From: hare@suse.de (Hannes Reinecke)
Subject: [PATCH 4/4] nvme: start ANATT timer on out-of-order state changes
Date: Thu, 7 Jun 2018 09:35:56 +0200 [thread overview]
Message-ID: <20180607073556.39050-5-hare@suse.de> (raw)
In-Reply-To: <20180607073556.39050-1-hare@suse.de>
If I/O is terminated due to an invalid ANA state we should be starting
the ANATT timer as we really should have received an ANA AEN signalling
the state change.
Signed-off-by: Hannes Reinecke <hare at suse.com>
---
drivers/nvme/host/multipath.c | 36 +++++++++++++++++++++++++++++-------
1 file changed, 29 insertions(+), 7 deletions(-)
diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c
index 7a37f5959ad4..7f34ae260ca9 100644
--- a/drivers/nvme/host/multipath.c
+++ b/drivers/nvme/host/multipath.c
@@ -49,7 +49,10 @@ static void nvme_update_ana_state(struct nvme_ns *ns, enum nvme_ana_state state)
void nvme_failover_req(struct request *req)
{
struct nvme_ns *ns = req->q->queuedata;
+ struct device *dev = disk_to_dev(ns->disk);
unsigned long flags;
+ enum nvme_ana_state ana_state;
+ bool ana_state_changed = false;
spin_lock_irqsave(&ns->head->requeue_lock, flags);
blk_steal_bios(&ns->head->requeue_list, req);
@@ -60,26 +63,45 @@ void nvme_failover_req(struct request *req)
* Reset the controller for any non-ANA error as we don't know what
* caused the error:
*/
+ ana_state = READ_ONCE(ns->ctrl->ana_state[ns->anagrpid]);
switch (nvme_req(req)->status & 0x7ff) {
case NVME_SC_ANA_TRANSITION:
- nvme_update_ana_state(ns, NVME_ANA_CHANGE);
- if (!timer_pending(&ns->anatt_timer)) {
- ns->anatt_timer.expires =
- ns->ctrl->anatt * HZ + jiffies;
- add_timer(&ns->anatt_timer);
+ if (ana_state != NVME_ANA_CHANGE) {
+ nvme_update_ana_state(ns, NVME_ANA_CHANGE);
+ ana_state_changed = true;
}
break;
case NVME_SC_ANA_PERSISTENT_LOSS:
- nvme_update_ana_state(ns, NVME_ANA_PERSISTENT_LOSS);
+ if (ana_state != NVME_ANA_PERSISTENT_LOSS) {
+ nvme_update_ana_state(ns, NVME_ANA_PERSISTENT_LOSS);
+ ana_state_changed = true;
+ }
break;
case NVME_SC_ANA_INACCESSIBLE:
- nvme_update_ana_state(ns, NVME_ANA_INACCESSIBLE);
+ if (ana_state != NVME_ANA_INACCESSIBLE) {
+ nvme_update_ana_state(ns, NVME_ANA_INACCESSIBLE);
+ ana_state_changed = true;
+ }
break;
default:
nvme_reset_ctrl(ns->ctrl);
break;
}
+ if (ana_state_changed) {
+ /*
+ * We should have received an ANA AEN signalling the state
+ * change. As we haven't (or we wouldn't ever reach here)
+ * give it a benefit of doubt to start the ANATT timer
+ * before resetting the controller.
+ */
+ if (!timer_pending(&ns->anatt_timer)) {
+ ns->anatt_timer.expires =
+ ns->ctrl->anatt * HZ + jiffies;
+ add_timer(&ns->anatt_timer);
+ }
+ }
+
kblockd_schedule_work(&ns->head->requeue_work);
}
--
2.12.3
next prev parent reply other threads:[~2018-06-07 7:35 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-07 7:35 [PATCH 0/4] nvme: ANATT handling Hannes Reinecke
2018-06-07 7:35 ` [PATCH 1/4] nvmet: make ANATT configurable Hannes Reinecke
2018-06-07 12:06 ` Christoph Hellwig
2018-06-07 12:42 ` Hannes Reinecke
2018-06-07 13:03 ` Sagi Grimberg
2018-06-07 7:35 ` [PATCH 2/4] nvmet: ANA transition timeout handling Hannes Reinecke
2018-06-07 12:07 ` Christoph Hellwig
2018-06-07 13:31 ` Hannes Reinecke
2018-06-07 13:41 ` Christoph Hellwig
2018-06-07 13:12 ` Sagi Grimberg
2018-06-07 7:35 ` [PATCH 3/4] nvme: " Hannes Reinecke
2018-06-07 12:09 ` Christoph Hellwig
2018-06-07 12:52 ` Hannes Reinecke
2018-06-07 13:11 ` Christoph Hellwig
2018-06-07 13:16 ` Sagi Grimberg
2018-06-07 13:26 ` Christoph Hellwig
2018-06-07 7:35 ` Hannes Reinecke [this message]
2018-06-07 12:11 ` [PATCH 4/4] nvme: start ANATT timer on out-of-order state changes Christoph Hellwig
2018-06-07 12:37 ` Hannes Reinecke
2018-06-07 13:10 ` Christoph Hellwig
2018-06-07 13:20 ` Hannes Reinecke
2018-06-07 13:46 ` Christoph Hellwig
2018-06-07 14:01 ` Hannes Reinecke
2018-06-07 14:22 ` Christoph Hellwig
2018-06-07 15:20 ` Sagi Grimberg
2018-06-07 13:20 ` Sagi Grimberg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180607073556.39050-5-hare@suse.de \
--to=hare@suse.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox