1 .\" Copyright (c) 2006, 2008 by Michael Kerrisk <mtk.manpages@gmail.com>
3 .\" Permission is granted to make and distribute verbatim copies of this
4 .\" manual provided the copyright notice and this permission notice are
5 .\" preserved on all copies.
7 .\" Permission is granted to copy and distribute modified versions of this
8 .\" manual under the conditions for verbatim copying, provided that the
9 .\" entire resulting derived work is distributed under the terms of a
10 .\" permission notice identical to this one.
12 .\" Since the Linux kernel and libraries are constantly changing, this
13 .\" manual page may be incorrect or out-of-date. The author(s) assume no
14 .\" responsibility for errors or omissions, or for damages resulting from
15 .\" the use of the information contained herein. The author(s) may not
16 .\" have taken the same level of care in the production of this manual,
17 .\" which is licensed free of charge, as they might when working
20 .\" Formatted or processed versions of this manual, if unaccompanied by
21 .\" the source, must acknowledge the copyright and authors of this work.
23 .TH CORE 5 2008-06-15 "Linux" "Linux Programmer's Manual"
25 core \- core dump file
27 The default action of certain signals is to cause a process to terminate
29 .IR "core dump file" ,
30 a disk file containing an image of the process's memory at
31 the time of termination.
32 A list of the signals which cause a process to dump core can be found in
35 A process can set its soft
37 resource limit to place an upper limit on the size of the core dump file
38 that will be produced if it receives a "core dump" signal; see
42 There are various circumstances in which a core dump file is
45 The process does not have permission to write the core file.
46 (By default the core file is called
48 and is created in the current working directory.
49 See below for details on naming.)
50 Writing the core file will fail if the directory in which
51 it is to be created is non-writable,
52 or if a file with the same name exists and
54 or is not a regular file
55 (e.g., it is a directory or a symbolic link).
57 A (writable, regular) file with the same name as would be used for the
58 core dump already exists, but there is more than one hard link to that
61 The file system where the core dump file would be created is full;
62 or has run out of inodes; or is mounted read-only;
63 or the user has reached their quota for the file system.
65 The directory in which the core dump file is to be created does
71 resource limits for a process are set to zero (see
74 The binary being executed by the process does not have read
77 The process is executing a set-user-ID (set-group-ID) program
78 that is owned by a user (group) other than the real user (group)
80 (However, see the description of the
83 operation, and the description of the
84 .I /proc/sys/fs/suid_dumpable
87 .SS Naming of core dump files
88 By default, a core dump file is named
91 .I /proc/sys/kernel/core_pattern
92 file (since Linux 2.6 and 2.4.21)
93 can be set to define a template that is used to name core dump files.
94 The template can contain % specifiers which are substituted
95 by the following values when a core file is created:
104 PID of dumped process
107 (numeric) real UID of dumped process
110 (numeric) real GID of dumped process
113 number of signal causing dump
116 time of dump, expressed as seconds since the Epoch (00:00h, 1\ Jan 1970, UTC)
119 hostname (same as \fInodename\fP returned by \fBuname\fP(2))
122 executable filename (without path prefix)
125 core file size soft resource limit of crashing process (since Linux 2.6.24)
129 A single % at the end of the template is dropped from the
130 core filename, as is the combination of a % followed by any
131 character other than those listed above.
132 All other characters in the template become a literal
133 part of the core filename.
134 The template may include \(aq/\(aq characters, which are interpreted
135 as delimiters for directory names.
136 The maximum size of the resulting core filename is 128 bytes (64 bytes
137 in kernels before 2.6.19).
138 The default value in this file is "core".
139 For backward compatibility, if
140 .I /proc/sys/kernel/core_pattern
141 does not include "%p" and
142 .I /proc/sys/kernel/core_uses_pid
144 is non-zero, then .PID will be appended to the core filename.
146 Since version 2.4, Linux has also provided
147 a more primitive method of controlling
148 the name of the core dump file.
150 .I /proc/sys/kernel/core_uses_pid
151 file contains the value 0, then a core dump file is simply named
153 If this file contains a non-zero value, then the core dump file includes
154 the process ID in a name of the form
156 .SS Piping core dumps to a program
157 Since kernel 2.6.19, Linux supports an alternate syntax for the
158 .I /proc/sys/kernel/core_pattern
160 If the first character of this file is a pipe symbol (\fB|\fP),
161 then the remainder of the line is interpreted as a program to be
163 Instead of being written to a disk file, the core dump is given as
164 standard input to the program.
