[Linux-ia64] realtime signals not held

From: Greg Edwards <edwardsg_at_sgi.com>
Date: 2003-01-17 08:56:08
Hi,

A part of one of our test suites that checks signal holding/releasing
between a parent and forked child appears to have problems with realtime
signals.  The test forks a child, has the child mask two signals, the
parent then sends the signals to the child, and the child should release
and handle one of the signals.  None of the realtime signals are ever
held, they are caught right away.

I've reproduced the behavior on several ia64 boxes, including a rawhide
box with glibc-2.3.1 and a 2.4.19 kernel.  The test runs fine on an ia32
box.  I've attached the files from the test, and just compile and run them
by:

$ gcc -Ddebug set_usig.c t_result.c sgrtcs02.c
$ ./a.out
...{edited}...
read_pipe: waiting...
write_pipe: sending ready.
read_pipe: received ready.
child: handler phase1: caught signal 45.
sgrtcs02    4  BROK  :  A signal was caught before being released.

Can anyone shed some light on what is going wrong here?  Thanks.

Regards,
Greg Edwards



Received on Thu Jan 16 13:56:22 2003

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:11 EST