linux-nvme.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: hch@lst.de (Christoph Hellwig)
Subject: [PATCH 1/9] nvme: don't hold nvmf_transports_rwsem for more than transport lookups
Date: Fri,  1 Jun 2018 09:11:20 +0200	[thread overview]
Message-ID: <20180601071128.7630-2-hch@lst.de> (raw)
In-Reply-To: <20180601071128.7630-1-hch@lst.de>

From: Johannes Thumshirn <jthumshirn@suse.de>

Only take nvmf_transports_rwsem when doing a lookup of registered
transports, so that a blocking ->create_ctrl doesn't prevent other
actions on /dev/nvme-fabrics.

Signed-off-by: Johannes Thumshirn <jthumshirn at suse.de>
[hch: increased lock hold time a bit to be safe, added a comment
 and updated the changelog]
Signed-off-by: Christoph Hellwig <hch at lst.de>
---
 drivers/nvme/host/fabrics.c | 3 ++-
 drivers/nvme/host/fabrics.h | 3 +++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/nvme/host/fabrics.c b/drivers/nvme/host/fabrics.c
index 5f5f7067c41d..fa32c1216409 100644
--- a/drivers/nvme/host/fabrics.c
+++ b/drivers/nvme/host/fabrics.c
@@ -952,6 +952,7 @@ nvmf_create_ctrl(struct device *dev, const char *buf, size_t count)
 		ret = -EBUSY;
 		goto out_unlock;
 	}
+	up_read(&nvmf_transports_rwsem);
 
 	ret = nvmf_check_required_opts(opts, ops->required_opts);
 	if (ret)
@@ -968,11 +969,11 @@ nvmf_create_ctrl(struct device *dev, const char *buf, size_t count)
 	}
 
 	module_put(ops->module);
-	up_read(&nvmf_transports_rwsem);
 	return ctrl;
 
 out_module_put:
 	module_put(ops->module);
+	goto out_free_opts;
 out_unlock:
 	up_read(&nvmf_transports_rwsem);
 out_free_opts:
diff --git a/drivers/nvme/host/fabrics.h b/drivers/nvme/host/fabrics.h
index 0cf0460a5c92..7491a0bbf711 100644
--- a/drivers/nvme/host/fabrics.h
+++ b/drivers/nvme/host/fabrics.h
@@ -124,6 +124,9 @@ struct nvmf_ctrl_options {
  *	1. At minimum, 'required_opts' and 'allowed_opts' should
  *	   be set to the same enum parsing options defined earlier.
  *	2. create_ctrl() must be defined (even if it does nothing)
+ *	3. struct nvmf_transport_ops must be statically allocated in the
+ *	   modules .bss section so that a pure module_get on @module
+ *	   prevents the memory from beeing freed.
  */
 struct nvmf_transport_ops {
 	struct list_head	entry;
-- 
2.17.0

  reply	other threads:[~2018-06-01  7:11 UTC|newest]

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-01  7:11 draft ANA support v2 Christoph Hellwig
2018-06-01  7:11 ` Christoph Hellwig [this message]
2018-06-03 12:26   ` [PATCH 1/9] nvme: don't hold nvmf_transports_rwsem for more than transport lookups Sagi Grimberg
2018-06-04  6:30   ` Hannes Reinecke
2018-06-01  7:11 ` [PATCH 2/9] nvme.h: add ANA definitions Christoph Hellwig
2018-06-03 12:27   ` Sagi Grimberg
2018-06-04  6:30   ` Hannes Reinecke
2018-06-04 11:58   ` Johannes Thumshirn
2018-06-01  7:11 ` [PATCH 3/9] nvme: add support for the log specific field Christoph Hellwig
2018-06-03 12:27   ` Sagi Grimberg
2018-06-04  6:31   ` Hannes Reinecke
2018-06-04 11:59   ` Johannes Thumshirn
2018-06-01  7:11 ` [PATCH 4/9] nvme: always failover on path or transport errors Christoph Hellwig
2018-06-01 15:27   ` Mike Snitzer
2018-06-04  6:29     ` Christoph Hellwig
2018-06-04 12:01       ` Johannes Thumshirn
2018-06-04 12:23       ` Mike Snitzer
2018-06-03 12:28   ` Sagi Grimberg
2018-06-04  6:32   ` Hannes Reinecke
2018-06-01  7:11 ` [PATCH 5/9] nvme: add ANA support Christoph Hellwig
2018-06-04  6:36   ` Hannes Reinecke
2018-06-04  7:03     ` Christoph Hellwig
2018-06-04  9:51       ` Hannes Reinecke
2018-06-04 12:31         ` Mike Snitzer
2018-06-04 13:37           ` Hannes Reinecke
2018-06-06 12:01   ` Popuri, Sriram
2018-06-06 12:13     ` Christoph Hellwig
2018-06-06 12:27       ` Popuri, Sriram
2018-06-06 12:50         ` Christoph Hellwig
2018-06-01  7:11 ` [PATCH 6/9] nvmet: track and limit the number of namespaces per subsystem Christoph Hellwig
2018-06-04  6:35   ` Hannes Reinecke
2018-06-04 12:04   ` Johannes Thumshirn
2018-06-01  7:11 ` [PATCH 7/9] nvmet: add minimal ANA support Christoph Hellwig
2018-06-04  6:41   ` Hannes Reinecke
2018-06-04 12:25   ` Johannes Thumshirn
2018-06-04 12:55     ` Christoph Hellwig
2018-06-01  7:11 ` [PATCH 8/9] nvmet: support configuring additional ANA groups Christoph Hellwig
2018-06-04  6:42   ` Hannes Reinecke
2018-06-04  6:53     ` Christoph Hellwig
2018-06-04  9:48       ` Hannes Reinecke
2018-06-01  7:11 ` [PATCH 9/9] nvmet: make ANATT configurable Christoph Hellwig
2018-06-04 10:21   ` Hannes Reinecke
2018-06-04 12:11     ` Christoph Hellwig

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=20180601071128.7630-2-hch@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).