Re: [dm-devel] linux-next - WARNING: at fs/block_dev.c:824 bd_link_disk_holder+0x92/0x1ac()

[Kay Sievers <kay sievers vrfy org>]

On Fri, Jan 14, 2011 at 16:07, Karel Zak <kzak redhat com> wrote:
> On Thu, Jan 13, 2011 at 05:10:02PM +0100, Kay Sievers wrote:
>> On Thu, Jan 13, 2011 at 16:59, Karel Zak <kzak redhat com> wrote:
>> > On Thu, Jan 13, 2011 at 03:43:38PM +0100, Kay Sievers wrote:
>> >> On Thu, Jan 13, 2011 at 15:30, Tejun Heo <tj kernel org> wrote:
>> >> > On Thu, Jan 13, 2011 at 3:25 PM, Milan Broz <mbroz redhat com> wrote:
>> >> >> Maybe, but this was not invented in DM/MD camp:-)
>> >> >> Probably Kay or Greg can answer why it was done this way?
>> >>
>> >> It's not from Greg or Kay. It just appeared some day in the context of dm. :)
>> >>
>> >> And yes, symlinks *look* nice and simple for the outside, but they are
>> >> not, and have all sorts of problems like non-atomic updates, make it
>> >
>> >  Sounds like sysfs implementation problem, right?
>>
>> It's a normal multi-file problem. It can by-definition not be atomic
>> without doing really weird locking things.
>
>  BTW, lsblk(8) and libblkid don't depend on the fact that slaves/holders
>  files are symlinks.
>
>  The important thing is the filename (/sys/block/.../slaves/<name>)
>  only. We don't follow the symlinks and we don't use readlink() there.
>
>  It means that you can replace the symlinks with regular files where
>  in the file contents is for example maj:min, etc.

I don't think we really can change anything here, there are more users of it.

If we would go changing things here, the file names should never be
the device name but the format "b8:32" for blockdevs, "n22" for the
network ifindex, ... which is the unchangeable unique kernel id -- not
depending on any possible device rename, or fast destroy/re-create or
anything like that.

Kay