On Tue, Oct 31, 2023 at 10:02:42AM +0100, Christoph Hellwig wrote: > Can you also pick up: > > "xfs: only remap the written blocks in xfs_reflink_end_cow_extent" > > ? > > Also this seems to a bit of a mix of fixes for 6.7 and big stuff that > is too late for the merge window. If by 'big stuff' you mean the MF_MEM_PRE_REMOVE patch, then yes, I agree that it's too late to be changing code outside xfs. Bumping that to 6.8 will disappoint Shiyang, regrettably. The patchsets for realtime units refactoring and typechecked rt-helpers (except for the xfs_rtalloc_args thing) I'd prefer to land in 6.7 for a few reasons. First, the blast radii are contained to the rtalloc subsystem of xfs. Second, I've been testing them for nearly a year now, I think they're ready from a QA perspective. The third selfish reason for wanting to get the xfs realtime stuff off my plate is that my goal for 6.8 is to try to eliminate the indirect ->iomap_begin/end calls from iomap. It'll be helpful for me to be able to focus exclusively on that since I'd really like your help making sure I do the transition correctly. :) --D
