Bug#269721: bug

[GOTO Masanori <gotom@debian.or.jp>]

reassign 269721 uclibc
thanks

At Sat, 28 May 2005 01:22:24 +0900,
GOTO Masanori wrote:
> > > >     - Also export __kernel_dev_t for util-linux/cryptoloop (Closes:
> > > >     220991).
> > > 
> > > This causes problems when building uclibc, since it defines
> > > __kernel_dev_t in its own headers.  I recently switched uclibc
> > > to use linux-kernel-headers instead of straight kernel headers
> > > for the same reason that glibc -- it's the Right Way To Do It.
> > > 
> > > This is wishlist, since I currently have a workaround in uclibc.
> > 
> > David, if so, then how to resolve util-linux report #220991?
> 
> I looked at uclibc, and I wondered __kernel_dev_t defines as unsigned
> short.  In 2.6 linux kernel adopts 32bit kernel dev_t.  Plus, it seems
> your hack is applied only for arm.  Why do we need to hide
> __kernel_dev_t from lkh headers?  I would like to hear your opinion.

As I explained before, I think it's not lkh bug.  I reassigned it to
uclibc source package.  David, if you think it should be fixed in lkh,
please reassign it back with the detailed information.  But actually
__kernel_dev_t is now 32bit in 2.6 kernel, so it's uclibc issue, IMHO.

Regards,
-- gotom