Subject: Re: pthread problem on MP system - is this the known issue?
To: Jason R Thorpe <thorpej@wasabisystems.com>
From: Jeff Rizzo <riz@boogers.sf.ca.us>
List: current-users
Date: 01/27/2003 09:34:41
On Mon, Jan 27, 2003 at 09:08:59AM -0800, Jason R Thorpe wrote:
> On Mon, Jan 27, 2003 at 08:55:14AM -0800, Jeff Rizzo wrote:
> 
> Please file a PR; this is a new one.  Can you try to reduce a test
> case out of the app in question?

I'd be happy to, but I will probably need a good deal of help.  

gdb seems to be giving me some problems... I don't think it's actually
dying where it says it is.  (As you can probably guess, I don't have
a whole lot of experience debugging threaded programs)

Here's what I get when starting it under gdb.  I set a breakpoint
at main(), then another breakpoint at the first function call from 
main(), but it dies quickly.  I'm guessing that our gdb may have some
problems with threads?  

Starting program: /home/riz/test -d -f /tmp/foo.conf

Breakpoint 1, main (argc=4, argv=0xbfbfb964) at main.c:22
22              detach_flag = 1;
(gdb) s
[New Thread 0]

Program received signal SIGBUS, Bus error.
[Switching to Thread 0]
0x48081d93 in pthread__switch_away () from /usr/lib/libpthread.so.0

The next two lines of source after "detach_flag=1;" are:

        filename = DEFAULT_CONFIG;
        i = 1;

... so I doubt those are the problems.  :)    I will try to find
the point where the first thread is started, since that's most likely
to yield results, but any suggestions on how to get gdb to play nice?

Thanks,
+j
-- 
Jeff Rizzo                                         http://boogers.sf.ca.us/~riz