On Wed, Jun 19, 2024 at 08:59:33AM +0100, John Garry wrote: > In this case, I would expect NOIOB >= atomic write boundary. > > Would it be sane to have a NOIOB < atomic write boundary in some other > config? > > I can support these possibilities, but the code will just get more complex. I'd be tempted to simply not support the case where NOIOB is not a multiple of the atomic write boundary for now and disable atomic writes with a big fat warning (and a good comment in the soure code). If users show up with a device that hits this and want to use atomic writes we can resolved it.