From a82afdfcb8c0df09776b6458af6b68fc58b2e87b Mon Sep 17 00:00:00 2001
From: Tejun Heo <tj@kernel.org>
Date: Fri, 3 Jul 2009 17:48:16 +0900
Subject: block: use the same failfast bits for bio and request

bio and request use the same set of failfast bits.  This patch makes
the following changes to simplify things.

* enumify BIO_RW* bits and reorder bits such that BIOS_RW_FAILFAST_*
  bits coincide with __REQ_FAILFAST_* bits.

* The above pushes BIO_RW_AHEAD out of sync with __REQ_FAILFAST_DEV
  but the matching is useless anyway.  init_request_from_bio() is
  responsible for setting FAILFAST bits on FS requests and non-FS
  requests never use BIO_RW_AHEAD.  Drop the code and comment from
  blk_rq_bio_prep().

* Define REQ_FAILFAST_MASK which is OR of all FAILFAST bits and
  simplify FAILFAST flags handling in init_request_from_bio().

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 include/linux/blkdev.h | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'include/linux/blkdev.h')

diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 69103e053c92..c3015736d814 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -93,6 +93,7 @@ enum rq_flag_bits {
 	__REQ_FAILFAST_DEV,	/* no driver retries of device errors */
 	__REQ_FAILFAST_TRANSPORT, /* no driver retries of transport errors */
 	__REQ_FAILFAST_DRIVER,	/* no driver retries of driver errors */
+	/* above flags must match BIO_RW_* */
 	__REQ_DISCARD,		/* request to discard sectors */
 	__REQ_SORTED,		/* elevator knows about this request */
 	__REQ_SOFTBARRIER,	/* may not be passed by ioscheduler */
@@ -143,6 +144,9 @@ enum rq_flag_bits {
 #define REQ_NOIDLE	(1 << __REQ_NOIDLE)
 #define REQ_IO_STAT	(1 << __REQ_IO_STAT)
 
+#define REQ_FAILFAST_MASK	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | \
+				 REQ_FAILFAST_DRIVER)
+
 #define BLK_MAX_CDB	16
 
 /*
-- 
cgit v1.2.3


From 80a761fd33cf812f771e212139157bf8f58d4b3f Mon Sep 17 00:00:00 2001
From: Tejun Heo <tj@kernel.org>
Date: Fri, 3 Jul 2009 17:48:17 +0900
Subject: block: implement mixed merge of different failfast requests

Failfast has characteristics from other attributes.  When issuing,
executing and successuflly completing requests, failfast doesn't make
any difference.  It only affects how a request is handled on failure.
Allowing requests with different failfast settings to be merged cause
normal IOs to fail prematurely while not allowing has performance
penalties as failfast is used for read aheads which are likely to be
located near in-flight or to-be-issued normal IOs.

This patch introduces the concept of 'mixed merge'.  A request is a
mixed merge if it is merge of segments which require different
handling on failure.  Currently the only mixable attributes are
failfast ones (or lack thereof).

When a bio with different failfast settings is added to an existing
request or requests of different failfast settings are merged, the
merged request is marked mixed.  Each bio carries failfast settings
and the request always tracks failfast state of the first bio.  When
the request fails, blk_rq_err_bytes() can be used to determine how
many bytes can be safely failed without crossing into an area which
requires further retrials.

This allows request merging regardless of failfast settings while
keeping the failure handling correct.

This patch only implements mixed merge but doesn't enable it.  The
next one will update SCSI to make use of mixed merge.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Niel Lambrechts <niel.lambrechts@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c       | 99 ++++++++++++++++++++++++++++++++++++++++++++++++++
 block/blk-merge.c      | 43 ++++++++++++++++++++++
 block/blk.h            |  1 +
 include/linux/blkdev.h | 23 +++++++++---
 4 files changed, 161 insertions(+), 5 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-core.c b/block/blk-core.c
index 4daae1ee2b23..c822239bcc9d 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1157,6 +1157,7 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 	const unsigned short prio = bio_prio(bio);
 	const int sync = bio_sync(bio);
 	const int unplug = bio_unplug(bio);
+	const unsigned int ff = bio->bi_rw & REQ_FAILFAST_MASK;
 	int rw_flags;
 
 	if (bio_barrier(bio) && bio_has_data(bio) &&
@@ -1186,6 +1187,9 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 
 		trace_block_bio_backmerge(q, bio);
 
+		if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff)
+			blk_rq_set_mixed_merge(req);
+
 		req->biotail->bi_next = bio;
 		req->biotail = bio;
 		req->__data_len += bytes;
@@ -1205,6 +1209,12 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 
 		trace_block_bio_frontmerge(q, bio);
 
+		if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff) {
+			blk_rq_set_mixed_merge(req);
+			req->cmd_flags &= ~REQ_FAILFAST_MASK;
+			req->cmd_flags |= ff;
+		}
+
 		bio->bi_next = req->bio;
 		req->bio = bio;
 
@@ -1649,6 +1659,50 @@ int blk_insert_cloned_request(struct request_queue *q, struct request *rq)
 }
 EXPORT_SYMBOL_GPL(blk_insert_cloned_request);
 
+/**
+ * blk_rq_err_bytes - determine number of bytes till the next failure boundary
+ * @rq: request to examine
+ *
+ * Description:
+ *     A request could be merge of IOs which require different failure
+ *     handling.  This function determines the number of bytes which
+ *     can be failed from the beginning of the request without
+ *     crossing into area which need to be retried further.
+ *
+ * Return:
+ *     The number of bytes to fail.
+ *
+ * Context:
+ *     queue_lock must be held.
+ */
+unsigned int blk_rq_err_bytes(const struct request *rq)
+{
+	unsigned int ff = rq->cmd_flags & REQ_FAILFAST_MASK;
+	unsigned int bytes = 0;
+	struct bio *bio;
+
+	if (!(rq->cmd_flags & REQ_MIXED_MERGE))
+		return blk_rq_bytes(rq);
+
+	/*
+	 * Currently the only 'mixing' which can happen is between
+	 * different fastfail types.  We can safely fail portions
+	 * which have all the failfast bits that the first one has -
+	 * the ones which are at least as eager to fail as the first
+	 * one.
+	 */
+	for (bio = rq->bio; bio; bio = bio->bi_next) {
+		if ((bio->bi_rw & ff) != ff)
+			break;
+		bytes += bio->bi_size;
+	}
+
+	/* this could lead to infinite loop */
+	BUG_ON(blk_rq_bytes(rq) && !bytes);
+	return bytes;
+}
+EXPORT_SYMBOL_GPL(blk_rq_err_bytes);
+
 static void blk_account_io_completion(struct request *req, unsigned int bytes)
 {
 	if (blk_do_io_stat(req)) {
@@ -1995,6 +2049,12 @@ bool blk_update_request(struct request *req, int error, unsigned int nr_bytes)
 	if (blk_fs_request(req) || blk_discard_rq(req))
 		req->__sector += total_bytes >> 9;
 
+	/* mixed attributes always follow the first bio */
+	if (req->cmd_flags & REQ_MIXED_MERGE) {
+		req->cmd_flags &= ~REQ_FAILFAST_MASK;
+		req->cmd_flags |= req->bio->bi_rw & REQ_FAILFAST_MASK;
+	}
+
 	/*
 	 * If total number of sectors is less than the first segment
 	 * size, something has gone terribly wrong.
@@ -2173,6 +2233,25 @@ bool blk_end_request_cur(struct request *rq, int error)
 }
 EXPORT_SYMBOL(blk_end_request_cur);
 
+/**
+ * blk_end_request_err - Finish a request till the next failure boundary.
+ * @rq: the request to finish till the next failure boundary for
+ * @error: must be negative errno
+ *
+ * Description:
+ *     Complete @rq till the next failure boundary.
+ *
+ * Return:
+ *     %false - we are done with this request
+ *     %true  - still buffers pending for this request
+ */
+bool blk_end_request_err(struct request *rq, int error)
+{
+	WARN_ON(error >= 0);
+	return blk_end_request(rq, error, blk_rq_err_bytes(rq));
+}
+EXPORT_SYMBOL_GPL(blk_end_request_err);
+
 /**
  * __blk_end_request - Helper function for drivers to complete the request.
  * @rq:       the request being processed
@@ -2232,6 +2311,26 @@ bool __blk_end_request_cur(struct request *rq, int error)
 }
 EXPORT_SYMBOL(__blk_end_request_cur);
 
+/**
+ * __blk_end_request_err - Finish a request till the next failure boundary.
+ * @rq: the request to finish till the next failure boundary for
+ * @error: must be negative errno
+ *
+ * Description:
+ *     Complete @rq till the next failure boundary.  Must be called
+ *     with queue lock held.
+ *
+ * Return:
+ *     %false - we are done with this request
+ *     %true  - still buffers pending for this request
+ */
+bool __blk_end_request_err(struct request *rq, int error)
+{
+	WARN_ON(error >= 0);
+	return __blk_end_request(rq, error, blk_rq_err_bytes(rq));
+}
+EXPORT_SYMBOL_GPL(__blk_end_request_err);
+
 void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
 		     struct bio *bio)
 {
diff --git a/block/blk-merge.c b/block/blk-merge.c
index e1999679a4d5..7c9ca01baa45 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -311,6 +311,36 @@ static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
 	return 1;
 }
 
+/**
+ * blk_rq_set_mixed_merge - mark a request as mixed merge
+ * @rq: request to mark as mixed merge
+ *
+ * Description:
+ *     @rq is about to be mixed merged.  Make sure the attributes
+ *     which can be mixed are set in each bio and mark @rq as mixed
+ *     merged.
+ */
+void blk_rq_set_mixed_merge(struct request *rq)
+{
+	unsigned int ff = rq->cmd_flags & REQ_FAILFAST_MASK;
+	struct bio *bio;
+
+	if (rq->cmd_flags & REQ_MIXED_MERGE)
+		return;
+
+	/*
+	 * @rq will no longer represent mixable attributes for all the
+	 * contained bios.  It will just track those of the first one.
+	 * Distributes the attributs to each bio.
+	 */
+	for (bio = rq->bio; bio; bio = bio->bi_next) {
+		WARN_ON_ONCE((bio->bi_rw & REQ_FAILFAST_MASK) &&
+			     (bio->bi_rw & REQ_FAILFAST_MASK) != ff);
+		bio->bi_rw |= ff;
+	}
+	rq->cmd_flags |= REQ_MIXED_MERGE;
+}
+
 static void blk_account_io_merge(struct request *req)
 {
 	if (blk_do_io_stat(req)) {
@@ -365,6 +395,19 @@ static int attempt_merge(struct request_queue *q, struct request *req,
 	if (!ll_merge_requests_fn(q, req, next))
 		return 0;
 
+	/*
+	 * If failfast settings disagree or any of the two is already
+	 * a mixed merge, mark both as mixed before proceeding.  This
+	 * makes sure that all involved bios have mixable attributes
+	 * set properly.
+	 */
+	if ((req->cmd_flags | next->cmd_flags) & REQ_MIXED_MERGE ||
+	    (req->cmd_flags & REQ_FAILFAST_MASK) !=
+	    (next->cmd_flags & REQ_FAILFAST_MASK)) {
+		blk_rq_set_mixed_merge(req);
+		blk_rq_set_mixed_merge(next);
+	}
+
 	/*
 	 * At this point we have either done a back merge
 	 * or front merge. We need the smaller start_time of
diff --git a/block/blk.h b/block/blk.h
index 3fae6add5430..5ee3d7e72feb 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -104,6 +104,7 @@ int ll_front_merge_fn(struct request_queue *q, struct request *req,
 int attempt_back_merge(struct request_queue *q, struct request *rq);
 int attempt_front_merge(struct request_queue *q, struct request *rq);
 void blk_recalc_rq_segments(struct request *rq);
+void blk_rq_set_mixed_merge(struct request *rq);
 
 void blk_queue_congestion_threshold(struct request_queue *q);
 
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index c3015736d814..650b6a9cb679 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -115,6 +115,7 @@ enum rq_flag_bits {
 	__REQ_INTEGRITY,	/* integrity metadata has been remapped */
 	__REQ_NOIDLE,		/* Don't anticipate more IO after this one */
 	__REQ_IO_STAT,		/* account I/O stat */
+	__REQ_MIXED_MERGE,	/* merge of different types, fail separately */
 	__REQ_NR_BITS,		/* stops here */
 };
 
@@ -143,6 +144,7 @@ enum rq_flag_bits {
 #define REQ_INTEGRITY	(1 << __REQ_INTEGRITY)
 #define REQ_NOIDLE	(1 << __REQ_NOIDLE)
 #define REQ_IO_STAT	(1 << __REQ_IO_STAT)
+#define REQ_MIXED_MERGE	(1 << __REQ_MIXED_MERGE)
 
