GCC SSP breaks xorg-server

0000:00:00.0 Host bridge: Intel Corporation 82855PM Processor to I/O Controller (rev 03)
0000:00:01.0 PCI bridge: Intel Corporation 82855PM Processor to AGP Controller (rev 03)
0000:00:1d.0 USB Controller: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) USB UHCI Controller #1 (rev 01)
0000:00:1d.1 USB Controller: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) USB UHCI Controller #2 (rev 01)
0000:00:1d.2 USB Controller: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) USB UHCI Controller #3 (rev 01)
0000:00:1d.7 USB Controller: Intel Corporation 82801DB/DBM (ICH4/ICH4-M) USB2 EHCI Controller (rev 01)
0000:00:1e.0 PCI bridge: Intel Corporation 82801 Mobile PCI Bridge (rev 81)
0000:00:1f.0 ISA bridge: Intel Corporation 82801DBM (ICH4-M) LPC Interface Bridge (rev 01)
0000:00:1f.1 IDE interface: Intel Corporation 82801DBM (ICH4-M) IDE Controller (rev 01)
0000:00:1f.3 SMBus: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) SMBus Controller (rev 01)
0000:00:1f.5 Multimedia audio controller: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) AC'97 Audio Controller (rev 01)
0000:00:1f.6 Modem: Intel Corporation 82801DB/DBL/DBM (ICH4/ICH4-L/ICH4-M) AC'97 Modem Controller (rev 01)
0000:01:00.0 VGA compatible controller: ATI Technologies Inc Radeon Mobility M7 LW [Radeon Mobility 7500]
0000:02:00.0 CardBus bridge: Texas Instruments PCI4520 PC card Cardbus Controller (rev 01)
0000:02:00.1 CardBus bridge: Texas Instruments PCI4520 PC card Cardbus Controller (rev 01)
0000:02:01.0 Ethernet controller: Intel Corporation 82540EP Gigabit Ethernet Controller (Mobile) (rev 03)

I see the same on my ThinkPad T42 with the same Radeon Mobility M7 LW that Johan has. It doesn't happen quite 100% of the time, and if the server manages to start up, it works fine. If I start the server with startx rather than gdm, that doesn't crash, so this is somehow triggered by something gdm does.

Here's a screenshot showing the problem. It's difficult to see, but a bit below the mouse cursor you can see that the text entry box for gdm has been drawn. The rest of the screen is a lighter color.

Desktop team: can you think of anything gdm might be doing at this time to trigger this bug?

Oh, it's also worth noting that this was happening to me with xserver-xorg-driver-ati as well, so the problem predates that upgrade and may not be this package's fault at all.

Can anyone confirm this problem on Debian? I have a sneaking suspicion it could be SSP-related

On a hunch, I rebuilt xorg-server and xserver-xorg-video-ati with -fno-stack-protector, and I am completely unable to reproduce the problem with those binaries. I restarted gdm 5 times from single-user mode and also rebooted the system 5 times. Given that the problem was extremely repeatable before, this is fairly compelling.

It's possible that this could explain my mysterious bug #54726 as well (different machine), though that one is not repeatable so it's difficult to tell.

I'm uploading my binary packages built with -fno-stack-protector to http://people.ubuntu.com/~mdz/bugs/54650/

Johan, could you try them and see if this corrects the problem for you as well?

Further testing:

I backed out to the xserver-xorg-video-ati in the archive, and the problem remained unreproducible. I backed out to the xserver-xorg-core in the archive, and the problem became easily reproducible again. So I think it should be sufficient to build only xorg-server with -fno-stack-protector. I'm uploading that change now, and Matthias has added this bug to the list of known SSP issues for further investigation:

https://wiki.ubuntu.com/GccSsp

xorg-server (1:1.1.1-0ubuntu3) edgy; urgency=low

  * Build with -fno-stack-protector (see LP#54650)

 -- Matt Zimmerman <email address hidden> Mon, 31 Jul 2006 16:28:59 -0700

Matt,

First a big thank you for this.
The update to xorg-server 1.1.1-0ubuntu3 works perfect here !
Glad i'am back on my T42.
Rebooted well 10 times and the server starts fine,first I thought that my
video card was broken but I have a dual boot with dapper but that was not possible.
Glad i'am back on my T42
I see on your webpage that you uploaded a packages xserver-xorg-video-ati_6.6.1-0ubuntu1_i386.deb,must I test this binary or leave it now
this way?

Thanks again !
Johan

On Tue, Aug 01, 2006 at 06:35:30PM -0000, Johan Van Hellemont wrote:
> First a big thank you for this.
> The update to xorg-server 1.1.1-0ubuntu3 works perfect here !
> Glad i'am back on my T42.
> Rebooted well 10 times and the server starts fine,first I thought that my
> video card was broken but I have a dual boot with dapper but that was not possible.
> Glad i'am back on my T42
> I see on your webpage that you uploaded a packages xserver-xorg-video-ati_6.6.1-0ubuntu1_i386.deb,must I test this binary or leave it now
> this way?

That's all I needed; thanks for helping with the testing.

--
 - mdz

Matt, could you try building the server in 7.04 without that option to see if it could be dropped?

I'd like to hear from the toolchain about this; since the problem was intermittent, we can't necessarily take a success report to mean that the option is now safe. If a problem was identified and fixed in GCC, we should feel confident in disabling the workaround, but otherwise, I'm inclined to be conservative and leave it as-is.

I'm unware of changes in gcc-4.1; we updated glibc to a newer version in the meantime.

Setting to "Incomplete". As discussed on the distro sprint in 2007/07, this should be rechecked in gutsy with the current gcc-4.1 and gcc-4.2.

I've completed a build using gcc 4.1.3 on Gutsy without the -fno-stack-protector option. I don't have an ATI system I can test this on, but it seems to work alright on my -nv system.

http://people.ubuntu.com/~bryce/Testing/xorg-server-stack-protect/xserver-xorg-core_1.3.0.0.dfsg-6ubuntu3~ppa1_i386.deb

Johan and/or Matt, could you try the above deb and see if the issue is still present?

Bryce

On Wed, Aug 01, 2007 at 06:37:56AM -0000, Bryce Harrington wrote:
> I've completed a build using gcc 4.1.3 on Gutsy without the -fno-stack-
> protector option. I don't have an ATI system I can test this on, but it
> seems to work alright on my -nv system.
>
> http://people.ubuntu.com/~bryce/Testing/xorg-server-stack-protect
> /xserver-xorg-core_1.3.0.0.dfsg-6ubuntu3~ppa1_i386.deb
>
> Johan and/or Matt, could you try the above deb and see if the issue is
> still present?

I don't have the system in question anymore, so I can't confirm.

--
 - mdz

I've built xorg-server_1.3.0.dfsg-12ubuntu1 locally, without -fno-stack-protector, and it works fine with my radeon 8500.

Bryce,

I tryed the above deb from you on my laptop with an Radeon Mobility M300 and everything works fine here.
Which tests can do I more?

ps:sorry for the late reply,wash on vacation.

This is now working alright for a number of people without compiling the server without ssp. Gutsy already has a version without -fno-stack-protection flag.