In the current implementation, final zone-mgmt request is issued with submit_bio_wait() which marks the bio REQ_SYNC. This is needed since immediate action is expected for zone-mgmt requests as these are blocking operations. This also bypasses the scheduler in the blk_mq_make_request() and dispatches the request directly into the hw ctx. This patch marks all the chained bios REQ_SYNC so that we can have above-mentioned behavior for non-final bios also. Reviewed-by: Damien Le Moal <damien.lemoal@xxxxxxx> Reviewed-by: Bob Liu <bob.liu@xxxxxxxxxx> Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@xxxxxxx> --- * Changes from V1 :- 1. Update commit description (Jens). --- block/blk-zoned.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block/blk-zoned.c b/block/blk-zoned.c index d00fcfd71dfe..05741c6f618b 100644 --- a/block/blk-zoned.c +++ b/block/blk-zoned.c @@ -198,7 +198,7 @@ int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op, break; } - bio->bi_opf = op; + bio->bi_opf = op | REQ_SYNC; bio->bi_iter.bi_sector = sector; sector += zone_sectors; -- 2.22.1
