1 .\" Copyright (c) 2006, 2008 by Michael Kerrisk <mtk.manpages@gmail.com>
3 .\" %%%LICENSE_START(VERBATIM)
4 .\" Permission is granted to make and distribute verbatim copies of this
5 .\" manual provided the copyright notice and this permission notice are
6 .\" preserved on all copies.
8 .\" Permission is granted to copy and distribute modified versions of this
9 .\" manual under the conditions for verbatim copying, provided that the
10 .\" entire resulting derived work is distributed under the terms of a
11 .\" permission notice identical to this one.
13 .\" Since the Linux kernel and libraries are constantly changing, this
14 .\" manual page may be incorrect or out-of-date. The author(s) assume no
15 .\" responsibility for errors or omissions, or for damages resulting from
16 .\" the use of the information contained herein. The author(s) may not
17 .\" have taken the same level of care in the production of this manual,
18 .\" which is licensed free of charge, as they might when working
21 .\" Formatted or processed versions of this manual, if unaccompanied by
22 .\" the source, must acknowledge the copyright and authors of this work.
25 .TH CORE 5 2017-09-15 "Linux" "Linux Programmer's Manual"
27 core \- core dump file
29 The default action of certain signals is to cause a process to terminate
31 .IR "core dump file" ,
32 a disk file containing an image of the process's memory at
33 the time of termination.
34 This image can be used in a debugger (e.g.,
36 to inspect the state of the program at the time that it terminated.
37 A list of the signals which cause a process to dump core can be found in
40 A process can set its soft
42 resource limit to place an upper limit on the size of the core dump file
43 that will be produced if it receives a "core dump" signal; see
47 There are various circumstances in which a core dump file is
50 The process does not have permission to write the core file.
51 (By default, the core file is called
57 is the ID of the process that dumped core,
58 and is created in the current working directory.
59 See below for details on naming.)
60 Writing the core file fails if the directory in which
61 it is to be created is nonwritable,
62 or if a file with the same name exists and
64 or is not a regular file
65 (e.g., it is a directory or a symbolic link).
67 A (writable, regular) file with the same name as would be used for the
68 core dump already exists, but there is more than one hard link to that
71 The filesystem where the core dump file would be created is full;
72 or has run out of inodes; or is mounted read-only;
73 or the user has reached their quota for the filesystem.
75 The directory in which the core dump file is to be created does
82 (file size) resource limits for the process are set to zero; see
84 and the documentation of the shell's
91 The binary being executed by the process does not have read
94 The process is executing a set-user-ID (set-group-ID) program
95 that is owned by a user (group) other than the real user (group)
97 or the process is executing a program that has file capabilities (see
98 .BR capabilities (7)).
99 (However, see the description of the
102 operation, and the description of the
103 .I /proc/sys/fs/suid_dumpable
104 .\" FIXME . Perhaps relocate discussion of /proc/sys/fs/suid_dumpable
105 .\" and PR_SET_DUMPABLE to this page?
109 .I /proc/sys/kernel/core_pattern
111 .I /proc/sys/kernel/core_uses_pid
112 contains the value 0.
113 (These files are described below.)
115 .I /proc/sys/kernel/core_pattern
117 .I /proc/sys/kernel/core_uses_pid
118 contains the value 1,
119 core dump files will have names of the form
121 and such files are hidden unless one uses the
127 .\" commit 046d662f481830e652ac34cd112249adde16452a
128 The kernel was configured without the
133 a core dump may exclude part of the address space of the process if the
138 On systems that employ
142 framework, core dumps may instead be placed in a location determined by
144 See below for further details.
146 .SS Naming of core dump files
147 By default, a core dump file is named
150 .I /proc/sys/kernel/core_pattern
151 file (since Linux 2.6 and 2.4.21)
152 can be set to define a template that is used to name core dump files.
