All of lore.kernel.org
 help / color / mirror / Atom feed
From: Christoph Hellwig <hch@lst.de>
To: Jan Kara <jack@suse.cz>,
	jaxboe@fusionio.com, tj@kernel.org, James.Bottomley@suse.de,
	linux-fsdevel@vger.kernel.org, linux-scsi@vger.kernel.org,
	tytso@mit.edu, chris.maso
Subject: [PATCH, RFC 1/2] relaxed cache flushes
Date: Tue, 3 Aug 2010 20:49:39 +0200	[thread overview]
Message-ID: <20100803184939.GA12198@lst.de> (raw)
In-Reply-To: <20100727175418.GF6820@quack.suse.cz>

So instead of cracking my head on the relaxed barriers I've decided to
do the easiet part first.  That is relaxing explicit cache flushes
done by blkdev_issue_flush.  These days these are handled as an
empty barrier, which is completely overkill.  Instead take advantage
of the way we now handle flushes, that is as REQ_FLUSH FS requests.

Do a few updates to the block layer so that we handle REQ_FLUSH
correctly and we can make blkdev_issue_flush submit them directly.

All request based block drivers should just work with it, but bio
based remappers will need some additional work.  The next patch
will do this for DM, but I haven't quite grasped the barrier code
in MD yet.  Despite doing a lot REQ_HARDBARRIER tests DRBD doesn't
actually advertize any ordered mode so it's not affected.  The
barrier handling in the loop driver is currently broken anyway,
and I'm still undecided if I want to fix it before or after
this conversion.


Index: linux-2.6/block/blk-barrier.c
===================================================================
--- linux-2.6.orig/block/blk-barrier.c	2010-08-03 20:26:50.259005954 +0200
+++ linux-2.6/block/blk-barrier.c	2010-08-03 20:33:39.580266216 +0200
@@ -151,25 +151,7 @@ static inline bool start_ordered(struct
 	q->ordered = q->next_ordered;
 	q->ordseq |= QUEUE_ORDSEQ_STARTED;
 
-	/*
-	 * For an empty barrier, there's no actual BAR request, which
-	 * in turn makes POSTFLUSH unnecessary.  Mask them off.
-	 */
-	if (!blk_rq_sectors(rq)) {
-		q->ordered &= ~(QUEUE_ORDERED_DO_BAR |
-				QUEUE_ORDERED_DO_POSTFLUSH);
-		/*
-		 * Empty barrier on a write-through device w/ ordered
-		 * tag has no command to issue and without any command
-		 * to issue, ordering by tag can't be used.  Drain
-		 * instead.
-		 */
-		if ((q->ordered & QUEUE_ORDERED_BY_TAG) &&
-		    !(q->ordered & QUEUE_ORDERED_DO_PREFLUSH)) {
-			q->ordered &= ~QUEUE_ORDERED_BY_TAG;
-			q->ordered |= QUEUE_ORDERED_BY_DRAIN;
-		}
-	}
+	BUG_ON(!blk_rq_sectors(rq));
 
 	/* stash away the original request */
 	blk_dequeue_request(rq);
@@ -311,6 +293,9 @@ int blkdev_issue_flush(struct block_devi
 	if (!q)
 		return -ENXIO;
 
+	if (!(q->next_ordered & QUEUE_ORDERED_DO_PREFLUSH))
+		return 0;
+
 	/*
 	 * some block devices may not have their queue correctly set up here
 	 * (e.g. loop device without a backing file) and so issuing a flush
@@ -327,7 +312,7 @@ int blkdev_issue_flush(struct block_devi
 		bio->bi_private = &wait;
 
