On 8/28/18 11:54 AM, Omar Sandoval wrote: > On Mon, Aug 27, 2018 at 08:52:34AM -0600, Jens Axboe wrote: >> On 8/27/18 1:24 AM, Ming Lei wrote: >>> Hi, >>> >>> There is fundamental issue in commit 328728630d9f2bf1 (scsi: core: avoid >>> host-wide host_busy counter for scsi_mq) because SCSI's host busy counter >>> may not be same with counter of blk-mq's inflight tags, especially in case >>> of none io scheduler. >>> >>> So please revert the two commits. >> >> Please do, they break changing write cache mode on my SATA drives. You >> can add my tested-by and/or reported-by to this. > > Jens, how does it break? Could we add a blktest for it? Yep, it's pretty easy: # echo "write through" > /sys/block/sde/device/scsi_disk/4:0:0:0/cache_type and it hangs. For a blktests test, probably check the current value and ensure we set it to a different value, then restore it. -- Jens Axboe
