firefox 19.0,2 crashing under FreeBSD 8

Craig Rodrigues rodrigc at FreeBSD.org
Fri Apr 5 01:34:38 UTC 2013


On Thu, Apr 4, 2013 at 3:24 PM, Jan Beich <jbeich at tormail.org> wrote:

> Craig Rodrigues <rodrigc at FreeBSD.org> writes:
>
> > Stacktrace 2:
> >
> > (gdb) where
> > #0  0x282e95f2 in flockfile () from /lib/libc.so.7
> > #1  0x282defb0 in vfprintf () from /lib/libc.so.7
> > #2  0x282cfd3e in fprintf () from /lib/libc.so.7
> > #3  0x282ec43a in .cerror () from /lib/libc.so.7
> > #4  0x283289a8 in __JCR_LIST__ () from /usr/local/lib/libffi.so.6
> > #5  0x28326fb7 in ffi_call_SYSV () from /usr/local/lib/libffi.so.6
> > #6  0x28326dee in ffi_call () from /usr/local/lib/libffi.so.6
> > #7  0x2a37c3dd in JS_SetCTypesCallbacks ()
> >    from /usr/local/lib/firefox/libxul.so
> > #8  0x333e1740 in ?? ()
> > #9  0x338b0ed4 in issetugid () from /usr/local/lib/compat/libc.so.6
>
> Can you show what's linked against libc.so.6 from misc/compat6x ?
> For example,
>
>   $ ldd -a /usr/local/lib/firefox/libxul.so
>


Take a look at:

http://people.freebsd.org/~rodrigc/qqq.txt.bz2

I don't see libc.so.6 anywhere in there, so I don't know how it could have
gotten into the
stack trace, unless the stack trace is mangled due to stack corruption.


>
> > I saw in this thread:
> >
> >
> http://lists.freebsd.org/pipermail/freebsd-ports/2013-February/081567.html
> >
> > That patching clang helped fix some stack issues with Firefox.
> >
> > Is this solution applicable to FreeBSD 8?
> > On FreeBSD 8, the firefox port is built with g++46, not clang.
>
> At this time only base clang is considered but during firefox 19
> lang/clang* could be picked up by default.
>


In ports/www/firefox/Makefile, there is this:

.if ${PORT_OPTIONS:MPGO}
USE_GCC?=       yes
USE_DISPLAY=            yes


I don't know if that triggers things, but I do,
cd /usr/ports/www/firefox
make -V USE_GCC

I get:

4.6+

--
Craig


More information about the freebsd-ports mailing list