Re: [xml] 2.4.27 thread problems

[Gary Pennington sun com]

On Tue, Nov 19, 2002 at 09:31:07AM -0500, Daniel Veillard wrote:
> On Tue, Nov 19, 2002 at 12:20:46PM +0000, Gary Pennington sun com wrote:
> > Hi,
> >
> > There are problems in the threading support when I build 2.4.27 on
> > Solaris with the Sun compilers (I haven't tried with gcc, but I think
> > the same problem would appear).
>
>   Hum, I haven't compiled with thread in a long time... But I remember
> Igor made some patches to unify Windows/Unix threading recently, that may
> be the cause of the problem.

There has been a huge amount of change to the threading support code since
2.4.26. It looks like there is a problem with the Unix builds. I can't
comment on the Windows build.

> > It looks like there has been a lot of change in the threading code since
> > 2.4.26. I've just spent an hour or so trying to work out what is broken,
> > but I'm not making much progress at the moment.
>
>    Igor ???
>
> > Does the library build properly on other platforms, e.g. Windows, Linux?
>
>    Just rebuilt with --with-threads on linux Red Hat 8.0, it compiles fine
> the general tests pass normally but not the thread one:

It compiles fine for me too, but the code it generates is not so fine :-(

> paphio:~/XML -> ./testThreads
> Segmentation fault
>
> Running under gdb , I see a bunch of threads being created ...
> [New Thread 2064637 (LWP 1319)]
> [New Thread 2072830 (LWP 1320)]
> [New Thread 2081023 (LWP 1321)]
> pthread_create: Interrupted system call
>
> Program exited with code 01.
> (gdb)
>
>    hum, that doesn't help :-(
>
> strace ends up the following way:
>
> --- SIGRTMIN (Real-time signal 0) ---
> <... rt_sigsuspend resumed> )           = -1 EINTR (Interrupted system call)
> sigreturn()                             = ? (mask now [RTMIN])
> write(4, "\2000\3 \1\0\0\0\2\0X\1\2000\3@\2000\3 \0\0\0\0\0\0\0\0"..., 148) = 148
> rt_sigprocmask(SIG_SETMASK, NULL, [RTMIN], 8) = 0
> rt_sigsuspend([] <unfinished ...>
> --- SIGRTMIN (Real-time signal 0) ---
> <... rt_sigsuspend resumed> )           = -1 EINTR (Interrupted system call)
> sigreturn()                             = ? (mask now [RTMIN])
> write(4, "\2000\3 \1\0\0\0\3 X\1\2000\3 \2000\3@\0\0\0\0\0\0\0\0"..., 148) = 148rt_sigprocmask(SIG_SETMASK, 
> NULL, [RTMIN], 8) = 0
> rt_sigsuspend([] <unfinished ...>
> --- SIGRTMIN (Real-time signal 0) ---
> <... rt_sigsuspend resumed> )           = -1 EINTR (Interrupted system call)
> --- SIGSEGV (Segmentation fault) ---
> +++ killed by SIGSEGV +++
> paphio:~/XML ->
>
>    not much better :-\
>
> Daniel
>
> -- 
> Daniel Veillard      | Red Hat Network https://rhn.redhat.com/
> veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
> http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/

I'll try to make some time to look at this later in the week, but I'm out
of time today.

Gary
-- 
Gary Pennington
Solaris Kernel Development,
Sun Microsystems
Gary Pennington sun com