1CORE(5) Linux Programmer's Manual CORE(5)
2
6 core - core dump file
7
9 The default action of certain signals is to cause a process to termi‐
10 nate and produce a core dump file, a file containing an image of the
11 process's memory at the time of termination. This image can be used in
12 a debugger (e.g., gdb(1)) to inspect the state of the program at the
13 time that it terminated. A list of the signals which cause a process
14 to dump core can be found in signal(7).
15 A process can set its soft RLIMIT_CORE resource limit to place an upper
17 limit on the size of the core dump file that will be produced if it
18 receives a "core dump" signal; see getrlimit(2) for details.
19 There are various circumstances in which a core dump file is not pro‐
21 duced:
22 * The process does not have permission to write the core file. (By
24 default, the core file is called core or core.pid, where pid is the
25 ID of the process that dumped core, and is created in the current
26 working directory. See below for details on naming.) Writing the
27 core file fails if the directory in which it is to be created is not
28 writable, or if a file with the same name exists and is not writable
29 or is not a regular file (e.g., it is a directory or a symbolic
30 link).
31 * A (writable, regular) file with the same name as would be used for
33 the core dump already exists, but there is more than one hard link
34 to that file.
35 * The filesystem where the core dump file would be created is full; or
37 has run out of inodes; or is mounted read-only; or the user has
38 reached their quota for the filesystem.
39 * The directory in which the core dump file is to be created does not
41 exist.
42 * The RLIMIT_CORE (core file size) or RLIMIT_FSIZE (file size)
44 resource limits for the process are set to zero; see getrlimit(2)
45 and the documentation of the shell's ulimit command (limit in
46 csh(1)).
47 * The binary being executed by the process does not have read permis‐
49 sion enabled. (This is a security measure to ensure that an exe‐
50 cutable whose contents are not readable does not produce a—possibly
51 readable—core dump containing an image of the executable.)
52 * The process is executing a set-user-ID (set-group-ID) program that
54 is owned by a user (group) other than the real user (group) ID of
55 the process, or the process is executing a program that has file
56 capabilities (see capabilities(7)). (However, see the description
57 of the prctl(2) PR_SET_DUMPABLE operation, and the description of
58 the /proc/sys/fs/suid_dumpable file in proc(5).)
59 * /proc/sys/kernel/core_pattern is empty and /proc/sys/ker‐
61 nel/core_uses_pid contains the value 0. (These files are described
62 below.) Note that if /proc/sys/kernel/core_pattern is empty and
63 /proc/sys/kernel/core_uses_pid contains the value 1, core dump files
64 will have names of the form .pid, and such files are hidden unless
65 one uses the ls(1) -a option.
66 * (Since Linux 3.7) The kernel was configured without the CONFIG_CORE‐
68 DUMP option.
69 In addition, a core dump may exclude part of the address space of the
71 process if the madvise(2) MADV_DONTDUMP flag was employed.
72 On systems that employ systemd(1) as the init framework, core dumps may
74 instead be placed in a location determined by systemd(1). See below
75 for further details.
76 Naming of core dump files
78 By default, a core dump file is named core, but the /proc/sys/ker‐
79 nel/core_pattern file (since Linux 2.6 and 2.4.21) can be set to define
80 a template that is used to name core dump files. The template can con‐
81 tain % specifiers which are substituted by the following values when a
82 core file is created:
83 %% A single % character.
85 %c Core file size soft resource limit of crashing process (since
86 Linux 2.6.24).
87 %d Dump mode—same as value returned by prctl(2) PR_GET_DUMPABLE
88 (since Linux 3.7).
89 %e The process or thread's comm value, which typically is the same
90 as the executable filename (without path prefix, and truncated
91 to a maximum of 15 characters), but may have been modified to
92 be something different; see the discussion of /proc/[pid]/comm
93 and /proc/[pid]/task/[tid]/comm in proc(5).
94 %E Pathname of executable, with slashes ('/') replaced by exclama‐
95 tion marks ('!') (since Linux 3.0).
96 %g Numeric real GID of dumped process.
97 %h Hostname (same as nodename returned by uname(2)).
98 %i TID of thread that triggered core dump, as seen in the PID
99 namespace in which the thread resides (since Linux 3.18).
100 %I TID of thread that triggered core dump, as seen in the initial
101 PID namespace (since Linux 3.18).
102 %p PID of dumped process, as seen in the PID namespace in which
103 the process resides.
