Bug 699711

the code got patched:

267,279c269,304
< if [ $1 = 0 -a -f %{_initrddir}/udev ]; then
<       if [ -x /sbin/pidof ]; then
<               pid=$(/sbin/pidof -c udevd)
<               while [ -n "$pid" ]; do
<                       for p in $pid; do
<                               kill $hard $p;
<                       done
<                       sleep 0.2
<                       pid=$(/sbin/pidof -c udevd)
<                       hard="-9"
<               done
<       fi
<       /sbin/chkconfig --del udev
---
> if [ "$1" -eq 0 -a -f %{_initrddir}/udev ]; then
>      if [ -f /proc/1/exe -a -d /proc/1/root ] && [ / -ef /proc/1/root ]; then
>           if [ -x /sbin/pidof ]; then
>               pid=$(/sbin/pidof -c udevd)
>               while [ -n "$pid" ]; do
>                   for p in $pid; do
>                        kill $hard $p 2>/dev/null
>                   done
>                   sleep 0.2
>                   pid=$(/sbin/pidof -c udevd)
>                   hard="-9"
>               done
>           fi
>      fi
>      /sbin/chkconfig --del udev
> fi
> if [ "$1" -eq 0 ]; then
>      /sbin/chkconfig --del udev-retry
> fi
> /sbin/chkconfig --del udev-post >/dev/null 2>&1 || :
> exit 0
> 
> %pre
> # kill daemon if we are not in a chroot
> if [ -f /proc/1/exe -a -d /proc/1/root ] && [ / -ef /proc/1/root ]; then
>    if test -x /sbin/udevd -a -x /sbin/pidof ; then
>       pid=$(/sbin/pidof -c udevd)
>       while [ -n "$pid" ]; do
>                   for p in $pid; do
>               kill $hard $p 2>/dev/null
>             done
>             sleep 0.2
>             pid=$(/sbin/pidof -c udevd)
>             hard="-9"
>       done
>    fi
284,294c309,311
< if [ "$1" -ge "1" -a -x /sbin/pidof ]; then
<       pid=$(/sbin/pidof -c udevd)
<       while [ -n "$pid" ]; do
<               for p in $pid; do
<                       kill $hard $p;
<               done
<               sleep 0.2
<               pid=$(/sbin/pidof -c udevd)
<               hard="-9"
<       done
<       /sbin/udevd -d
---
> # start daemon if we are not in a chroot
> if [ -f /proc/1/exe -a -d /proc/1/root ] && [ / -ef /proc/1/root ]; then
>         /sbin/udevd -d >/dev/null 2>&1 || :


however, the fix doesn't do what the bug/fix description says - it doesn't detect trick like "chroot /" ... but quick googling says there's no safe way to detect chroot and comparing inodes seems to be the best guess - at least it is good enough to workaround the problem

just to demonstrate ...

.live.[root@ppcp-5s-m1 /]# chroot chroot /bin/bash -c 'if [ -f /proc/1/exe -a -d /proc/1/root ] && [ / -ef /proc/1/root ]; then echo "not chroot" ; else echo "chroot"; fi'
chroot
.live.[root@ppcp-5s-m1 /]# chroot / /bin/bash -c 'if [ -f /proc/1/exe -a -d /proc/1/root ] && [ / -ef /proc/1/root ]; then echo "not chroot" ; else echo "chroot"; fi'
not chroot

(where ./chroot contains minimal system to run bash)

    Technical note added. If any revisions are required, please edit the "Technical Notes" field
    accordingly. All revisions will be proofread by the Engineering Content Services team.
    
    New Contents:
Cause

The user runs fsck.gfs2 on a GFS2 file system in verbose mode (with the -v parameter).

Consequence

The user sees alarming error messages indicating that the master and root inode are not marked "In use" by the file system and an indication that the problem has been corrected:

System inode for 'master' is located at block 23 (0x17)
The inode exists but the block is not marked 'in use'; fixing it.
System inode for 'root' is located at block 22 (0x16)
The inode exists but the block is not marked 'in use'; fixing it.

There is no problem with either inode; the error messages are just misleading.

Fix

The code has been changed to set both the master and root inodes as "in use" in the in-core block map.

Result

As a result, fsck.gfs2 now realizes the master and root inodes are properly marked, so the error messages are not printed.

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2011-1046.html