On Wed, May 23, 2018 at 08:38:06AM +1000, Dave Chinner wrote: > Ok, I missed that detail as it's in a different patch. It looks like > if (pos > EOF) it will zeroed. But in this case I think that pos == > EOF and so it was reading instead. That smells like off-by-one bug > to me. This has been fixed in the tree I pushed yesterday already.
