Re: [reiserfs-dev] Re: [PATCH] sparc32: wrong type of nlink_t

[Chris Mason <mason@suse.com>]

On Thu, 2002-09-05 at 10:17, Oleg Drokin wrote:
> Hello!
> 
> On Thu, Sep 05, 2002 at 10:03:44AM -0400, Chris Mason wrote:
> 
> > read the -noleaf description on the find man page to see why we need to
> > set the directory link count to 1 when we are lying to userspace about
> > the actual link count on directories. 
> 
> There is nothing about nlink == 1 means assume -noleaf, so it should not work
> with old way too, right? Have anybody verified? ;)

I remember that happening during the initial discussions for the link
patch.  1 was chosen as the best way to do it, since it was a flag to
various programs that the unix directory link convention was not being
followed.

> 
> Actually patch might be easily modified to represent i_nlink == 1 for
> large directories, but still maintain correct on-disk nlink count.

Right.

> (and show maximal possible nlink count for regular files. Hm,
> I wonder if tar and stuff would break if met with file that have
> 67000 hardlinks ;) ).

Certainly seems like it would on sparc at least ;-)

-chris