104 %P PID of dumped process, as seen in the initial PID namespace
105 (since Linux 3.12).
106 %s Number of signal causing dump.
107 %t Time of dump, expressed as seconds since the Epoch, 1970-01-01
108 00:00:00 +0000 (UTC).
109 %u Numeric real UID of dumped process.
110 A single % at the end of the template is dropped from the core file‐
112 name, as is the combination of a % followed by any character other than
113 those listed above. All other characters in the template become a lit‐
114 eral part of the core filename. The template may include '/' charac‐
115 ters, which are interpreted as delimiters for directory names. The
116 maximum size of the resulting core filename is 128 bytes (64 bytes in
117 kernels before 2.6.19). The default value in this file is "core". For
118 backward compatibility, if /proc/sys/kernel/core_pattern does not
119 include %p and /proc/sys/kernel/core_uses_pid (see below) is nonzero,
120 then .PID will be appended to the core filename.
121 Paths are interpreted according to the settings that are active for the
123 crashing process. That means the crashing process's mount namespace
124 (see mount_namespaces(7)), its current working directory (found via
125 getcwd(2)), and its root directory (see chroot(2)).
126 Since version 2.4, Linux has also provided a more primitive method of
128 controlling the name of the core dump file. If the /proc/sys/ker‐
129 nel/core_uses_pid file contains the value 0, then a core dump file is
130 simply named core. If this file contains a nonzero value, then the
131 core dump file includes the process ID in a name of the form core.PID.
132 Since Linux 3.6, if /proc/sys/fs/suid_dumpable is set to 2 ("suid‐
134 safe"), the pattern must be either an absolute pathname (starting with
135 a leading '/' character) or a pipe, as defined below.
136 Piping core dumps to a program
138 Since kernel 2.6.19, Linux supports an alternate syntax for the
139 /proc/sys/kernel/core_pattern file. If the first character of this
140 file is a pipe symbol (|), then the remainder of the line is inter‐
141 preted as the command-line for a user-space program (or script) that is
142 to be executed.
143 Since kernel 5.3.0, the pipe template is split on spaces into an argu‐
145 ment list before the template parameters are expanded. In earlier ker‐
146 nels, the template parameters are expanded first and the resulting
147 string is split on spaces into an argument list. This means that in
148 earlier kernels executable names added by the %e and %E template param‐
149 eters could get split into multiple arguments. So the core dump han‐
150 dler needs to put the executable names as the last argument and ensure
151 it joins all parts of the executable name using spaces. Executable
152 names with multiple spaces in them are not correctly represented in
153 earlier kernels, meaning that the core dump handler needs to use mecha‐
154 nisms to find the executable name.
155 Instead of being written to a file, the core dump is given as standard
157 input to the program. Note the following points:
158 * The program must be specified using an absolute pathname (or a path‐
160 name relative to the root directory, /), and must immediately follow
161 the '|' character.
162 * The command-line arguments can include any of the % specifiers
164 listed above. For example, to pass the PID of the process that is
165 being dumped, specify %p in an argument.
166 * The process created to run the program runs as user and group root.
168 * Running as root does not confer any exceptional security bypasses.
170 Namely, LSMs (e.g., SELinux) are still active and may prevent the
171 handler from accessing details about the crashed process via
172 /proc/[pid].
173 * The program pathname is interpreted with respect to the initial
175 mount namespace as it is always executed there. It is not affected
176 by the settings (e.g., root directory, mount namespace, current
177 working directory) of the crashing process.
178 * The process runs in the initial namespaces (PID, mount, user, and so
180 on) and not in the namespaces of the crashing process. One can uti‐
181 lize specifiers such as %P to find the right /proc/[pid] directory
182 and probe/enter the crashing process's namespaces if needed.
183 * The process starts with its current working directory as the root
185 directory. If desired, it is possible change to the working direc‐
186 tory of the dumping process by employing the value provided by the
187 %P specifier to change to the location of the dumping process via
188 /proc/[pid]/cwd.
189 * Command-line arguments can be supplied to the program (since Linux
191 2.6.24), delimited by white space (up to a total line length of 128
192 bytes).
193 * The RLIMIT_CORE limit is not enforced for core dumps that are piped
195 to a program via this mechanism.
196 /proc/sys/kernel/core_pipe_limit
198 When collecting core dumps via a pipe to a user-space program, it can
199 be useful for the collecting program to gather data about the crashing
200 process from that process's /proc/[pid] directory. In order to do this
201 safely, the kernel must wait for the program collecting the core dump
202 to exit, so as not to remove the crashing process's /proc/[pid] files
203 prematurely. This in turn creates the possibility that a misbehaving
204 collecting program can block the reaping of a crashed process by simply
205 never exiting.
