On Fri, Feb 21, 2020 at 08:19:43AM -0800, Darrick J. Wong wrote: > Ok, we were doing it wrong. Should this series have fixed that as the > first patch (so that we could push it into old kernels) followed by the > actual icdinode field removal? Maybe I could squeeze that in after patch 1. Before that we'd get weird mismatch, which is how I arrived at the current series. > (Granted nobody seems to have complained...) Right now the use case seems entirely theoretical, but there are multiple series out on fsdevel that aim to change that.
