vm_start_mutex use in IcedTeaNPPlugin.cc looks fishy

[Deepak Bhole]

* pcpa at mandriva.com.br <pcpa at mandriva.com.br> [2012-05-23 12:46]:
> Deepak Bhole escreveu:
> > Hi Martin,
> >
> > * Martin Olsson <mnemo at minimum.se> [2012-05-23 11:44]:
> >> The usage of "vm_start_mutex" in IcedTeaNPPlugin.cc looks buggy to me.
> >> Won't each thread just create and lock it's own mutex, thereby voiding the
> >> raison d'être of the mutex?
> >>
> >
> > Yep. I just noticed this a couple of days ago myself when trying to
> > investigate zombie VMs someone was seeing (though it was unrelated).
> 
>   I have a patch for it in
> http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=599
> I ended up adding another unrelated patch to the same bug report
> by mistake, see the icedtea6-1.8.2-mutex_and_leak.patch
> attachment. BTW, there is an issue with the bugzilla, that does
> not allow to see text/plain in the browser.
> 
> > Haven't gotten around to creating a patch for it, but please feel free
> > to. If not, it is still on my list (though not critical since I have
> > never seen a bug due to it).
> 
>   I remember it corrected problems for me, with multiple instances
> of jmol, but did not test it without the patch for a long time.
> 

Thanks Paulo! I took a slightly different approach and have posted a
patch for review here for this issue:
http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2012-May/018700.html

As for the second issue (msg_queue_mutex), the patch looks good to me. I
will do some quick testing to ensure all is good and commit it on your
behalf. I'll post a reply here when done.

Thanks again,
Deepak

> > Cheers,
> > Deepak
> 
> Paulo
>