 	bio_get(bio);
-	submit_bio(WRITE_BARRIER, bio);
+	submit_bio(WRITE_SYNC | REQ_FLUSH, bio);
 	if (test_bit(BLKDEV_WAIT, &flags)) {
 		wait_for_completion(&wait);
 		/*
Index: linux-2.6/block/elevator.c
===================================================================
--- linux-2.6.orig/block/elevator.c	2010-08-03 20:26:50.268024322 +0200
+++ linux-2.6/block/elevator.c	2010-08-03 20:32:11.949256478 +0200
@@ -423,7 +423,8 @@ void elv_dispatch_sort(struct request_qu
 	q->nr_sorted--;
 
 	boundary = q->end_sector;
-	stop_flags = REQ_SOFTBARRIER | REQ_HARDBARRIER | REQ_STARTED;
+	stop_flags = REQ_SOFTBARRIER | REQ_HARDBARRIER | REQ_STARTED | \
+		     REQ_FLUSH;
 	list_for_each_prev(entry, &q->queue_head) {
 		struct request *pos = list_entry_rq(entry);
 
Index: linux-2.6/include/linux/bio.h
===================================================================
--- linux-2.6.orig/include/linux/bio.h	2010-08-03 20:26:50.298255570 +0200
+++ linux-2.6/include/linux/bio.h	2010-08-03 20:46:48.367257736 +0200
@@ -153,6 +153,7 @@ enum rq_flag_bits {
 	__REQ_META,		/* metadata io request */
 	__REQ_DISCARD,		/* request to discard sectors */
 	__REQ_NOIDLE,		/* don't anticipate more IO after this one */
+	__REQ_FLUSH,		/* request for cache flush */
 
 	/* bio only flags */
 	__REQ_UNPLUG,		/* unplug the immediately after submission */
@@ -174,7 +175,6 @@ enum rq_flag_bits {
 	__REQ_ALLOCED,		/* request came from our alloc pool */
 	__REQ_COPY_USER,	/* contains copies of user pages */
 	__REQ_INTEGRITY,	/* integrity metadata has been remapped */
-	__REQ_FLUSH,		/* request for cache flush */
 	__REQ_IO_STAT,		/* account I/O stat */
 	__REQ_MIXED_MERGE,	/* merge of different types, fail separately */
 	__REQ_NR_BITS,		/* stops here */
@@ -189,12 +189,13 @@ enum rq_flag_bits {
 #define REQ_META		(1 << __REQ_META)
 #define REQ_DISCARD		(1 << __REQ_DISCARD)
 #define REQ_NOIDLE		(1 << __REQ_NOIDLE)
+#define REQ_FLUSH		(1 << __REQ_FLUSH)
 
 #define REQ_FAILFAST_MASK \
 	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
 #define REQ_COMMON_MASK \
 	(REQ_WRITE | REQ_FAILFAST_MASK | REQ_HARDBARRIER | REQ_SYNC | \
-	 REQ_META| REQ_DISCARD | REQ_NOIDLE)
+	 REQ_META| REQ_DISCARD | REQ_NOIDLE | REQ_FLUSH)
 
 #define REQ_UNPLUG		(1 << __REQ_UNPLUG)
 #define REQ_RAHEAD		(1 << __REQ_RAHEAD)
@@ -214,7 +215,6 @@ enum rq_flag_bits {
 #define REQ_ALLOCED		(1 << __REQ_ALLOCED)
 #define REQ_COPY_USER		(1 << __REQ_COPY_USER)
 #define REQ_INTEGRITY		(1 << __REQ_INTEGRITY)
-#define REQ_FLUSH		(1 << __REQ_FLUSH)
 #define REQ_IO_STAT		(1 << __REQ_IO_STAT)
 #define REQ_MIXED_MERGE		(1 << __REQ_MIXED_MERGE)
 