165 Note the following points:
167 The program must be specified using an absolute pathname (or a
168 pathname relative to the root directory, \fI/\fP),
169 and must immediately follow the '|' character.
171 The process created to run the program runs as user and group
174 Command-line arguments can be supplied to the
175 program (since kernel 2.6.24),
176 delimited by white space (up to a total line length of 128 bytes).
178 The command-line arguments can include any of
179 the % specifiers listed above.
180 For example, to pass the PID of the process that is being dumped, specify
183 .SS Controlling which mappings are written to the core dump
184 Since kernel 2.6.23, the Linux-specific
185 .IR /proc/PID/coredump_filter
186 file can be used to control which memory segments are written to the
187 core dump file in the event that a core dump is performed for the
188 process with the corresponding process ID.
190 The value in the file is a bit mask of memory mapping types (see
192 If a bit is set in the mask, then memory mappings of the
193 corresponding type are dumped; otherwise they are not dumped.
194 The bits in this file have the following meanings:
200 Dump anonymous private mappings.
203 Dump anonymous shared mappings.
206 Dump file-backed private mappings.
209 Dump file-backed shared mappings.
210 .\" file-backed shared mappings of course also update the underlying
218 this reflects traditional Linux behavior and means that
219 only anonymous memory segments are dumped.
221 Memory-mapped I/O pages such as frame buffer are never dumped, and
222 virtual DSO pages are always dumped, regardless of the
226 A child process created via
233 value is preserved across an
236 It can be useful to set
238 in the parent shell before running a program, for example:
242 $ echo 0x7 > /proc/self/coredump_filter
247 This file is only provided if the kernel was built with the
248 CONFIG_ELF_CORE configuration option.
253 command can be used to obtain a core dump of a running process.
255 If a multithreaded process (or, more precisely, a process that
256 shares its memory with another process by being created with the
260 dumps core, then the process ID is always appended to the core filename,
261 unless the process ID was already included elsewhere in the
262 filename via a %p specification in
263 .IR /proc/sys/kernel/core_pattern .
264 (This is primarily useful when employing the LinuxThreads implementation,
265 where each thread of a process has a different PID.)
266 .\" Always including the PID in the name of the core file made
267 .\" sense for LinuxThreads, where each thread had a unique PID,
268 .\" but doesn't seem to serve any purpose with NPTL, where all the
269 .\" threads in a process share the same PID (as POSIX.1 requires).
270 .\" Probably the behavior is maintained so that applications using
271 .\" LinuxThreads continue appending the PID (the kernel has no easy
272 .\" way of telling which threading implementation the userspace
273 .\" application is using). -- mtk, April 2006
275 The program below can be used to demonstrate the use of the
277 .I /proc/sys/kernel/core_pattern
279 The following shell session demonstrates the use of this program
280 (compiled to create an executable named
281 .IR core_pattern_pipe_test ):
285 $ cc -o core_pattern_pipe_test core_pattern_pipe_test.c
288 # echo "|$PWD/core_pattern_pipe_test %p UID=%u GID=%g sig=%s" > \\
289 /proc/sys/kernel/core_pattern
292 .I type control-backslash
296 argc[0]=</home/mtk/core_pattern_pipe_test>
301 Total bytes in core dump: 282624
305 The source code of the program is as follows:
308 /* core_pattern_pipe_test.c */
311 #include <sys/stat.h>
318 #define BUF_SIZE 1024
321 main(int argc, char *argv[])
329 /* Change our current working directory to that of the
332 snprintf(cwd, PATH_MAX, "/proc/%s/cwd", argv[1]);
335 /* Write output to file "core.info" in that directory */
337 fp = fopen("core.info", "w+");
341 /* Display command\-line arguments given to core_pattern
344 fprintf(fp, "argc=%d\\n", argc);
345 for (j = 0; j < argc; j++)
346 fprintf(fp, "argc[%d]=<%s>\\n", j, argv[j]);
348 /* Count bytes in standard input (the core dump) */
351 while ((nread = read(STDIN_FILENO, buf, BUF_SIZE)) > 0)
353 fprintf(fp, "Total bytes in core dump: %d\\n", tot);