Didier:
> Le 26/06/2017 à 23:57, Steve Litt a écrit :
> > On Mon, 26 Jun 2017 12:25:40 +0200
> > Didier Kryn <k...@in2p3.fr> wrote:
> >> mdev is fine for a server. It is lacking a few features w/r udev:
> >>
> >> 1) It doesn't build /dev/disk/[by-id | by-label | by-partuuid |
> >> by-path | by-uuid]
> >>
> >> I tried to add this feature by the mean an mdev-called
> >> script ~ a year ago in a Busybox OS, and succeeded easily with
> >> by-label and by-uuid. I don't know the usage of the others anyway.
> > Could you please post that shellscript and a couple sentences about how
> > to install it?
> >
> Please find it in attachment. The archive contains four files:
>
> README
> etc/init.d/mdev -- The start/stop script for Sysvinit
> etc/mdev.conf -- The necessary configuration file for mdev
> sbin/mdev-disk -- An experimental script to manage /dev/disk/by-label
> -- and /dev/disk/by-uuid
>
> I might restart experimenting with it some day...
Hmm, something similar could be used to create thoose /dev/disk/by-*,
for a static /dev:
find /dev -type b | xargs blkid | sort | some_script_to_populate_by_xxx
But if that is only a transformed /sys/class/block, wouldn't it be
easier if there was a /sys/class/uuid or similar ?
Also man blkid says:
# man blkid | grep -A6 ' -L label$'
-L label
Look up the device that uses this label (equal to: -l -o device
-t LABEL=<label>). This lookup method is able to reliably use
/dev/disk/by-label udev symlinks (dependent on a setting in
/etc/blkid.conf). Avoid using the symlinks directly; it is not
reliable to use the symlinks without verification. The -L
option works on systems with and without udev.
so if /dev/disk/by-* isn't reliable, why do we have them ?
Regards,
/Karl Hammar
-----------------------------------------------------------------------
Aspö Data
Lilla Aspö 148
S-742 94 Östhammar
Sweden
+46 173 140 57
_______________________________________________
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
