On Fri, Jan 31, 2020 at 01:43:19PM +0000, Johannes Thumshirn wrote: > > I have a local version now, where btrfs_read_dev_one_super() uses > read_cache_page_gfp() and btrfs_scratch_superblocks() uses > write_one_page() offloading all the I/O to the pagecache. So far this > works as expected. > > Now when I started converting write_dev_supers() and friends I wasn't > sure if I can/should mix up read_cache_page_gfp() and submit_bio_wait(). > I could also convert it to write_one_page() but this way we would loose > integrity checking for the super block. > > Any advice? You can mix it, although it needs some big fat comments explaining what is going on there to the reader.
