Re: 2.4 has O_SYNC bug ?

On Sat, Mar 22, 2003 at 02:37:53PM -0500, Doug McNaught wrote:
> Rob van Nieuwkerk <[e-mail]> writes:
> 
> > But the strange thing is this:  always after 30s the kernel performs
> > extra writes to the CF.  It seems it's flushing some kind of dirty buffer
> > from the buffer cache.  But there is not supposed to be any dirty buffer:
> > all data should have been written already to the CF because the partition
> > was opened with O_SYNC !
> 
> noatime?
> 
> -Doug

As he stated he was using no filesystem, but a treating the partition as
a raw file, I suppose it can't be atime-refreshing... perhaps using
O_DIRECT will solve it?

Greets, Antonio.

[sorry for double response Doug, I forgot to reply to list earlier]