Re: 2.6.24-rc6-git7: Reported regressions from 2.6.23

[Greg KH <greg@kroah.com>]

On Tue, Jan 01, 2008 at 02:26:06PM -0800, Linus Torvalds wrote:
> 
> 
> On Tue, 1 Jan 2008, Rafael J. Wysocki wrote:
> > 
> > Subject	: Could not set non-blocking flag with 2.6.24-rc5
> > Submitter	: Tino Keitel <tino.keitel@gmx.de>
> > Date		: 2007-12-13 16:27
> > References	: http://lkml.org/lkml/2007/12/13/392
> > 		  http://bugzilla.kernel.org/show_bug.cgi?id=9557
> > Handled-By	: 
> > Patch		: 
> 
> That strace shows that trying to open /dev/null fails with ENXIO:
> 
> 	[pid  6050] open("/dev/null", O_RDONLY <unfinished ...>
> 	[pid  6050] <... open resumed> )        = -1 ENXIO (No such device or address)
> 
> and everything goes downhill from there.
> 
> It would be worth looking at your /dev/null to see what kind of (broken) 
> device node it is, and get a clue about *why* it is broken.
> 
> Greg, any udev breakage that could affect /dev/null? 

I do not know of any such breakage.

But, as this is in a chroot, hopefully udev has properly populated the
chrooted /dev tree for the process to provide the /dev/null device node?

Tino, did you provide a /dev/ for your chroot?  Did you use udev to
create it, or did you use a static /dev tree?

thanks,

greg k-h