Re: [PATCH] iomap: Make sure iomap_end is called after iomap_begin

[Matthew Wilcox <willy@infradead.org>]

On Thu, Jun 18, 2020 at 06:56:39AM -0700, Christoph Hellwig wrote:
> On Thu, Jun 18, 2020 at 02:37:37PM +0200, Andreas Gruenbacher wrote:
> > On Thu, Jun 18, 2020 at 2:32 PM Matthew Wilcox <willy@infradead.org> wrote:
> > > On Wed, Jun 17, 2020 at 06:39:01PM -0700, Darrick J. Wong wrote:
> > > > > -   if (WARN_ON(iomap.offset > pos))
> > > > > -           return -EIO;
> > > > > -   if (WARN_ON(iomap.length == 0))
> > > > > -           return -EIO;
> > > > > +   if (WARN_ON(iomap.offset > pos) || WARN_ON(iomap.length == 0)) {
> > > >
> > > > Why combine these WARN_ON?  Before, you could distinguish between your
> > > > iomap_begin method returning zero length vs. bad offset.
> > >
> > > Does it matter?  They're both the same problem -- the filesystem has
> > > returned an invalid iomap.  I'd go further and combine the two:
> > >
> > >         if (WARN_ON(iomap.offset > pos || iomap.length == 0)) {
> > >
> > > that'll save a few bytes of .text
> > 
> > That would be fine by me as well. Christoph may have wanted separate
> > warnings for a particular reason though.
> 
> Yes.  The line number in the WARN_ON will tell you which condition
> you if they are separate, which is really useful to diagnose what is
> going on.

Thinking about it, wouldn't the second test be better replaced with:

	if (WARN_ON(iomap.offset + iomap.length <= pos))

in case the filesystem returns an extent which finishes before pos?
This would be a superset of the test for length being 0.