9 a.m.
I've asked on IRC and did a google search but I cannot seem to find out how
rh/fedora uses partition labels.. What program do I use to view/set these
labels?
I've only heard about labels in BSD, so I'm kinda new to this and trying to
learn how rh/fedora does things..
TIA
-r
--
Rodney "meff" Gordon II -+- meff <at> pobox <dot> com
9:03 a.m.
# tune2fs -l /dev/xx
for view
# tune2fs -L NEW_LABEL /dev/xx
for change label
Rodney Gordon II wrote:
I've asked on IRC and did a google search but I cannot seem to
find out how
rh/fedora uses partition labels.. What program do I use to view/set these
labels?
I've only heard about labels in BSD, so I'm kinda new to this and trying to
learn how rh/fedora does things..
TIA
-r
9:42 a.m.
On Thu, Aug 05, 2004 at 04:03:48PM +0200, Alain Thiebauld wrote:
# tune2fs -l /dev/xx
for view
# tune2fs -L NEW_LABEL /dev/xx
for change label
So I assume it's an ext thing only? Not that I have other filesystems, just
wondering..
Thanks! :)
-r
--
Rodney "meff" Gordon II -+- meff <at> pobox <dot> com
9:43 a.m.
On Thu, 5 Aug 2004, Rodney Gordon II wrote:
On Thu, Aug 05, 2004 at 04:03:48PM +0200, Alain Thiebauld wrote:
> # tune2fs -l /dev/xx
> for view
>
>
> # tune2fs -L NEW_LABEL /dev/xx
> for change label
>
So I assume it's an ext thing only? Not that I have other
filesystems, just wondering..
there's also just "e2label".
rday
10:28 a.m.
Am Do, den 05.08.2004 schrieb Rodney Gordon II um 16:42:
So I assume it's an ext thing only? Not that I have other
filesystems, just
wondering..
-r
As far as I know those labels are not tied to any filesystem, nor
dependent on any. If you use i.e. Partition Magic you will find the
possibility to set a label, so does the old fdisk on DOS.
The Linux kernel requires BSD disk label support compiled in
(CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
as kernel command line in the grub.conf on Fedora (Redhat) systems.
Alexander
--
Alexander Dalloz | Enger, Germany | GPG key 1024D/ED695653 1999-07-13
Fedora GNU/Linux Core 2 (Tettnang) kernel 2.6.7-1.494.2.2smp
Serendipity 17:23:53 up 1 day, 10:51, load average: 0.44, 0.75, 0.95
1:22 p.m.
As far as I know those labels are not tied to any filesystem, nor
dependent on any. If you use i.e. Partition Magic you will find the
possibility to set a label, so does the old fdisk on DOS.
The Linux kernel requires BSD disk label support compiled in
(CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
as kernel command line in the grub.conf on Fedora (Redhat) systems.
While I think what you're saying is true (I don't use labeling at the
partition level) the labels that e2labl displays and modifies are
indeed an ext2/3 thing and apply to the filesystem, not the partition.
Devlabel, which I have not messed with, can supposedly be used to
label devices in a filesyste-independant manner by associating a
symlink with the device's UUID (see 'man devlabel' for more).
--Brad
1:23 p.m.
On Thu, 5 Aug 2004, Brad Smith wrote:
and someone else wrote:
> As far as I know those labels are not tied to any filesystem,
nor
> dependent on any. If you use i.e. Partition Magic you will find the
> possibility to set a label, so does the old fdisk on DOS.
>
> The Linux kernel requires BSD disk label support compiled in
> (CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
> as kernel command line in the grub.conf on Fedora (Redhat) systems.
really? i've used e2label many times for labelling filesystems, and
i've never built a kernel with BSD labelling. are you sure about
this? (time to go RTFS, i guess.)
rday
2:47 p.m.
Am Do, den 05.08.2004 schrieb Robert P. J. Day um 20:23:
>> The Linux kernel requires BSD disk label support compiled
in
>> (CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
>> as kernel command line in the grub.conf on Fedora (Redhat) systems.
really? i've used e2label many times for labelling filesystems, and
i've never built a kernel with BSD labelling. are you sure about
this? (time to go RTFS, i guess.)
rday
Hm, I am becoming a bit uncertain when you say that the labels in fstab
work for you on a custom kernel without BSD disklabel support and
because I can't find the source of my "knowledge". If it would be not a
big deal and not last that long, I would compile a kernel without
support for BSD disklabels and try it out. So for now I hope someone
with better and certain knowledge about the topic can state about it
without trial and error.
Alexander
--
Alexander Dalloz | Enger, Germany | GPG key 1024D/ED695653 1999-07-13
Fedora GNU/Linux Core 2 (Tettnang) kernel 2.6.7-1.494.2.2smp
Serendipity 21:42:24 up 1 day, 15:10, load average: 0.03, 0.21, 0.26
3:04 p.m.
On Thu, 5 Aug 2004, Alexander Dalloz wrote:
Am Do, den 05.08.2004 schrieb Robert P. J. Day um 20:23:
>>> The Linux kernel requires BSD disk label support compiled in
>>> (CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
>>> as kernel command line in the grub.conf on Fedora (Redhat) systems.
>
> really? i've used e2label many times for labelling filesystems, and
> i've never built a kernel with BSD labelling. are you sure about
> this? (time to go RTFS, i guess.)
>
> rday
Hm, I am becoming a bit uncertain when you say that the labels in fstab
work for you on a custom kernel without BSD disklabel support and
because I can't find the source of my "knowledge". If it would be not a
big deal and not last that long, I would compile a kernel without
support for BSD disklabels and try it out.
i don't have the resources to do that at this second, but i just took
a look at the Kconfig file in fs/partitions, and here's the entry for
that feature:
config BSD_DISKLABEL
bool "BSD disklabel (FreeBSD partition tables) support"
depends on PARTITION_ADVANCED && MSDOS_PARTITION
help
FreeBSD uses its own hard disk partition scheme on your PC. It
requires only one entry in the primary partition table of your disk
and manages it similarly to DOS extended partitions, putting in its
first sector a new partition table in BSD disklabel format. Saying Y
here allows you to read these disklabels and further mount FreeBSD
partitions from within Linux if you have also said Y to "UFS
file system support", above. If you don't know what all this is
about, say N.
and i know i've built lots of kernels without this feature, and done
lots of e2labels, and never had a problem. so if i've missed getting
burned by this all this time, i've been indescribably lucky.
rday
p.s. just making this as difficult as possible, rather than actually
test it, i took a look at the include file include/linux/ext2_fs.h in
the kernel source tree, where we find:
struct ext2_super_block {
__u32 s_inodes_count; /* Inodes count */
__u32 s_blocks_count; /* Blocks count */
...
char s_volume_name[16]; /* volume name */ <--- hmmm
...
and the man page for e2label:
... Ext2 filesystem labels can be at most 16 characters long; if
new-label is longer than 16 characters, e2label will truncate it and
print a warning message....
16 character limits in both places. coincidence? (yes, i'm clearly
bored today.)
3:36 p.m.
On Thu, 05 Aug 2004 21:47:47 +0200, Alexander Dalloz wrote:
Am Do, den 05.08.2004 schrieb Robert P. J. Day um 20:23:
> >> The Linux kernel requires BSD disk label support compiled in
> >> (CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
> >> as kernel command line in the grub.conf on Fedora (Redhat) systems.
>
> really? i've used e2label many times for labelling filesystems, and
> i've never built a kernel with BSD labelling. are you sure about
> this? (time to go RTFS, i guess.)
>
> rday
Hm, I am becoming a bit uncertain when you say that the labels in fstab
work for you on a custom kernel without BSD disklabel support and
because I can't find the source of my "knowledge". If it would be not a
big deal and not last that long, I would compile a kernel without
support for BSD disklabels and try it out. So for now I hope someone
with better and certain knowledge about the topic can state about it
without trial and error.
BSD disk labels are not the same than ext2/ext3 volume labels. Robert is
right. ext2/ext3 labels are a superblock thing (you can display the label
with "dumpe2fs", too).
3:33 p.m.
