coredump开关

openwrt中一般默认没有打开coredump功能，所以我们要先开启改功能，按照下面的配置开启之后，如果进程异常崩溃了，则会在/tmp目录下生成coredump文件。

ulimit -c unlimited

coredump文件格式core_pattern

echo "/tmp/core-%e-%p-%t" > /proc/sys/kernel/core_pattern

%c 转储文件的大小上限

%e 所dump的文件名

%g 所dump的进程的实际组ID

%h 主机名

%p 所dump的进程PID

%s 导致本次coredump的信号

%t 转储时刻(由1970年1月1日起计的秒数)

%u 所dump进程的实际用户ID

echo "1" > /proc/sys/fs/suid_dumpable

       /proc/sys/fs/suid_dumpable (since Linux 2.6.13)
              The value in this file determines whether core dump files are produced for set-
              user-ID  or otherwise protected/tainted binaries.  Three different integer val‐
              ues can be specified:

              0 (default)
                     This provides the traditional (pre-Linux 2.6.13) behavior.  A core  dump
                     will  not  be  produced  for a process which has changed credentials (by
                     calling seteuid(2), setgid(2), or similar, or by executing a set-user-ID
                     or  set-group-ID  program) or whose binary does not have read permission
                     enabled.

              1 ("debug")
                     All processes dump core when possible.  The core dump is  owned  by  the
                     filesystem  user  ID  of the dumping process and no security is applied.
                     This is intended  for  system  debugging  situations  only.   Ptrace  is
                     unchecked.

              2 ("suidsafe")
                     Any  binary which normally would not be dumped (see "0" above) is dumped
                     readable by root only.  This allows the user to  remove  the  core  dump
                     file  but  not to read it.  For security reasons core dumps in this mode
                     will not overwrite one another or other files.  This mode is appropriate
                     when  administrators  are attempting to debug problems in a normal envi‐
                     ronment.

                     Additionally, since Linux 3.6, /proc/sys/kernel/core_pattern must either
                     be  an  absolute  pathname  or  a  pipe command, as detailed in core(5).
                     Warnings will be written to the kernel log if core_pattern does not fol‐
                     low these rules, and no core dump will be produced.

手动触发coredump

kill -s SIGABRT pid 或者 kill -6 pid

kill -s SIGSEGV pid 或者 kill -11 pid

man 5 core可以查看linux对coredump的说明。