153 The template can contain % specifiers which are substituted
154 by the following values when a core file is created:
163 core file size soft resource limit of crashing process (since Linux 2.6.24)
166 .\" Added in git commit 12a2b4b2241e318b4f6df31228e4272d2c2968a1
167 dump mode\(emsame as value returned by
173 executable filename (without path prefix)
176 pathname of executable,
177 with slashes (\(aq/\(aq) replaced by exclamation marks (\(aq!\(aq)
181 (numeric) real GID of dumped process
184 hostname (same as \fInodename\fP returned by \fBuname\fP(2))
187 TID of thread that triggered core dump,
188 as seen in the PID namespace in which the thread resides
189 .\" commit b03023ecbdb76c1dec86b41ed80b123c22783220
193 TID of thread that triggered core dump, as seen in the initial PID namespace
194 .\" commit b03023ecbdb76c1dec86b41ed80b123c22783220
198 PID of dumped process,
199 as seen in the PID namespace in which the process resides
202 .\" Added in git commit 65aafb1e7484b7434a0c1d4c593191ebe5776a2f
203 PID of dumped process, as seen in the initial PID namespace
207 number of signal causing dump
210 time of dump, expressed as seconds since the
211 Epoch, 1970-01-01 00:00:00 +0000 (UTC)
214 (numeric) real UID of dumped process
218 A single % at the end of the template is dropped from the
219 core filename, as is the combination of a % followed by any
220 character other than those listed above.
221 All other characters in the template become a literal
222 part of the core filename.
223 The template may include \(aq/\(aq characters, which are interpreted
224 as delimiters for directory names.
225 The maximum size of the resulting core filename is 128 bytes (64 bytes
226 in kernels before 2.6.19).
227 The default value in this file is "core".
228 For backward compatibility, if
229 .I /proc/sys/kernel/core_pattern
233 .I /proc/sys/kernel/core_uses_pid
235 is nonzero, then .PID will be appended to the core filename.
237 Paths are interpreted according to the settings that are active for the
239 That means the crashing process's mount namespace (see
240 .BR mount_namespaces (7)),
241 its current working directory (found via
243 and its root directory (see
246 Since version 2.4, Linux has also provided
247 a more primitive method of controlling
248 the name of the core dump file.
250 .I /proc/sys/kernel/core_uses_pid
251 file contains the value 0, then a core dump file is simply named
253 If this file contains a nonzero value, then the core dump file includes
254 the process ID in a name of the form
258 .\" 9520628e8ceb69fa9a4aee6b57f22675d9e1b709
260 .I /proc/sys/fs/suid_dumpable
261 is set to 2 ("suidsafe"), the pattern must be either an absolute pathname
262 (starting with a leading \(aq/\(aq character) or a pipe, as defined below.
263 .SS Piping core dumps to a program
264 Since kernel 2.6.19, Linux supports an alternate syntax for the
265 .I /proc/sys/kernel/core_pattern
267 If the first character of this file is a pipe symbol (\fB|\fP),
268 then the remainder of the line is interpreted as the command-line for
269 a user-space program (or script) that is to be executed.
270 Instead of being written to a disk file, the core dump is given as
271 standard input to the program.
272 Note the following points:
274 The program must be specified using an absolute pathname (or a
275 pathname relative to the root directory, \fI/\fP),
276 and must immediately follow the '|' character.
278 The command-line arguments can include any of
279 the % specifiers listed above.
280 For example, to pass the PID of the process that is being dumped, specify
284 The process created to run the program runs as user and group
289 does not confer any exceptional security bypasses.
290 Namely, LSMs (e.g., SELinux) are still active and may prevent the handler
291 from accessing details about the crashed process via
294 The program pathname is interpreted with respect to the initial mount namespace
295 as it is always executed there.
296 It is not affected by the settings
297 (e.g., root directory, mount namespace, current working directory)
298 of the crashing process.
300 The process runs in the initial namespaces
301 (PID, mount, user, and so on)
302 and not in the namespaces of the crashing process.
303 One can utilize specifiers such as
307 directory and probe/enter the crashing process's namespaces if needed.
309 The process starts with its current working directory
310 as the root directory.
311 If desired, it is possible change to the working directory of
312 the dumping process by employing the value provided by the
314 specifier to change to the location of the dumping process via
315 .IR /proc/[pid]/cwd .
