On Tue, Jun 10, 2025 at 11:23:27AM -0700, Joanne Koong wrote: > count is unused by ->writeback_folio(), so it's always just zero. But > I see what you're saying now. I should just increment count after > doing the ->writeback_folio() call and then we could just leave the > "if (!count)" check untouched. I like this suggestion a lot, i'll make > this change in v2. I think you'll need to add count. Not only to support the block mappings through the common interface, but also to deal with the case where due to races with e.g. truncate we end up not kicking off any writeback at all. In that case the folio_end_writeback here needs to happen.
