On 1/9/23 5:38 PM, Jens Axboe wrote: > On 1/9/23 5:32?PM, Damien Le Moal wrote: >> It may be way simpler to rely on: >> 1) none scheduler >> 2) some light re-ordering of write requests in the driver itself to avoid >> any requeue to higher level (essentially, handle requeueing in the driver >> itself). > > Let's not start adding requeueing back into the drivers, that was a > giant mess that we tried to get out of for a while. > > And all of this strict ordering mess really makes me think that we > should just have a special scheduler mandated for devices that have that > kind of stupid constraints... Or, probably better, a stacked scheduler where the bottom one can be zone away. Then we can get rid of littering the entire stack and IO schedulers with silly blk_queue_pipeline_zoned_writes() or blk_is_zoned_write() etc. It really is a mess right now... -- Jens Axboe
