Re: /dev/dm-0

On Sun, Mar 26, 2006 at 12:24:56PM +0200, Luca Berra wrote:
>On Mon, Mar 20, 2006 at 04:21:39PM -0500, Peter Jones wrote:
>>On Mon, 2006-03-20 at 10:55 -0500, Phillip Susi wrote:
>>>I have /dev/mapper/via_hfciifae as the base raid device, and partitions 
>>>on the device with a number appended to the above.  I believe that the 
>>>dmraid utility creates devices with this style of name, and if you are 
>>>seeing /dev/dm-0 it is because udev is creating that node because as far 
>>>as the kernel knows, the name of the device ( as exported by sysfs ) is 
>>>"dm-0".  Udev probably shouldn't be creating device nodes for device 
>>>mapper devices, but if it does, you can simply ignore them. 
>>
>>Well, really we should fix the kernel, and make these /dev/dm-$DMNAME
>>(and thus /sys/block/dm-$DMNAME )
>
>maybe mapper/$DMNAME (and thus /sys/block/mapper!$DMNAME) so we are
>consistent with other drivers and udev will use /dev/mapper/$DMNAME
>
>this should be fairly easy to change,
>just add a name parameter to 
>
>dm_create(), dm_create_with_minor(), create_aux and alloc_dev() in
>drivers/md/dm.c
>
>and use that parameter instead of "dm-%s" when setting disk_name just
>before add_disk in alloc_dev()
>
>then pass param->name as the name parameter to dm_createXXXXX() in
>drivers/md/dm-ioctl.c:dev_create()
>
>if there are no objections i might even be tempted to send a patch.
>
oh, yes, we would also need to plug into the rename ioctl.
but i am not sure how this should be done.
i don't feel like unregister_blkdev and add again is a good idea.

L.

-- 
Luca Berra -- [email protected]
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \