[kib@freebsd.org: [src] cvs commit: src/sys/amd64/linux32 linux32_sysvec.c]

[Scot Hetzel]

On 11/3/06, Alexander Leidinger <Alexander at leidinger.net> wrote:
> Quoting Scot Hetzel <swhetzel at gmail.com> (from Thu, 2 Nov 2006
> 11:56:23 -0600):
>
> > execve(/compat/linux/usr/lib/../bin/wine-preloader)
> > :
> > Nov  2 08:36:53 hp010 kernel: linux(1380): set_thread_area(-1,
> > 9c2666c0, fffff, 1, 0, 0, 1, 0, 1
> > Nov  2 08:36:53 hp010 kernel: )
> > Nov  2 08:36:53 hp010 kernel: linux(1380): set_tid_address(0x48)
> > Nov  2 08:39:10 hp010 kernel: pid 1380 (wine-preloader), uid 0: exited
> > on signal 11 (core dumped)
>
> Is this with the patch Roman announced, or with the
> linuxolator-p4.diff on Leidinger.net?
>
This is with the Oct-29th linuxolator patch on Leidinger.net

> > This failure got as far as loading wine-preloader, before
> > wine-preloader core dumped, /compat/linux/usr/bin/wine was running, as
> > it was used to start wine-preloader.
> >
> > Instead, it fails now fails with:
> >
> > fatal process exception: page fault, fault VA = 0x8
> > Segmentation fault (core dumped)
> >
> > messages:
> > Nov  2 08:47:17 hp010 kernel: linux(1251): set_thread_area(-1,
> > 282ad6c0, fffff, 1, 0, 0, 1, 0, 1
> > Nov  2 08:47:17 hp010 kernel: )
> > Nov  2 08:47:17 hp010 kernel: pid 1251 (wine), uid 0: exited on signal
> > 11 (core dumped)
> >
> > Which occurs after wine executes set_thread_area.
>
> Did you reboot in between? There's a bug with TLS on amd64 which only
> allows to use TLS once. And I think Roman didn't fixed it yet (but
> have a look at the mailarchives, I think he asked for a test of some
> changes revently).
>
I tried it again, and if I reboot before running wineprefixcreate,
then I am able to reproduce the first error with wine-preloader.

So I guess it is being hit by this bug in TLS.

Scot

-- 
DISCLAIMER:
No electrons were mamed while sending this message. Only slightly bruised.