On Thu, 2004-08-05 at 16:24, netmask wrote:
Also, you must use initrd to use labels in fstab
This is not true. An initrd is only _required_ if you have not compiled
into the kernel a module necessary for boot.
> The Linux kernel requires BSD disk label support compiled in
> (CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
> as kernel command line in the grub.conf on Fedora (Redhat) systems.
--
C. Linus Hicks <lhicks(a)nc.rr.com>
6:38 p.m.
On 05 Aug 2004 16:33:50 -0400, C. Linus Hicks wrote:
On Thu, 2004-08-05 at 16:24, netmask wrote:
> Also, you must use initrd to use labels in fstab
This is not true. An initrd is only _required_ if you have not compiled
into the kernel a module necessary for boot.
Wrong. You can only mount partitions via their ext2/ext3 volume labels if
the initrd contains a mount command which understands the labels, e.g.
nash.
6:40 p.m.
On Fri, 6 Aug 2004, Michael Schwendt wrote:
On 05 Aug 2004 16:33:50 -0400, C. Linus Hicks wrote:
> On Thu, 2004-08-05 at 16:24, netmask wrote:
>> Also, you must use initrd to use labels in fstab
>
> This is not true. An initrd is only _required_ if you have not compiled
> into the kernel a module necessary for boot.
Wrong. You can only mount partitions via their ext2/ext3 volume labels if
the initrd contains a mount command which understands the labels, e.g.
nash.
only partially true. as i understand it (and i asked about this way
back in history once upon a time), you only need an initrd if you want
to mount the *root* filesystem by label.
the explanation i got back then was that initially mounting the root
filesystem is done by the kernel in kernel space, where there is no
knowledge of labels -- you have to use device names. once that's
done, the remaining mounts out of /etc/fstab are done in user space,
and labels are fine.
if you use an initrd, then the mounting of the root filesystem will
(somehow) be able to use labels -- perhaps the initrd forces the
mounting to happen in user space?
can someone clarify this?
rday
7:04 p.m.
only partially true. as i understand it (and i asked about this way
back in
history once upon a time), you only need an initrd if you want to mount the
*root* filesystem by label.
if you use an initrd, then the mounting of the root filesystem will (somehow)
be able to use labels -- perhaps the initrd forces the mounting to happen in
user space?
from man mkinitrd.. This leaves me to believe that yes, you /only/ need to use
initrd if your root partition is utilizing labels.. as is default in
redhat/fedora.
If a label is used, as in root=LABEL=rootPart the initrd will search all
available devices for an ext2 or ext3 filesystem with the appropriate label,
and mount that device as the root filesystem.
7:32 p.m.
On Thu, 5 Aug 2004 19:40:09 -0400 (EDT), Robert P. J. Day wrote:
On Fri, 6 Aug 2004, Michael Schwendt wrote:
> On 05 Aug 2004 16:33:50 -0400, C. Linus Hicks wrote:
>
>> On Thu, 2004-08-05 at 16:24, netmask wrote:
>>> Also, you must use initrd to use labels in fstab
>>
>> This is not true. An initrd is only _required_ if you have not compiled
>> into the kernel a module necessary for boot.
>
> Wrong. You can only mount partitions via their ext2/ext3 volume labels if
> the initrd contains a mount command which understands the labels, e.g.
> nash.
only partially true. as i understand it (and i asked about this way
back in history once upon a time), you only need an initrd if you want
to mount the *root* filesystem by label.
That's what I meant, root=LABEL=foo in kernel command-line. Sorry.
6:44 p.m.
Once upon a time, Michael Schwendt <fedora(a)wir-sind-cool.org> said:
On 05 Aug 2004 16:33:50 -0400, C. Linus Hicks wrote:
> This is not true. An initrd is only _required_ if you have not compiled
> into the kernel a module necessary for boot.
Wrong. You can only mount partitions via their ext2/ext3 volume labels if
the initrd contains a mount command which understands the labels, e.g.
nash.
Note that this only affects the root filesystem. All other filesystems
are mounted using /sbin/mount on the root filesystem.