317 Command-line arguments can be supplied to the
318 program (since Linux 2.6.24),
319 delimited by white space (up to a total line length of 128 bytes).
323 limit is not enforced for core dumps that are piped to a program
326 .SS /proc/sys/kernel/core_pipe_limit
327 When collecting core dumps via a pipe to a user-space program,
328 it can be useful for the collecting program to gather data about
329 the crashing process from that process's
332 In order to do this safely,
333 the kernel must wait for the program collecting the core dump to exit,
334 so as not to remove the crashing process's
337 This in turn creates the
338 possibility that a misbehaving collecting program can block
339 the reaping of a crashed process by simply never exiting.
342 .\" commit a293980c2e261bd5b0d2a77340dd04f684caff58
344 .I /proc/sys/kernel/core_pipe_limit
345 can be used to defend against this possibility.
346 The value in this file defines how many concurrent crashing
347 processes may be piped to user-space programs in parallel.
348 If this value is exceeded, then those crashing processes above this value
349 are noted in the kernel log and their core dumps are skipped.
351 A value of 0 in this file is special.
352 It indicates that unlimited processes may be captured in parallel,
353 but that no waiting will take place (i.e., the collecting
354 program is not guaranteed access to
355 .IR /proc/<crashing-PID> ).
356 The default value for this file is 0.
358 .SS Controlling which mappings are written to the core dump
359 Since kernel 2.6.23, the Linux-specific
360 .IR /proc/[pid]/coredump_filter
361 file can be used to control which memory segments are written to the
362 core dump file in the event that a core dump is performed for the
363 process with the corresponding process ID.
365 The value in the file is a bit mask of memory mapping types (see
367 If a bit is set in the mask, then memory mappings of the
368 corresponding type are dumped; otherwise they are not dumped.
369 The bits in this file have the following meanings:
375 Dump anonymous private mappings.
378 Dump anonymous shared mappings.
381 Dump file-backed private mappings.
384 Dump file-backed shared mappings.
385 .\" file-backed shared mappings of course also update the underlying
388 bit 4 (since Linux 2.6.24)
391 bit 5 (since Linux 2.6.28)
392 Dump private huge pages.
394 bit 6 (since Linux 2.6.28)
395 Dump shared huge pages.
397 bit 7 (since Linux 4.4)
398 .\" commit ab27a8d04b32b6ee8c30c14c4afd1058e8addc82
399 Dump private DAX pages.
401 bit 8 (since Linux 4.4)
402 .\" commit ab27a8d04b32b6ee8c30c14c4afd1058e8addc82
403 Dump shared DAX pages.
407 By default, the following bits are set: 0, 1, 4 (if the
408 .B CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
409 kernel configuration option is enabled), and 5.
410 This default can be modified at boot time using the
414 The value of this file is displayed in hexadecimal.
415 (The default value is thus displayed as 33.)
417 Memory-mapped I/O pages such as frame buffer are never dumped, and
420 pages are always dumped, regardless of the
424 A child process created via
426 inherits its parent's
431 value is preserved across an
434 It can be useful to set
436 in the parent shell before running a program, for example:
440 .RB "$" " echo 0x7 > /proc/self/coredump_filter"
441 .RB "$" " ./some_program"
445 This file is provided only if the kernel was built with the
447 configuration option.
449 .SS Core dumps and systemd
453 framework, core dumps may be placed in a location determined by
459 feature that allows piping core dumps to a program.
460 One can verify this by checking whether core dumps are being piped to the
461 .BR systemd-coredump (8)
466 $ \fBcat /proc/sys/kernel/core_pattern\fP
467 |/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %e
471 In this case, core dumps will be placed in the location configured for
472 .BR systemd-coredump (8),
475 compressed files in the directory
476 .IR /var/lib/systemd/coredump/ .
