On 3/29/19 8:34 AM, Martin K. Petersen wrote: > > Jens, > >> I told you this for the initial posting, and the objection still >> stands. Adding 40 bytes to struct bio is a no-go. >> >> So that's a big NAK on that series. > > I think you missed Bob's comment that this will go in the existing > bio_integrity field. I believe the main purpose of the series is to > solicit feedback on the callback approach. I didn't miss that, but it fixes nothing. That will unify the 40 bytes with 8 bytes, we're still growing the bio by a LOT. And we can't even nicely hide this behind some ifdef, sine distros enable everything and hence we're solving nothing by doing that. One potential solution could be to setup a specific bio_set for this, and allocate your read bios out of that. Growing struct bio by an enormous amount for this is just a non-starter to begin with, end of story. -- Jens Axboe