206 Since Linux 2.6.32, the /proc/sys/kernel/core_pipe_limit can be used to
208 defend against this possibility. The value in this file defines how
209 many concurrent crashing processes may be piped to user-space programs
210 in parallel. If this value is exceeded, then those crashing processes
211 above this value are noted in the kernel log and their core dumps are
212 skipped.
213 A value of 0 in this file is special. It indicates that unlimited pro‐
215 cesses may be captured in parallel, but that no waiting will take place
216 (i.e., the collecting program is not guaranteed access to /proc/<crash‐
217 ing-PID>). The default value for this file is 0.
218 Controlling which mappings are written to the core dump
220 Since kernel 2.6.23, the Linux-specific /proc/[pid]/coredump_filter
221 file can be used to control which memory segments are written to the
222 core dump file in the event that a core dump is performed for the
223 process with the corresponding process ID.
224 The value in the file is a bit mask of memory mapping types (see
226 mmap(2)). If a bit is set in the mask, then memory mappings of the
227 corresponding type are dumped; otherwise they are not dumped. The bits
228 in this file have the following meanings:
229 bit 0 Dump anonymous private mappings.
231 bit 1 Dump anonymous shared mappings.
232 bit 2 Dump file-backed private mappings.
233 bit 3 Dump file-backed shared mappings.
234 bit 4 (since Linux 2.6.24)
235 Dump ELF headers.
236 bit 5 (since Linux 2.6.28)
237 Dump private huge pages.
238 bit 6 (since Linux 2.6.28)
239 Dump shared huge pages.
240 bit 7 (since Linux 4.4)
241 Dump private DAX pages.
242 bit 8 (since Linux 4.4)
243 Dump shared DAX pages.
244 By default, the following bits are set: 0, 1, 4 (if the CON‐
246 FIG_CORE_DUMP_DEFAULT_ELF_HEADERS kernel configuration option is
247 enabled), and 5. This default can be modified at boot time using the
248 coredump_filter boot option.
249 The value of this file is displayed in hexadecimal. (The default value
251 is thus displayed as 33.)
252 Memory-mapped I/O pages such as frame buffer are never dumped, and vir‐
254 tual DSO (vdso(7)) pages are always dumped, regardless of the core‐
255 dump_filter value.
256 A child process created via fork(2) inherits its parent's coredump_fil‐
258 ter value; the coredump_filter value is preserved across an execve(2).
259 It can be useful to set coredump_filter in the parent shell before run‐
261 ning a program, for example:
262 $ echo 0x7 > /proc/self/coredump_filter
264 $ ./some_program
265 This file is provided only if the kernel was built with the CON‐
267 FIG_ELF_CORE configuration option.
268 Core dumps and systemd
270 On systems using the systemd(1) init framework, core dumps may be
271 placed in a location determined by systemd(1). To do this, systemd(1)
272 employs the core_pattern feature that allows piping core dumps to a
273 program. One can verify this by checking whether core dumps are being
274 piped to the systemd-coredump(8) program:
275 $ cat /proc/sys/kernel/core_pattern
277 |/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %e
278 In this case, core dumps will be placed in the location configured for
280 systemd-coredump(8), typically as lz4(1) compressed files in the direc‐
281 tory /var/lib/systemd/coredump/. One can list the core dumps that have
282 been recorded by systemd-coredump(8) using coredumpctl(1):
283 $ coredumpctl list | tail -5
285 Wed 2017-10-11 22:25:30 CEST 2748 1000 1000 3 present /usr/bin/sleep
286 Thu 2017-10-12 06:29:10 CEST 2716 1000 1000 3 present /usr/bin/sleep
287 Thu 2017-10-12 06:30:50 CEST 2767 1000 1000 3 present /usr/bin/sleep
288 Thu 2017-10-12 06:37:40 CEST 2918 1000 1000 3 present /usr/bin/cat
289 Thu 2017-10-12 08:13:07 CEST 2955 1000 1000 3 present /usr/bin/cat
290 The information shown for each core dump includes the date and time of
292 the dump, the PID, UID, and GID of the dumping process, the signal
293 number that caused the core dump, and the pathname of the executable
294 that was being run by the dumped process. Various options to core‐
295 dumpctl(1) allow a specified coredump file to be pulled from the sys‐
296 temd(1) location into a specified file. For example, to extract the
297 core dump for PID 2955 shown above to a file named core in the current
298 directory, one could use:
