Bug#435186: xorg: Crash

[Julien Cristau <jcristau@debian.org>]

reopen 435186
reassign 435186 nvidia-glx
retitle 435186 tries to mmap() /dev/zero with PROT_EXEC
thanks

On Mon, Jul 30, 2007 at 16:37:04 +0200, Piotr Kaczuba wrote:

> I had the following error message:
> 
> Backtrace:
> 0: /usr/bin/X(xf86SigHandler+0x81) [0x80c8631]
> 1: [0xffffe500]
> 
> Fatal server error:
> Caught signal 11.  Server aborting
> 
> No mention of nvidia. Although I do use the nvidia proprietary driver.
> Downgrading to udev 0.105-4 solved the problem.
> 
> A strace of X revealed the reason for the crash:
> 
> > 13597 open("/dev/zero", O_RDWR)         = 19
> > 13597 mmap2(NULL, 8192, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE, 19, 0x2000) = -1 EPERM (Operation not permitted)
> > 13597 close(19)                         = 0
> > 13597 --- SIGSEGV (Segmentation fault) @ 0 (0) ---
> 
> With udev 0.105-4 the mmap succeeds.
> The reason for EPERM is that the mmap call specifies PROT_EXEC but since
> udev 0.113-1 /dev is mounted with the noexec mount option.
> 
I don't see anything in the xserver code that would do that on a quick
look.  Reassigning to nvidia-glx for now.

Cheers,
Julien