On Wed, 9 Jan 2019 at 19:44, Mark Syms <mark.s...@citrix.com> wrote: > > So, actually the original comparison in the assert (dodgy scaling factor not > withstanding) was probably correct in that we don't ever want to remove all > blocks from the inode as one of them is used for the inode itself? Or do we > still think it should allow for change to be the negative of current blocks?
It doesn't make a difference; all we care about is that we don't go negative. I think the updated comparison better reflects that goal. Andreas