--
Chris Adams <cmadams(a)hiwaay.net>
Systems and Network Administrator - HiWAAY Internet Services
I don't speak for anybody but myself - that's enough trouble.
6:51 p.m.
> Also, you must use initrd to use labels in fstab
This is not true. An initrd is only _required_ if you have not compiled
into the kernel a module necessary for boot.
I'm sorry, but that's just not the case.
You /must/ have a ram disk to use a label on your boot partition. Think about
it for a second, it makes perfect sense.
http://www.redhat.com/docs/manuals/enterprise/RHEL-3-Manual/sysadmin-guid...
If the system uses the ext3 file system, a SCSI controller, or uses labels to
reference partitions in /etc/fstab, an initial RAM disk is needed. The initial
RAM disk allows a modular kernel to have access to modules that it might need
to boot from before the kernel has access to the device where the modules
normally reside.
8:25 p.m.
On Thu, 2004-08-05 at 19:51, netmask wrote:
>> Also, you must use initrd to use labels in fstab
>
> This is not true. An initrd is only _required_ if you have not compiled
> into the kernel a module necessary for boot.
I'm sorry, but that's just not the case.
You /must/ have a ram disk to use a label on your boot partition. Think about
it for a second, it makes perfect sense.
http://www.redhat.com/docs/manuals/enterprise/RHEL-3-Manual/sysadmin-guid...
If the system uses the ext3 file system, a SCSI controller, or uses labels to
reference partitions in /etc/fstab, an initial RAM disk is needed. The initial
RAM disk allows a modular kernel to have access to modules that it might need
to boot from before the kernel has access to the device where the modules
normally reside.
Please note that part of what I said was "not compiled into the kernel a
module necessary for boot." Then you said "allows a modular kernel to
have access to modules that it might need to boot".
Compiling those modules into the kernel obviates the need for loading a
module. I regularly build custom kernels and never use an initrd with
them. I have never tried specifying a partition label for the boot, but
I know I have mounted partitions via labels with no initrd.
--
C. Linus Hicks <lhicks(a)nc.rr.com>
11:08 p.m.
>
> If the system uses the ext3 file system, a SCSI controller, or uses labels to
> reference partitions in /etc/fstab, an initial RAM disk is needed. The initial
> RAM disk allows a modular kernel to have access to modules that it might need
> to boot from before the kernel has access to the device where the modules
> normally reside.
Please note that part of what I said was "not compiled into the kernel a
module necessary for boot." Then you said "allows a modular kernel to
have access to modules that it might need to boot".
I didn't say anything.. that's from Redhat
Compiling those modules into the kernel obviates the need for loading a
module. I regularly build custom kernels and never use an initrd with
them. I have never tried specifying a partition label for the boot, but
I know I have mounted partitions via labels with no initrd.
That's very nice for you.. Many of us having been building "custom" (aka,
building..) kernels for our selves for many years as well.
Nobody is saying you need initrd for LABELS.. you need it for LABELS on your
boot partition. The logic is VERY simple to understand.
3:31 p.m.
On Thu, 2004-08-05 at 11:28, Alexander Dalloz wrote:
Am Do, den 05.08.2004 schrieb Rodney Gordon II um 16:42:
> So I assume it's an ext thing only? Not that I have other filesystems, just
> wondering..
> -r
As far as I know those labels are not tied to any filesystem, nor
dependent on any. If you use i.e. Partition Magic you will find the
possibility to set a label, so does the old fdisk on DOS.
The Linux kernel requires BSD disk label support compiled in
(CONFIG_BSD_DISKLABEL=y) to use the labels with fstab and how it is used
as kernel command line in the grub.conf on Fedora (Redhat) systems.
BSD labels have nothing to do with ext2/3 partition labels. They are a
completely different thing.
--
C. Linus Hicks <lhicks(a)nc.rr.com>
7230
days inactive
7231
days old
20 comments
9 participants
participants (9)
-
Alain Thiebauld -
Alexander Dalloz -
Brad Smith -
C. Linus Hicks -
Chris Adams -
Michael Schwendt -
netmask -
Robert P. J. Day -
Rodney Gordon II