Re: Duplicate slice entries in /dev

From: Dimitry Andric <dimitry_at_andric.com>
Date: Fri, 13 Feb 2009 17:12:59 +0100
On 2009-02-13 16:01, John Baldwin wrote:
>>> Any chance you might have both GEOM_MBR and GEOM_PART_MBR in your
>>> kernel configuration?
>> I have both GEOM_PART_GPT and GEOM_MBR, but not GEOM_PART_MBR.
> Since GEOM_PART_MBR is in DEFAULTS you effectively do have both of them in 
> there, however.

Hmm, on my -CURRENT test box, I get this problem with duplicated /dev
entries when I kldload GEOM_BSD (while GEOM_PART_BSD is already
statically linked into the kernel).  It seems GEOM_MBR isn't even
needed:

$ ls -l /dev/ad0*
crw-r-----  1 root  operator    0,  81 Feb 13 17:05 /dev/ad0
crw-r-----  1 root  operator    0,  84 Feb 13 17:05 /dev/ad0a
crw-r-----  1 root  operator    0,  84 Feb 13 17:05 /dev/ad0a
crw-r-----  1 root  operator    0,  92 Feb 13 17:09 /dev/ad0aa
crw-r-----  1 root  operator    0,  93 Feb 13 17:09 /dev/ad0ab
crw-r-----  1 root  operator    0,  94 Feb 13 17:09 /dev/ad0ac
crw-r-----  1 root  operator    0,  85 Feb 13 17:05 /dev/ad0b
crw-r-----  1 root  operator    0,  85 Feb 13 17:05 /dev/ad0b
crw-r-----  1 root  operator    0,  97 Feb 13 17:09 /dev/ad0c

Especially the ad0a[a-c] entries are funny, it seems to find a disklabel
within a disklabel. :)

Probably each of the geom modules registers the device(s) it found with
devd, causing nodes to appear in devfs.

Maybe these geom modules could be exclusionary, e.g. GEOM_PART_BSD
should refuse to load, if GEOM_BSD is loaded, and vice versa?  (And
possibly the same for GEOM_PART_MBR and GEOM_MBR.)
Received on Fri Feb 13 2009 - 15:13:01 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:42 UTC