linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mike Snitzer <snitzer@redhat.com>
To: dm-devel@redhat.com
Cc: Kiyoshi Ueda <k-ueda@ct.jp.nec.com>,
	Nikanth Karthikesan <knikanth@suse.de>,
	linux-kernel@vger.kernel.org, Jens Axboe <jens.axboe@oracle.com>,
	"Jun'ichi Nomura" <j-nomura@ce.jp.nec.com>,
	Vivek Goyal <vgoyal@redhat.com>
Subject: [RFC PATCH 3/2] dm: bio-based device must not register elevator in sysfs
Date: Thu, 13 May 2010 01:02:53 -0400	[thread overview]
Message-ID: <20100513050252.GC25523@redhat.com> (raw)
In-Reply-To: <20100513043108.GB25523@redhat.com>

Properly clear io scheduler state from sysfs if transitioning from a
request-based table back to bio-based.  Unregister the elevator from
sysfs (removes 'iosched/') and have the 'scheduler' sysfs attribute
properly report "none" for bio-based DM.

Adjust elv_iosched_show() to return "none" if !blk_queue_stackable.
This allows the block layer to take bio-based DM into consideration.

These changes address the following (albeit obscure) corner-case:
   # dmsetup create --notable bio-based
   # echo "0 100 multipath ..." | dmsetup load bio-based
   # echo "0 100 linear ..."    | dmsetup load bio-based
   # dmsetup resume bio-based
   # ls /sys/block/dm-0/queue/
   ... iosched ...

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
---
 block/elevator.c      |    2 +-
 drivers/md/dm-table.c |    1 +
 drivers/md/dm.c       |   41 ++++++++++++++++++++++++++++++++++++++---
 drivers/md/dm.h       |    1 +
 4 files changed, 41 insertions(+), 4 deletions(-)

diff --git a/block/elevator.c b/block/elevator.c
index 6ec9137..168112e 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -1088,7 +1088,7 @@ ssize_t elv_iosched_show(struct request_queue *q, char *name)
 	struct elevator_type *__e;
 	int len = 0;
 
-	if (!q->elevator)
+	if (!q->elevator || !blk_queue_stackable(q))
 		return sprintf(name, "none\n");
 
 	elv = e->elevator_type;
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 64a8578..dec8295 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -803,6 +803,7 @@ int dm_table_set_type(struct dm_table *t)
 	if (bio_based) {
 		/* We must use this table as bio-based */
 		t->type = DM_TYPE_BIO_BASED;
+		dm_clear_request_based_queue(t->md);
 		return 0;
 	}
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index b2171be..08a4745 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1957,9 +1957,9 @@ bad_module_get:
 int dm_init_request_based_queue(struct mapped_device *md)
 {
 	struct request_queue *q = NULL;
+	int register_queue = 0;
 
-	if (!md->queue)
-		return 0;
+	BUG_ON(!md->queue);
 
 	/*
 	 * Avoid re-initializing the queue (and leaking the existing
@@ -1973,9 +1973,22 @@ int dm_init_request_based_queue(struct mapped_device *md)
 		md->queue = q;
 		md->saved_make_request_fn = md->queue->make_request_fn;
 		blk_queue_make_request(md->queue, dm_request);
-		elv_register_queue(md->queue);
+		register_queue = 1;
+	} else if (!md->queue->request_fn) {
+		/*
+		 * Handle corner-case where bio-based table was cleared
+		 * and requested-based table was loaded in its place.
+		 * - re-instate the ->request_fn that was cleared
+		 *   by dm_clear_request_based_queue().
+		 * - register the queue's 'iosched/' sysfs attributes
+		 */
+		md->queue->request_fn = dm_request_fn;
+		register_queue = 1;
 	}
 
