Re: [PATCH RFC] __bd_forget should wait for inodes using the mapping

[Chris Mason <mason@suse.com>]

On Fri, 2004-06-18 at 16:26, Andrew Morton wrote:
> Chris Mason <mason@suse.com> wrote:
> >
> > Maybe the real bug is the FS inode should never have ended up in the
> > dirty list.
> 
> It'd be interesting to find out where and why it is being dirtied (atime?),
> but even if we can prevent that from happening, people can still do things
> like chmod on it, so we're back in the same situation.
> 
> There's tight coupling between writing back the inode and writing back its
> pages, and at times it has caused problems.  It's not clear _why_ there
> should be such a coupling but it's never been a sufficient problem to
> justify ripping it all up.
> 
> >  This should all work fine if the bdev inode were the only
> > one to ever hit a dirty list.
> 
> Something like this?

[ skip writing block-special inodes ]

Hmmm, any risk in missing data integrity syncs because of this?

-chris