Index: linux-2.6/include/linux/blkdev.h
===================================================================
--- linux-2.6.orig/include/linux/blkdev.h	2010-08-03 20:26:50.311003929 +0200
+++ linux-2.6/include/linux/blkdev.h	2010-08-03 20:32:11.956036684 +0200
@@ -589,7 +589,8 @@ static inline void blk_clear_queue_full(
  * it already be started by driver.
  */
 #define RQ_NOMERGE_FLAGS	\
-	(REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
+	(REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER | \
+	 REQ_FLUSH)
 #define rq_mergeable(rq)	\
 	(!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
 	 (((rq)->cmd_flags & REQ_DISCARD) || \
Index: linux-2.6/block/blk-core.c
===================================================================
--- linux-2.6.orig/block/blk-core.c	2010-08-03 20:26:50.275003649 +0200
+++ linux-2.6/block/blk-core.c	2010-08-03 20:32:11.960004138 +0200
@@ -1203,7 +1203,7 @@ static int __make_request(struct request
 	const unsigned int ff = bio->bi_rw & REQ_FAILFAST_MASK;
 	int rw_flags;
 
-	if ((bio->bi_rw & REQ_HARDBARRIER) &&
+	if ((bio->bi_rw & (REQ_HARDBARRIER|REQ_FLUSH)) &&
 	    (q->next_ordered == QUEUE_ORDERED_NONE)) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
@@ -1217,7 +1217,7 @@ static int __make_request(struct request
 
 	spin_lock_irq(q->queue_lock);
 
-	if (unlikely((bio->bi_rw & REQ_HARDBARRIER)) || elv_queue_empty(q))
+	if ((bio->bi_rw & (REQ_HARDBARRIER|REQ_FLUSH)) || elv_queue_empty(q))
 		goto get_rq;
 
 	el_ret = elv_merge(q, &req, bio);

  parent reply	other threads:[~2010-08-03 18:49 UTC|newest]

