On Wednesday 06 September 2006 12:38, Wookey wrote:
> On 06-09-06 11:36 +1200, Charles Manning wrote:
> > Folks
> >
> > Sorry I thought I had checked in a fix, but I had not.
> >
> > It is now sorted in cvs
> >
> > http://www.aleph1.co.uk/cgi-bin/viewcvs.cgi/yaffs2/yaffs_guts.c?r1=1.35&r
> >2=1.36
> >
> >
> > The problem that was occurring would only happen if you did a
> > modification like a chmod just before closing the file, then a reset. The
> > chmod would say that the file was no longer dirty (but it was, because
> > there's still dirty in the cache.
> >
> > The solution is to only mark the file clean if the cache is also empty. I
> > chose to do this rather than Artis' fix because that can result in far
> > more writes (when you might not be expecting them).
>
> So far as I cen see from preusing the code this fix needs to go in the
> yaffs1 codebase too. Is it in fact not necessary for some reason?
Yes, this should be applied to yaffs1 too.
yaffs1 fixes are limited to "broken stuff". Performance fixes only get applied
to yaffs2.
>
> I wonder if it also fixes Laurence Withers' example about renaming a
> file just before closing it in :
> http://aleph1.co.uk/lurker/list/message/20060817.102559.b5a65eed.en.html
>
> Renaming would alter the header only in the same way that the chmod
> would, and thus suffer the same problem with caching.
This could explain things and I'd like to see the results of running with this
new patch.
-- CHarles
(text/plain)