devel/nspr's self-test failing: `mutex is on list'

[Jeremy Messenger]

On Thu, 19 Oct 2006 21:17:16 -0500, Mikhail Teterin  
<mi+mx at aldan.algebra.com> wrote:

> Unfortunately, the tests are not being run automatically every time the  
> port
> builds, so we don't know, when the problem began...
>
> To reproduce:
>
> 	% cd /usr/ports/devel/nspr
> 	% make build test
>
> [...]
>         forktest                        FAILED
>         forktest:       Wait one second before accept
>         forktest:       Wait one second before connect
>         forktest:       Connecting to port 51686
>         forktest:       Writing message "Hello world!"
>         forktest:       Accepting connection at port 51686
>         forktest:       Received "Hello world!" from the client
>         forktest:       The message is received correctly
>         forktest:       Fork succeeded.  Parent process continues.
>         forktest:       Wait one second before accept
>         forktest:       Wait one second before connect
>         forktest:       Fatal error 'mutex is on list' at line 540 in  
> file /usr/src/lib/libpthread/thread/thr_mutex.c (errno = 43)
>         forktest:       Fork succeeded.  Child process continues.
>         forktest:       Accepting connection at port 56714
>         forktest:       Connecting to port 56714
>         forktest:       Writing message "Hello world!"
>         forktest:       Received "Hello world!" from the client
>         forktest:       The message is received correctly
>         forktest:       Parent process exits.
>         forktest:       FAILED
> [...]
>
> I'm unsure, whether the bug is in the test itself or in the pthread,
> or in the nspr-library's threading implementation, but the test used
> to pass...
>
> Would a thread-expert care to investigate, please? Thanks!

It's known and there has a PR (ports/100785) for it with no solution. I  
have a backtrace that forktest dumped a core if you know how to dig in and  
solve it.

http://people.freebsd.org/~mezz/gdb/gdb-nspr-forktest.txt

marcus said, 'Looks like a potential symbol conflict, but without full  
libpthread and libc symbols, it's hard to tell.' I have have done compiled  
these with the debug but I haven't hear anything so far.

Cheers,
Mezz

> 	-mi


-- 
mezz7 at cox.net  -  mezz at FreeBSD.org
FreeBSD GNOME Team  -  FreeBSD Multimedia Hat (ports, not src)
http://www.FreeBSD.org/gnome/  -  gnome at FreeBSD.org
http://wiki.freebsd.org/multimedia  -  multimedia at FreeBSD.org