+	if (register_queue)
+		elv_register_queue(md->queue);
+
 	/*
 	 * Request-based dm devices cannot be stacked on top of bio-based dm
 	 * devices.  The type of this dm device has not been decided yet.
@@ -1996,6 +2009,28 @@ int dm_init_request_based_queue(struct mapped_device *md)
 	return 1;
 }
 
+void dm_clear_request_based_queue(struct mapped_device *md)
+{
+	BUG_ON(!md->queue);
+
+	if (!md->queue->elevator)
+		return; /* already bio-based so return */
+
+	/*
+	 * Clear ->request_fn, treat NULL as a flag to indicate a
+	 * bio-based table was loaded after a request-based table.
+	 */
+	md->queue->request_fn = NULL;
+
+	/*
+	 * Unregister the request-based queue's elevator from sysfs.
+	 * NOTE: md->queue doesn't have the QUEUE_FLAG_STACKABLE flag
+	 * set so elv_iosched_show() will report the allocated queue's
+	 * type as "none" through the 'scheduler' sysfs attribute.
+	 */
+	elv_unregister_queue(md->queue);
+}
+
 static void unlock_fs(struct mapped_device *md);
 
 static void free_dev(struct mapped_device *md)
diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index 489feba..95aec64 100644
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -114,6 +114,7 @@ struct kobject *dm_kobject(struct mapped_device *md);
 struct mapped_device *dm_get_from_kobject(struct kobject *kobj);
 
 int dm_init_request_based_queue(struct mapped_device *md);
+void dm_clear_request_based_queue(struct mapped_device *md);
 
 /*
  * Targets for linear and striped mappings

  reply	other threads:[~2010-05-13  5:03 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-05-10 22:55 [RFC PATCH 1/2] block: allow initialization of previously allocated request_queue Mike Snitzer
2010-05-10 22:55 ` [RFC PATCH 2/2] dm: only initialize full request_queue for request-based device Mike Snitzer
2010-05-11  4:23   ` Kiyoshi Ueda
2010-05-11 13:15     ` Mike Snitzer
2010-05-12  8:23       ` Kiyoshi Ueda
2010-05-13  3:57         ` Mike Snitzer
2010-05-14  8:06           ` Kiyoshi Ueda
2010-05-14 14:08             ` Mike Snitzer
2010-05-17  9:27               ` Kiyoshi Ueda
2010-05-17 17:27                 ` Mike Snitzer
2010-05-18  8:32                   ` Kiyoshi Ueda
2010-05-18 13:46                     ` Mike Snitzer
2010-05-19  5:57                       ` Kiyoshi Ueda
2010-05-19 12:01                         ` Mike Snitzer
2010-05-19 14:39                           ` Mike Snitzer
2010-05-19 14:45                             ` Mike Snitzer
2010-05-20 11:21                             ` Kiyoshi Ueda
2010-05-20 17:07                               ` Mike Snitzer
2010-05-21  8:32                                 ` Kiyoshi Ueda
2010-05-21 13:34                                   ` Mike Snitzer
2010-05-24  9:58                                     ` Kiyoshi Ueda
2010-05-19 21:51                           ` Mike Snitzer
2010-05-13  4:31   ` [PATCH 2/2 v2] " Mike Snitzer
2010-05-13  5:02     ` Mike Snitzer [this message]
2010-05-13 22:14       ` [PATCH 3/2 v2] dm: bio-based device must not register elevator in sysfs Mike Snitzer
2010-05-11  6:55 ` [RFC PATCH 1/2] block: allow initialization of previously allocated request_queue Jens Axboe
2010-05-11 13:18   ` Mike Snitzer
2010-05-11 13:21     ` Jens Axboe

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=20100513050252.GC25523@redhat.com \
    --to=snitzer@redhat.com \
    --cc=dm-devel@redhat.com \
    --cc=j-nomura@ce.jp.nec.com \
    --cc=jens.axboe@oracle.com \
    --cc=k-ueda@ct.jp.nec.com \
    --cc=knikanth@suse.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=vgoyal@redhat.com \
    /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).