUX_2.6
506 __kernel_vsyscall LINUX_2.6
507 .TE
508 .if t \{\
509 .in
510 .ft P
511 \}
512 .SS i386 functions
513 .\" See linux/arch/x86/vdso/vdso32/vdso32.lds.S
514 The table below lists the symbols exported by the vDSO.
515 .if t \{\
516 .ft CW
517 \}
518 .TS
519 l l.
520 symbol version
521 _
522 __kernel_sigreturn LINUX_2.5
523 __kernel_rt_sigreturn LINUX_2.5
524 __kernel_vsyscall LINUX_2.5
525 .\" Added in 7a59ed415f5b57469e22e41fc4188d5399e0b194 and updated
526 .\" in 37c975545ec63320789962bf307f000f08fabd48.
527 __vdso_clock_gettime LINUX_2.6 (exported since Linux 3.15)
528 __vdso_gettimeofday LINUX_2.6 (exported since Linux 3.15)
529 __vdso_time LINUX_2.6 (exported since Linux 3.15)
530 .TE
531 .if t \{\
532 .in
533 .ft P
534 \}
535 .SS x86-64 functions
536 .\" See linux/arch/x86/vdso/vdso.lds.S
537 The table below lists the symbols exported by the vDSO.
538 All of these symbols are also available without the "__vdso_" prefix, but
539 you should ignore those and stick to the names below.
540 .if t \{\
541 .ft CW
542 \}
543 .TS
544 l l.
545 symbol version
546 _
547 __vdso_clock_gettime LINUX_2.6
548 __vdso_getcpu LINUX_2.6
549 __vdso_gettimeofday LINUX_2.6
550 __vdso_time LINUX_2.6
551 .TE
552 .if t \{\
553 .in
554 .ft P
555 \}
556 .SS x86/x32 functions
557 .\" See linux/arch/x86/vdso/vdso32.lds.S
558 The table below lists the symbols exported by the vDSO.
559 .if t \{\
560 .ft CW
561 \}
562 .TS
563 l l.
564 symbol version
565 _
566 __vdso_clock_gettime LINUX_2.6
567 __vdso_getcpu LINUX_2.6
568 __vdso_gettimeofday LINUX_2.6
569 __vdso_time LINUX_2.6
570 .TE
571 .if t \{\
572 .in
573 .ft P
574 \}
575 .SS History
576 The vDSO was originally just a single function\(emthe vsyscall.
577 In older kernels, you might see that name
578 in a process's memory map rather than "vdso".
579 Over time, people realized that this mechanism
580 was a great way to pass more functionality
581 to user space, so it was reconceived as a vDSO in the current format.
582 .SH SEE ALSO
583 .BR syscalls (2),
584 .BR getauxval (3),
585 .BR proc (5)
586 .PP
587 The documents, examples, and source code in the Linux source code tree:
588 .PP
589 .in +4n
590 .EX
591 Documentation/ABI/stable/vdso
592 Documentation/ia64/fsys.txt
593 Documentation/vDSO/* (includes examples of using the vDSO)
594
595 find arch/ \-iname \(aq*vdso*\(aq \-o \-iname \(aq*gate*\(aq
596 .EE
597 .in