topic Re: Unexpected SIGCHLD in Operating System - Linux

Unexpected SIGCHLD

Joost Bijen — Thu, 07 Jun 2007 08:08:38 GMT

Hi,

A program I wrote is started through inetd and handles input/output through stdin/stdout. It runs fine on my HP-UX 9000/800 single CPU box.

When moving towards another 9000/800 with 4 CPU the program unexpectedly crashes and does so after receiving a SIGCHLD. Can it be that the difference is in the multiprocessor arrangement? The program on the first machine does not receive the SIGCHLD...

I do not create children myself (in C that is...) but I have linked some BEA Tuxedo 8.1 libraries that may create threads that I am not aware of.

Any hints on where to start looking for this behaviour?

Thanks,
Joost

Re: Unexpected SIGCHLD

A. Clay Stephenson — Thu, 07 Jun 2007 09:04:27 GMT

I suspect that your single CPU box was working more or less by accident in that the real problem is timing related. Of course, without a stack trace or at least some tusc output, it's not possible to know what is wrong. The fix may be as simple as adding a signal handler for SIGCHLD.

Re: Unexpected SIGCHLD

Joost Bijen — Mon, 11 Jun 2007 01:05:44 GMT

The tusc trace provided me with exactly what I needed. I turned out that I hit the maxfiles kernel limit somewhere so I changed my piece of code that was the cause.

What I still do not know why this causes a SIGCHLD. nice to find out but I'm happy already :-)