Thread overview: 155+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-07-27 16:56 [RFC] relaxed barrier semantics Christoph Hellwig
2010-07-27 17:54 ` Jan Kara
2010-07-27 18:35   ` Vivek Goyal
2010-07-27 18:42     ` James Bottomley
2010-07-27 18:51       ` Ric Wheeler
2010-07-27 19:43       ` Christoph Hellwig
2010-07-27 19:38     ` Christoph Hellwig
2010-07-28  8:08     ` Tejun Heo
2010-07-28  8:20       ` Tejun Heo
2010-07-28 13:55         ` Vladislav Bolkhovitin
2010-07-28 14:23           ` Tejun Heo
2010-07-28 14:37             ` James Bottomley
2010-07-28 14:44               ` Tejun Heo
2010-07-28 16:17                 ` Vladislav Bolkhovitin
2010-07-28 16:17               ` Vladislav Bolkhovitin
2010-07-28 16:16             ` Vladislav Bolkhovitin
2010-07-28  8:24       ` Christoph Hellwig
2010-07-28  8:40         ` Tejun Heo
2010-07-28  8:50           ` Christoph Hellwig
2010-07-28  8:58             ` Tejun Heo
2010-07-28  9:00               ` Christoph Hellwig
2010-07-28  9:11                 ` Hannes Reinecke
2010-07-28  9:16                   ` Christoph Hellwig
2010-07-28  9:24                     ` Tejun Heo
2010-07-28  9:38                       ` Christoph Hellwig
2010-07-28  9:28                   ` Steven Whitehouse
2010-07-28  9:35                     ` READ_META semantics, was " Christoph Hellwig
2010-07-28 13:52                       ` Jeff Moyer
2010-07-28  9:17                 ` Tejun Heo
2010-07-28  9:28                   ` Christoph Hellwig
2010-07-28  9:48                     ` Tejun Heo
2010-07-28 10:19                     ` Steven Whitehouse
2010-07-28 11:45                       ` Christoph Hellwig
2010-07-28 12:47                     ` Jan Kara
2010-07-28 23:00                       ` Christoph Hellwig
2010-07-29 10:45                         ` Jan Kara
2010-07-29 16:54                           ` Joel Becker
2010-07-29 17:02                             ` Christoph Hellwig
2010-07-29 17:02                             ` Christoph Hellwig
2010-07-29  1:44                     ` Ted Ts'o
2010-07-29  2:43                       ` Vivek Goyal
2010-07-29  2:43                       ` Vivek Goyal
2010-07-29  8:42                         ` Christoph Hellwig
2010-07-29 20:02                           ` Vivek Goyal
2010-07-29 20:06                             ` Christoph Hellwig
2010-07-30  3:17                               ` Vivek Goyal
2010-07-30  7:07                                 ` Christoph Hellwig
2010-07-30  7:41                                   ` Vivek Goyal
2010-08-02 18:28                                   ` [RFC PATCH] Flush only barriers (Was: Re: [RFC] relaxed barrier semantics) Vivek Goyal
2010-08-03 13:03                                     ` Christoph Hellwig
2010-08-04 15:29                                       ` Vivek Goyal
2010-08-04 16:21                                         ` Christoph Hellwig
2010-07-29  8:31                       ` [RFC] relaxed barrier semantics Christoph Hellwig
2010-07-29 11:16                         ` Jan Kara
2010-07-29 13:00                         ` extfs reliability Vladislav Bolkhovitin
2010-07-29 13:08                           ` Christoph Hellwig
2010-07-29 14:12                             ` Vladislav Bolkhovitin
2010-07-29 14:34                               ` Jan Kara
2010-07-29 18:20                                 ` Vladislav Bolkhovitin
2010-07-29 18:49                                 ` Vladislav Bolkhovitin
2010-07-29 14:26                           ` Jan Kara
2010-07-29 18:20                             ` Vladislav Bolkhovitin
2010-07-29 18:58                           ` Ted Ts'o
2010-07-29 19:44                       ` [RFC] relaxed barrier semantics Ric Wheeler
2010-07-29 19:49                         ` Christoph Hellwig
2010-07-29 19:56                           ` Ric Wheeler
2010-07-29 19:59                             ` James Bottomley
2010-07-29 20:03                               ` Christoph Hellwig
2010-07-29 20:07                                 ` James Bottomley
2010-07-29 20:11                                   ` Christoph Hellwig
2010-07-30 12:45                                     ` Vladislav Bolkhovitin
2010-07-30 12:56                                       ` Christoph Hellwig
2010-08-04  1:58                                     ` Jamie Lokier
2010-07-30 12:46                                 ` Vladislav Bolkhovitin
2010-07-30 12:57                                   ` Christoph Hellwig
2010-07-30 13:09                                     ` Vladislav Bolkhovitin
2010-07-30 13:12                                       ` Christoph Hellwig
2010-07-30 17:40                                         ` Vladislav Bolkhovitin
2010-07-29 20:58                               ` Ric Wheeler
2010-07-29 22:30                             ` Andreas Dilger
2010-07-29 23:04                               ` Ted Ts'o
2010-07-29 23:08                                 ` Ric Wheeler
2010-07-29 23:08                                 ` Ric Wheeler
