From: hch@lst.de (Christoph Hellwig)
Subject: [PATCH 1/4] nvme-fabrics: refactor queue ready check
Date: Wed, 13 Jun 2018 10:16:16 +0200 [thread overview]
Message-ID: <20180613081616.GA23377@lst.de> (raw)
In-Reply-To: <99751a08-8485-787f-95bc-862b8f9d02ed@broadcom.com>
On Tue, Jun 12, 2018@02:45:16PM -0700, James Smart wrote:
> Additionally, if a user-mode delete request is made while in the middle of
> a RECONNECT, and the io the if_ready is looking at is part of the reconnect
> - we're allowing the connect command to go out/complete even though we're
> in the middle of deleting it.?? May well be fine - but I don't know why
> we're letting all this io through simply because we didn't want to add a
> check for a property_set() on a reset or delete.
so what about something like this on top of the series (untested so far):
diff --git a/drivers/nvme/host/fabrics.c b/drivers/nvme/host/fabrics.c
index dc87719a5c9e..a36a74b2cb8e 100644
--- a/drivers/nvme/host/fabrics.c
+++ b/drivers/nvme/host/fabrics.c
@@ -556,24 +556,31 @@ EXPORT_SYMBOL_GPL(nvmf_fail_nonready_command);
bool __nvmf_check_ready(struct nvme_ctrl *ctrl, struct request *rq,
bool queue_live)
{
- if (ctrl->state == NVME_CTRL_DEAD)
- return false;
+ struct nvme_request *req = nvme_req(rq);
/*
- * If the queue is not live the only thing allowed is a connect
- * command to actually set the queue live.
+ * If we are in some state of setup or teardown only allow
+ * internally generated commands.
*/
- if (!queue_live &&
- (nvme_req(rq)->cmd->common.opcode != nvme_fabrics_command ||
- nvme_req(rq)->cmd->fabrics.fctype != nvme_fabrics_type_connect))
+ if (!blk_rq_is_passthrough(rq) || (req->flags & NVME_REQ_USERCMD))
return false;
/*
- * If we are in some state of setup or teardown only allow
- * internally generated commands.
+ * Only allow commands on a live queue, except for the connect command,
+ * which is require to set the queue live in the appropinquate states.
*/
- return blk_rq_is_passthrough(rq) &&
- !(nvme_req(rq)->flags & NVME_REQ_USERCMD);
+ switch (ctrl->state) {
+ case NVME_CTRL_DEAD:
+ return false;
+ case NVME_CTRL_NEW:
+ case NVME_CTRL_CONNECTING:
+ if (req->cmd->common.opcode == nvme_fabrics_command &&
+ req->cmd->fabrics.fctype == nvme_fabrics_type_connect)
+ return true;
+ /*FALLTHRU*/
+ default:
+ return queue_live;
+ }
}
EXPORT_SYMBOL_GPL(__nvmf_check_ready);
next prev parent reply other threads:[~2018-06-13 8:16 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-11 15:46 queue ready fixes and cleanups Christoph Hellwig
2018-06-11 15:46 ` [PATCH 1/4] nvme-fabrics: refactor queue ready check Christoph Hellwig
2018-06-12 8:23 ` Johannes Thumshirn
2018-06-12 21:45 ` James Smart
2018-06-13 8:16 ` Christoph Hellwig [this message]
2018-06-13 15:54 ` James Smart
2018-06-14 12:19 ` Christoph Hellwig
2018-06-13 19:48 ` James Smart
2018-06-11 15:46 ` [PATCH 2/4] nvme-fabrics: handle queue ready fast path inline Christoph Hellwig
2018-06-12 8:24 ` Johannes Thumshirn
2018-06-12 21:47 ` James Smart
2018-06-11 15:46 ` [PATCH 3/4] nvme-fabrics: handle the admin-only case properly in nvmf_check_ready Christoph Hellwig
2018-06-12 8:25 ` Johannes Thumshirn
2018-06-12 21:48 ` James Smart
2018-06-13 7:59 ` Christoph Hellwig
2018-06-11 15:46 ` [PATCH 4/4] nvme-fabrics: reverse polarity in __nvmf_check_ready Christoph Hellwig
2018-06-12 8:26 ` Johannes Thumshirn
2018-06-12 11:08 ` Hannes Reinecke
2018-06-12 22:22 ` James Smart
2018-06-12 22:05 ` James Smart
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=20180613081616.GA23377@lst.de \
--to=hch@lst.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.