RE: getdents64 problem in 2.6.23

["Joakim Tjernlund" <joakim.tjernlund@transmode.se>]

> -----Original Message-----
> From: linux-mtd-bounces@lists.infradead.org 
> [mailto:linux-mtd-bounces@lists.infradead.org] On Behalf Of 
> Joakim Tjernlund
> Sent: den 30 oktober 2007 00:42
> To: David Woodhouse
> Cc: 'Linux-MTD Mailing List'; 'Jörn Engel'; 'Jamie Lokier'
> Subject: RE: getdents64 problem in 2.6.23
> 
> On Mon, 2007-10-29 at 18:30 +0100, Joakim Tjernlund wrote:
> > On Mon, 2007-10-29 at 13:58 +0100, Joakim Tjernlund wrote:
> > > On Sun, 2007-10-28 at 20:26 -0400, David Woodhouse wrote:
> > > > On Mon, 2007-10-29 at 00:58 +0100, Joakim Tjernlund wrote:
> > > > > David, before rewriting getdents et. all may I ask 
> you to apply the
> > > > > patch I sent today? Then we will have something that 
> works and go
> > > > > from there.
> > > > 
> > > > I'm not happy about having fd->raw still pointing to 
> nodes which may no
> > > > longer exist. Can't we set it to NULL (and cope with 
> potential fallout)?
> > > 
> > > Thats a diffrent story. Now I fixed the bug so it behaves 
> like the 
> > > !mark_node_obsolete case. Walking through all ->raw references is
> > > something I will save for a rainy dag, I need to fix a 
> few other bugs in
> > > our system ATM.
> > 
> > ehh, my patch may be somewhat wrong, need to reflect a bit 
> on that. I
> > might even get to the ->raw = NULL thing :)
> 
> ok, this should do it I hope. I also have ->raw = NULL in 
> progess, but I
> would like to know if this patch is accpetable to solve this problem.
> 
>  Jocke

David, just sent 2 patches to the list to do your bidding :)
I am done now unless you have something more to add.

   Jocke