299 $ coredumpctl dump 2955 -o core
301 For more extensive details, see the coredumpctl(1) manual page.
303 To (persistently) disable the systemd(1) mechanism that archives core
305 dumps, restoring to something more like traditional Linux behavior, one
306 can set an override for the systemd(1) mechanism, using something like:
307 # echo "kernel.core_pattern=core.%p" > \
309 /etc/sysctl.d/50-coredump.conf
310 # /lib/systemd/systemd-sysctl
311 It is also possible to temporarily (i.e., until the next reboot) change
313 the core_pattern setting using a command such as the following (which
314 causes the names of core dump files to include the executable name as
315 well as the number of the signal which triggered the core dump):
316 # sysctl -w kernel.core_pattern="%e-%s.core"
318
320 The gdb(1) gcore command can be used to obtain a core dump of a running
321 process.
322 In Linux versions up to and including 2.6.27, if a multithreaded
324 process (or, more precisely, a process that shares its memory with
325 another process by being created with the CLONE_VM flag of clone(2))
326 dumps core, then the process ID is always appended to the core file‐
327 name, unless the process ID was already included elsewhere in the file‐
328 name via a %p specification in /proc/sys/kernel/core_pattern. (This is
329 primarily useful when employing the obsolete LinuxThreads implementa‐
330 tion, where each thread of a process has a different PID.)
331
333 The program below can be used to demonstrate the use of the pipe syntax
334 in the /proc/sys/kernel/core_pattern file. The following shell session
335 demonstrates the use of this program (compiled to create an executable
336 named core_pattern_pipe_test):
337 $ cc -o core_pattern_pipe_test core_pattern_pipe_test.c
339 $ su
340 Password:
341 # echo "|$PWD/core_pattern_pipe_test %p UID=%u GID=%g sig=%s" > \
342 /proc/sys/kernel/core_pattern
343 # exit
344 $ sleep 100
345 ^\ # type control-backslash
346 Quit (core dumped)
347 $ cat core.info
348 argc=5
349 argc[0]=</home/mtk/core_pattern_pipe_test>
350 argc[1]=<20575>
351 argc[2]=<UID=1000>
352 argc[3]=<GID=100>
353 argc[4]=<sig=3>
354 Total bytes in core dump: 282624
355 Program source
357 /* core_pattern_pipe_test.c */
359 #define _GNU_SOURCE
361 #include <sys/stat.h>
362 #include <fcntl.h>
363 #include <limits.h>
364 #include <stdio.h>
365 #include <stdlib.h>
366 #include <unistd.h>
367 #define BUF_SIZE 1024
369 int
371 main(int argc, char *argv[])
372 {
373 int tot, j;
374 ssize_t nread;
375 char buf[BUF_SIZE];
376 FILE *fp;
377 char cwd[PATH_MAX];
378 /* Change our current working directory to that of the
380 crashing process */
381 snprintf(cwd, PATH_MAX, "/proc/%s/cwd", argv[1]);
383 chdir(cwd);
384 /* Write output to file "core.info" in that directory */
386 fp = fopen("core.info", "w+");
388 if (fp == NULL)
389 exit(EXIT_FAILURE);
390 /* Display command-line arguments given to core_pattern
392 pipe program */
393 fprintf(fp, "argc=%d\n", argc);
395 for (j = 0; j < argc; j++)
396 fprintf(fp, "argc[%d]=<%s>\n", j, argv[j]);
397 /* Count bytes in standard input (the core dump) */
399 tot = 0;
401 while ((nread = read(STDIN_FILENO, buf, BUF_SIZE)) > 0)
402 tot += nread;
403 fprintf(fp, "Total bytes in core dump: %d\n", tot);
404 fclose(fp);
406 exit(EXIT_SUCCESS);
407 }
408
410 bash(1), coredumpctl(1), gdb(1), getrlimit(2), mmap(2), prctl(2),
411 sigaction(2), elf(5), proc(5), pthreads(7), signal(7), systemd-core‐
412 dump(8)
413
415 This page is part of release 5.07 of the Linux man-pages project. A
416 description of the project, information about reporting bugs, and the
417 latest version of this page, can be found at
418 https://www.kernel.org/doc/man-pages/.
419Linux 2020-06-09 CORE(5)