2010-07-29 23:28                                 ` James Bottomley
2010-07-29 23:37                                   ` James Bottomley
2010-07-30  0:19                                     ` Ted Ts'o
2010-07-30 12:56                                   ` Vladislav Bolkhovitin
2010-07-30  7:11                                 ` Christoph Hellwig
2010-07-30  7:11                                 ` Christoph Hellwig
2010-07-30 12:56                                 ` Vladislav Bolkhovitin
2010-07-30 13:07                                   ` Tejun Heo
2010-07-30 13:22                                     ` Vladislav Bolkhovitin
2010-07-30 13:27                                       ` Vladislav Bolkhovitin
2010-07-30 13:09                                   ` Christoph Hellwig
2010-07-30 13:25                                     ` Vladislav Bolkhovitin
2010-07-30 13:34                                       ` Christoph Hellwig
2010-07-30 13:44                                         ` Vladislav Bolkhovitin
2010-07-30 14:20                                           ` Christoph Hellwig
2010-07-31  0:47                                             ` Jan Kara
2010-07-31  9:12                                               ` Christoph Hellwig
2010-08-02 13:14                                                 ` Jan Kara
2010-08-02 10:38                                               ` Vladislav Bolkhovitin
2010-08-02 12:48                                                 ` Christoph Hellwig
2010-08-02 19:03                                                   ` xfs rm performance Vladislav Bolkhovitin
2010-08-02 19:18                                                     ` Christoph Hellwig
2010-08-05 19:31                                                       ` Vladislav Bolkhovitin
2010-08-02 19:01                                             ` [RFC] relaxed barrier semantics Vladislav Bolkhovitin
2010-08-02 19:26                                               ` Christoph Hellwig
2010-07-30 12:56                                 ` Vladislav Bolkhovitin
2010-07-31  0:35                         ` Jan Kara
2010-07-29 19:44                       ` Ric Wheeler
2010-08-02 16:47                     ` Ryusuke Konishi
2010-08-02 17:39                     ` Chris Mason
2010-08-05 13:11                       ` Vladislav Bolkhovitin
2010-08-05 13:32                         ` Chris Mason
2010-08-05 14:52                           ` Hannes Reinecke
2010-08-05 14:52                           ` Hannes Reinecke
2010-08-05 15:17                             ` Chris Mason
2010-08-05 17:07                             ` Christoph Hellwig
2010-08-05 19:48                           ` Vladislav Bolkhovitin
2010-08-05 19:48                           ` Vladislav Bolkhovitin
2010-08-05 19:50                             ` Christoph Hellwig
2010-08-05 20:05                               ` Vladislav Bolkhovitin
2010-08-06 14:56                                 ` Hannes Reinecke
2010-08-06 18:38                                   ` Vladislav Bolkhovitin
2010-08-06 23:38                                     ` Christoph Hellwig
2010-08-06 23:34                                   ` Christoph Hellwig
2010-08-05 17:09                         ` Christoph Hellwig
2010-08-05 19:32                           ` Vladislav Bolkhovitin
2010-08-05 19:40                             ` Christoph Hellwig
2010-08-05 13:11                       ` Vladislav Bolkhovitin
2010-07-28 13:56                   ` Vladislav Bolkhovitin
2010-07-28 14:42                 ` Vivek Goyal
2010-07-27 19:37   ` Christoph Hellwig
2010-08-03 18:49   ` Christoph Hellwig [this message]
2010-08-03 18:51     ` [PATCH, RFC 2/2] dm: support REQ_FLUSH directly Christoph Hellwig
2010-08-04  4:57       ` Kiyoshi Ueda
2010-08-04  8:54         ` Christoph Hellwig
2010-08-05  2:16           ` Jun'ichi Nomura
2010-08-26 22:50             ` Mike Snitzer
2010-08-27  0:40               ` Mike Snitzer
2010-08-27  1:20                 ` Jamie Lokier
2010-08-27  1:43               ` Jun'ichi Nomura
2010-08-27  4:08                 ` Mike Snitzer
2010-08-27  5:52                   ` Jun'ichi Nomura
2010-08-27 14:13                     ` Mike Snitzer
2010-08-30  4:45                       ` Jun'ichi Nomura
2010-08-30  8:33                         ` Tejun Heo
2010-08-30 12:43                           ` Mike Snitzer
2010-08-30 12:45                             ` Tejun Heo
2010-08-06 16:04     ` [PATCH, RFC] relaxed barriers Tejun Heo
2010-08-06 23:34       ` Christoph Hellwig
2010-08-07 10:13       ` [PATCH REPOST " Tejun Heo
2010-08-08 14:31         ` Christoph Hellwig
2010-08-09 14:50           ` Tejun Heo

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=20100803184939.GA12198@lst.de \
    --to=hch@lst.de \
    --cc=James.Bottomley@suse.de \
    --cc=dm-devel@redhat.com \
    --cc=jack@suse.cz \
    --cc=jaxboe@fusionio.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=tj@kernel.org \
    --cc=tytso@mit.edu \
    /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.