Before we either iterate on tags we must guarantee that the
hw queues are stopped and no inflight .queue_rq is active.
Thats what blk_mq_quiesce_queue is for, so use it where appropriate.
Note that before we cleanup the request queue, we must unquiesce in
order to guarantee to not block on requests queued during quiescing.
Changes from v2:
- removed admin_q requeue list kick altogether (suggested by Ming)
Changes from v1:
- dropped the xen_blockfront patch, not sure if it was helpful at all
- removed the requeue kick in nvme drivers while making nvme unconditionally
kick the requeue list when requeuing a request instead of doing so when
unquiescing the queues. Note that this change was a two step change to
keep bisection possible.
- fixed possible hangs with unquiescing before cleanup of the request queue
in mtip32xx and nvme drivers.
- added check that the request started in mtip32xx itration callouts.
- collected review tags.
*** SUBJECT HERE ***
*** BLURB HERE ***
Sagi Grimberg (8):
nvme-rdma: quiesce/unquiesce admin_q instead of start/stop its hw
queues
nvme-fc: quiesce/unquiesce admin_q instead of start/stop its hw queues
nvme-loop: quiesce/unquiesce admin_q instead of start/stop its hw
queues
nvme-pci: quiesce/unquiesce admin_q instead of start/stop its hw
queues
nvme: kick requeue list when requeueing a request instead of when
starting the queues
nbd: quiesce request queues to make sure no submissions are inflight
mtip32xx: quiesce request queues to make sure no submissions are
inflight
virtio_blk: quiesce/unquiesce live IO when entering PM states
drivers/block/mtip32xx/mtip32xx.c | 19 +++++++++++++++----
drivers/block/nbd.c | 4 ++--
drivers/block/virtio_blk.c | 4 ++--
drivers/nvme/host/core.c | 19 ++-----------------
drivers/nvme/host/fc.c | 5 +++--
drivers/nvme/host/pci.c | 6 +++---
drivers/nvme/host/rdma.c | 7 ++++---
drivers/nvme/target/loop.c | 3 ++-
8 files changed, 33 insertions(+), 34 deletions(-)
--
2.7.4
