Hi, The 1st patch kills the atomic host-wide counter of host_busy. The 2nd patch bypass the atomic LUN-wide connter of device_busy for fast SSD device. V3: - use non-atomic set/clear bit operations as suggested by Bart - kill single field struct for storing count of in-flight requests - add patch to bypass the atomic LUN-wide counter of device_busy for fast SSD device V2: - introduce SCMD_STATE_INFLIGHT for getting accurate host busy via blk_mq_tagset_busy_iter() - verified that original Jens's report[1] is fixed - verified that SCSI timeout/abort works fine Ming Lei (2): scsi: core: avoid host-wide host_busy counter for scsi_mq scsi: core: don't limit per-LUN queue depth for SSD drivers/scsi/hosts.c | 19 +++++++++++++- drivers/scsi/scsi.c | 2 +- drivers/scsi/scsi_lib.c | 57 +++++++++++++++++++++++----------------- drivers/scsi/scsi_priv.h | 2 +- include/scsi/scsi_cmnd.h | 1 + include/scsi/scsi_host.h | 1 - 6 files changed, 54 insertions(+), 28 deletions(-) Cc: Jens Axboe <axboe@xxxxxxxxx> Cc: Ewan D. Milne <emilne@xxxxxxxxxx> Cc: Omar Sandoval <osandov@xxxxxx>, Cc: "Martin K. Petersen" <martin.petersen@xxxxxxxxxx>, Cc: James Bottomley <james.bottomley@xxxxxxxxxxxxxxxxxxxxx>, Cc: Christoph Hellwig <hch@xxxxxx>, Cc: Kashyap Desai <kashyap.desai@xxxxxxxxxxxx> Cc: Hannes Reinecke <hare@xxxxxxx> Cc: Laurence Oberman <loberman@xxxxxxxxxx> Cc: Bart Van Assche <bart.vanassche@xxxxxxx> -- 2.20.1
