klibc - Jul 2004 - Patch for bug in nuke_dirent in run-init.c

Hi

In run-init.c there is a minor bug which can cause major problems i.e.
not able to boot systems with udev-enabled initramfs on systems with
multiple cdrom drives. In nuke_dirent () in run-init.c the stat call
should be replaced with a lstat call, else nuking symbolic links which
have been created before the link target will fail. Was somewhat tricky
to debug... The bug gets triggered by udev with the /dev/cdrom symbolic
link rule (from the default udev rules file) if there is more than one
ide cdrom drive connected.

Attached the small patch against klibc-0.153.

Regards,

J?rg

PS: Last Saturday I've sent a patch to add xfs support to fstype and
haven't got any replies yet, any reasons?


-- 
J?rg Billeter <j@bitron.ch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: klibc-0.153-lstat-1.patch
Type: text/x-patch
Size: 451 bytes
Desc: not available
Url :
http://www.zytor.com/pipermail/klibc/attachments/20040729/53debfc0/klibc-0.153-lstat-1.bin

J?rg Billeter wrote:
> Hi
> 
> In run-init.c there is a minor bug which can cause major problems i.e.
> not able to boot systems with udev-enabled initramfs on systems with
> multiple cdrom drives. In nuke_dirent () in run-init.c the stat call
> should be replaced with a lstat call, else nuking symbolic links which
> have been created before the link target will fail. Was somewhat tricky
> to debug... The bug gets triggered by udev with the /dev/cdrom symbolic
> link rule (from the default udev rules file) if there is more than one
> ide cdrom drive connected.
> 
> Attached the small patch against klibc-0.153.
> 
Thanks!
> 
> PS: Last Saturday I've sent a patch to add xfs support to fstype and
> haven't got any replies yet, any reasons?
> 
I'm travelling, that's all...	

	-hpa