xfs_chver(1M)xfs_chver(1M)NAMExfs_chver - change the version of a filesystem to enable the extent
unwritten version.
SYNOPSISxfs_chver [ -f ] [ -l logversion ] [ -c|+c ] [ -n ] -a
xfs_chver [ -f ] [ -l logversion ] [ -c|+c ] [ -n ] devices
DESCRIPTIONxfs_chver uses the xfs_db command to change the version of a filesystem
to enable the extent unwritten version. This version is needed to close
a security hole where a process can sometimes read old data off disk.
Using the -n option one can see the current version number and what the
new version number would be without having the change take effect.
Using the -l option one can change the log version to either version 1 or
version 2, where logversion is either 1 or 2 respectively.
Using the +c option one can enable a filesystem to use lazy counters.
Note that you must run xfs_repair(1M) after setting this option to build
the internal state that is required to support this functionality.
Using the -c option one can disable lazy counters if it is enabled. Note
that you must run xfs_repair(1M) after clearing this option to ensure
that the internal state of the filesystem is consistent.
The command must be run by the super-user. The command either does all
devices found in fstab(4) when the -a option is used or the given devices
from the second form above.
The command either does the changes or creates scripts which can later be
applied using the xfs_db command. If the -f option is used, scripts are
created and not applied. Without the -f option, the scripts are created,
applied using xfs_db, and removed.
Devices that are mounted cannot be converted. This is always true for
the root filesystem. Administrators must unmount filesystems before
xfs_chver or xfs_db can convert the version. If a device being converted
is found in mtab(4), the conversion is not attempted and an appropriate
message is printed. It is safe to run this command on mounted devices as
they are just skipped.
It is also safe to re-run the command multiple times on the same devices.
If the device is already converted or inaccessible, a simple message is
printed and the command continues with the next device.
The -f option causes the xfs_chver command to reduce error checking and
create file(s) which can be passed to xfs_db. These files are created in
the current directory in the form <device>.xfs_chver. These files can be
applied to the filesystem, invoke the following command:
Page 1
xfs_chver(1M)xfs_chver(1M)
xfs_db -x <device> < <device>.xfs_chver
Page 2
xfs_chver(1M)xfs_chver(1M)
When the -f options is given, the following conditions are not checked
for the device(s):
o - is the device the root filesystem,
o - is the device mounted,
o - can the user write to the device.
When the -f option is not used, the device is checked with /dev/root. If
a device has the same st_rdev, as the root device, i.e. /dev/root, the
conversion is not attempted and a message is printed. The root can only
be converted if another filesystem is used as a root and the "current
root" filesystem is accessed from the other root.
Another alternative to convert the root is to use the -f option. The
created file should be saved on the root to convert. Next, boot the
miniroot and mount the root to convert. Next, copy the file just created
with the -f option and the xfs_db command (usually found in
/usr/sbin/xfs_db) onto the mini root. Now, unmount the root to convert.
Then run the command:
xfs_db -x <device> < <script>
where <device> is the root device and <script> is the file created with
the -f option above.
EXIT STATUS
A status of 1 is returned if the usage is incorrect, e.g. -a is used with
a device name or an unrecognized parameter is detected. All other cases
return a status of 0, even if a filesystem is skipped.
BUGS
The xfs_db man page does not exist, yet.
cat <script> | xfs_db -x <device> does not work. Use xfs_db -x <device>
< <script>.
SEE ALSOxfs_db(1M), mstat(4), fstab(4), mkfs_xfs(1M), xfs_repair(1M), xfs(4).
Page 3