477 One can list the core dumps that have been recorded by
478 .BR systemd-coredump (8)
484 $ \fBcoredumpctl list | tail -5\fP
485 Wed 2017-10-11 22:25:30 CEST 2748 1000 1000 3 present /usr/bin/sleep
486 Thu 2017-10-12 06:29:10 CEST 2716 1000 1000 3 present /usr/bin/sleep
487 Thu 2017-10-12 06:30:50 CEST 2767 1000 1000 3 present /usr/bin/sleep
488 Thu 2017-10-12 06:37:40 CEST 2918 1000 1000 3 present /usr/bin/cat
489 Thu 2017-10-12 08:13:07 CEST 2955 1000 1000 3 present /usr/bin/cat
493 The information shown for each core dump includes the date and time
494 of the dump, the PID, UID, and GID of the dumping process,
495 the signal number that caused the core dump,
496 and the pathname of the executable that was being run by the dumped process.
499 allow a specified coredump file to be pulled from the
501 location into a specified file.
502 For example, to extract the core dump for PID 2955 shown above to a file named
504 in the current directory, one could use:
508 $ \fBcoredumpctl dump 2955 -o core\fP
512 For more extensive details, see the
518 mechanism that archives core dumps, restoring to something more like
519 traditional Linux behavior, one can set an override for the
521 mechanism, using something like:
525 # echo "kernel.core_pattern=core.%p" > /etc/sysctl.d/50-coredump.conf
526 # /lib/systemd/systemd-sysctl
535 command can be used to obtain a core dump of a running process.
537 In Linux versions up to and including 2.6.27,
538 .\" Changed with commit 6409324b385f3f63a03645b4422e3be67348d922
539 if a multithreaded process (or, more precisely, a process that
540 shares its memory with another process by being created with the
544 dumps core, then the process ID is always appended to the core filename,
545 unless the process ID was already included elsewhere in the
549 .IR /proc/sys/kernel/core_pattern .
550 (This is primarily useful when employing the obsolete
551 LinuxThreads implementation,
552 where each thread of a process has a different PID.)
553 .\" Always including the PID in the name of the core file made
554 .\" sense for LinuxThreads, where each thread had a unique PID,
555 .\" but doesn't seem to serve any purpose with NPTL, where all the
556 .\" threads in a process share the same PID (as POSIX.1 requires).
557 .\" Probably the behavior is maintained so that applications using
558 .\" LinuxThreads continue appending the PID (the kernel has no easy
559 .\" way of telling which threading implementation the user-space
560 .\" application is using). -- mtk, April 2006
562 The program below can be used to demonstrate the use of the
564 .I /proc/sys/kernel/core_pattern
566 The following shell session demonstrates the use of this program
567 (compiled to create an executable named
568 .IR core_pattern_pipe_test ):
572 .RB "$" " cc \-o core_pattern_pipe_test core_pattern_pipe_test.c"
575 .RB "#" " echo \(dq|$PWD/core_pattern_pipe_test %p \
576 UID=%u GID=%g sig=%s\(dq > \e"
577 .B " /proc/sys/kernel/core_pattern"
580 .BR "^\e" " # type control-backslash"
582 .RB "$" " cat core.info"
584 argc[0]=</home/mtk/core_pattern_pipe_test>
589 Total bytes in core dump: 282624
595 /* core_pattern_pipe_test.c */
598 #include <sys/stat.h>
605 #define BUF_SIZE 1024
608 main(int argc, char *argv[])
616 /* Change our current working directory to that of the
619 snprintf(cwd, PATH_MAX, "/proc/%s/cwd", argv[1]);
622 /* Write output to file "core.info" in that directory */
624 fp = fopen("core.info", "w+");
628 /* Display command\-line arguments given to core_pattern
631 fprintf(fp, "argc=%d\\n", argc);
632 for (j = 0; j < argc; j++)
633 fprintf(fp, "argc[%d]=<%s>\\n", j, argv[j]);
635 /* Count bytes in standard input (the core dump) */
638 while ((nread = read(STDIN_FILENO, buf, BUF_SIZE)) > 0)
640 fprintf(fp, "Total bytes in core dump: %d\\n", tot);
658 .BR systemd-coredump (8)