On Fri, Feb 01, 2019 at 07:46:42AM -0500, Brian Foster wrote: > Unless I'm missing something where this code is required for correctness > or somehow provides a measureable performance benefit, I think we're > better off keeping this path as simple as possible. It should be fairly > easy to provide test coverage for the page granularity truncate handling > code (generic/524 may already cover this). I'm not so sure about that > for the block granularity handling without having some kind of > additional instrumentation. It should not be required for correctness. I've done a first round of testing with the check removed and it seems to be doing fine.
