On Fri, 22 Mar 2024, Ming Lei wrote: > If logical block becomes mismatched by reconfiguration, the whole DM stack can't work: > > - at the beginning, DM is over NVMe(512 bs), DM & NVMe lbs is 512 > - later, nvme is reconfigured and its lbs becomes 4k, but DM's lbs can't > be updated > - then unaligned IO is submitted to NVMe > > So DM _never_ works with mis-matched logical block size because of > reconfigure, and same with MD. It can work. If a filesystem has 4k block size and it is used on a block device with misconfigured with 512-byte sector size, it will work. The buffer cache also prefers using page-sized blocks rather than logical block size, so it will work too. > > But yes, it works with 64K PAGE_SIZE. > > Fine, but I still think PAGE_SIZE is hard to follow than logical block > size. > > Thanks, > Ming The temporary buffer has one page, so I restricted the alignment to PAGE_SIZE. logical block size is always <= PAGE_SIZE. Mikulas
