On Fri, 15 Mar 2013 08:48:46 -0500
Pat Riehecky <[log in to unmask]> wrote:

> On 03/15/2013 05:54 AM, Andras Horvath wrote:
> > Hi,
> >
> > Yesterday came a huge update regarding the kernel (2.6.32-358.2.1)
> > and the X (xorg and libs).
> >
> > System is SL 6.3, 64 bit.
> >
> > Since this update the sandbox windows cannot be resized with
> > Firefox. I run FF in sandbox with this command:
> >
> > sandbox -i ~/.mozilla/plugins -t sandbox_net_t -X -- firefox
> >
> > AFAIK the ability to resize an X sandbox window is a special patch
> > from TUV which therefore makes this ability to be available only on
> > RHEL (and so SL) and not on Fedora. This one came with an update
> > when the system version was at 6.2. Before that it was not
> > available.
> >
> > Another regression I seem to have noticed is that the adobe flash
> > plugin crashes or simply doesn't behave properly. Sometimes the
> > flash area on the web page stays blank on some specific sites.
> >
> > I tested these in FF with an empty profile too, but the same
> > happens. Didn't have any problem before yesterday.
> >
> > I'd appreciate any help. Thanks!
> >
> >
> > Andras
> 
> hmmmmm
> 
> I fear I've a ton of questions, so forgive my rather shotgun approach.
> 
> I do hate X updates as security errata, not all dependencies are
> listed as 'Requires' which makes for all sorts of fun.....
> 
> 
> 
> There is a new policycoreutils build for 6.4, nothing that I pushed
> yesterday listed it as a requirement so it wasn't pushed.  But it is
> the rpm providing 'sandbox'.  Maybe we've got a bugfix or two here....
> 
> Can I have you try updating those rpms:
> 
> http://ftp2.scientificlinux.org/linux/scientific/6rolling/x86_64/os/repoview/letter_p.group.html
> 
> (you can also do yum --releasever=6rolling update policycoreutil* if
> that is more convenient)
 
Ok, I'll update it and write the results.

> Are you having the flash problems in a sandbox or just naively?  If
> it is in a sandbox, does the problem persist when run naively?

In both. Also I just remembered, that I'm using flash plugin from my
mozilla dir (~/.mozilla/plugins) and I pass that dir to the sandboxed
FF - so I tested the flash in sandbox without passing the plugin, and
it's giving the same result. BTW, other flash contents like Youtube
videos work fine (flash and not the HTML5 version).

> Just to verify, you've booted into the new kernel, right? If not can
> I have you check and see if the problems follow you there?  (Forgive
> the seeming obviousness of the question, but sometimes the answer is
> 'no')

(No prob.) Yes, I booted into the new kernel.

> Do the problems you've reported persist on the older kernel?

Yes, problem is there with the old kernel too.

> I show the current adobe flash rpm as
> flash-plugin-11.2.202.275-release.x86_64 , it was released earlier in
> the week.  Did you get that update?

Yes I did.

> Any chance I can see a copy of your /etc/X11/xorg.conf?

I have none.

> If you've got the binary only drivers, what version are they and
> where did you get them?

I have an Intel GPU with a driver coming with the official kernel. No
any 3rd party driver anywhere on my system.

$ lspci | grep -i vga
00:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960
Integrated Graphics Controller (primary) (rev 0c)

> Any chance for your X logs from a session where you've got the
> problems?

I have several lines of these in my Xorg.0.log:

AUDIT: Thu Mar 14 21:38:35 2013: 2409: client 2 disconnected
AUDIT: Thu Mar 14 21:38:35 2013: 2409: client 2 connected from local
host ( uid=0 gid=0 pid=2503 ) Auth name: MIT-MAGIC-COOKIE-1 ID: 197

Can it be reported by SELinux and so might I have problems with SELinux
access deny somehow?

> I'll look at some stuff and see what I can see,
> 
> Pat

Also, I created a fresh install of SL 6.3 in a VM and the problem comes
up there too.

Thanks!