Signal #14 message - more details required

[Philip Kernohan]

Hello,

we have an application that occassionally puts out a message:

process (pid=13224) terminated on un-caught signal (sig=14), no core file was generated

In /usr/include/sys/sysinfo.h it appears that signal 14 is defined as:

define FPE_ASSISTX 14 /* HP Assist exception trap */

Can anybody suggest more details on this signal as it what it implies? This one has stumped our development people for a few days now.

Could the logging to the syslog be overwhelmed to the extent that a process gives this signal message and gives up writing to the syslog.

Any suggestions greatly appreciated.

Regards,
Philip

[A. Clay Stephenson]

Hi Phillip,

I think that you are really seeing SIGALRM. Meaning that the process was sent an alarm clock signal without a signal handler.

The typical way this occurs is that you do a setjmp() but fail to include an alarm(0) call to cancel the alarm signal in case the code executes normally (i.e. does not timeout).

Check your code for alarm() system calls. This could also happen because sleep gets called in an unexpected place (it too uses alarm()).

Hope this helps, Clay

[Klaus Crusius]

You can use "kill -l" to see a list of signal numbers and corresponding values.
"man 5 signal" gives a comprehensive description what the signals are used for.


Kind regards, Klaus