 #define REQ_FAILFAST_MASK	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | \
 				 REQ_FAILFAST_DRIVER)
@@ -832,11 +834,13 @@ static inline void blk_run_address_space(struct address_space *mapping)
 }
 
 /*
- * blk_rq_pos()		: the current sector
- * blk_rq_bytes()	: bytes left in the entire request
- * blk_rq_cur_bytes()	: bytes left in the current segment
- * blk_rq_sectors()	: sectors left in the entire request
- * blk_rq_cur_sectors()	: sectors left in the current segment
+ * blk_rq_pos()			: the current sector
+ * blk_rq_bytes()		: bytes left in the entire request
+ * blk_rq_cur_bytes()		: bytes left in the current segment
+ * blk_rq_err_bytes()		: bytes left till the next error boundary
+ * blk_rq_sectors()		: sectors left in the entire request
+ * blk_rq_cur_sectors()		: sectors left in the current segment
+ * blk_rq_err_sectors()		: sectors left till the next error boundary
  */
 static inline sector_t blk_rq_pos(const struct request *rq)
 {
@@ -853,6 +857,8 @@ static inline int blk_rq_cur_bytes(const struct request *rq)
 	return rq->bio ? bio_cur_bytes(rq->bio) : 0;
 }
 
+extern unsigned int blk_rq_err_bytes(const struct request *rq);
+
 static inline unsigned int blk_rq_sectors(const struct request *rq)
 {
 	return blk_rq_bytes(rq) >> 9;
@@ -863,6 +869,11 @@ static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
 	return blk_rq_cur_bytes(rq) >> 9;
 }
 
+static inline unsigned int blk_rq_err_sectors(const struct request *rq)
+{
+	return blk_rq_err_bytes(rq) >> 9;
+}
+
 /*
  * Request issue related functions.
  */
@@ -889,10 +900,12 @@ extern bool blk_end_request(struct request *rq, int error,
 			    unsigned int nr_bytes);
 extern void blk_end_request_all(struct request *rq, int error);
 extern bool blk_end_request_cur(struct request *rq, int error);
+extern bool blk_end_request_err(struct request *rq, int error);
 extern bool __blk_end_request(struct request *rq, int error,
 			      unsigned int nr_bytes);
 extern void __blk_end_request_all(struct request *rq, int error);
 extern bool __blk_end_request_cur(struct request *rq, int error);
+extern bool __blk_end_request_err(struct request *rq, int error);
 
 extern void blk_complete_request(struct request *);
 extern void __blk_complete_request(struct request *);
-- 
cgit v1.2.3


From 1f98a13f623e0ef666690a18c1250335fc6d7ef1 Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Fri, 11 Sep 2009 14:32:04 +0200
Subject: bio: first step in sanitizing the bio->bi_rw flag testing

Get rid of any functions that test for these bits and make callers
use bio_rw_flagged() directly. Then it is at least directly apparent
what variable and flag they check.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c            | 25 +++++++++++++------------
 block/cfq-iosched.c         |  2 +-
 block/elevator.c            |  3 ++-
 drivers/block/loop.c        |  2 +-
 drivers/md/dm-raid1.c       |  2 +-
 drivers/md/dm-stripe.c      |  2 +-
 drivers/md/dm.c             | 12 ++++++------
 drivers/md/linear.c         |  2 +-
 drivers/md/multipath.c      |  4 ++--
 drivers/md/raid0.c          |  2 +-
 drivers/md/raid1.c          | 14 ++++++++------
 drivers/md/raid10.c         |  6 +++---
 drivers/md/raid5.c          |  2 +-
 drivers/staging/dst/dcore.c |  5 +++--
 fs/btrfs/volumes.c          |  4 ++--
 include/linux/bio.h         | 25 +++++++------------------
 include/linux/blkdev.h      |  2 +-
 17 files changed, 54 insertions(+), 60 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-core.c b/block/blk-core.c
index c822239bcc9d..52559715cb90 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1114,24 +1114,24 @@ void init_request_from_bio(struct request *req, struct bio *bio)
 	 * Inherit FAILFAST from bio (for read-ahead, and explicit
 	 * FAILFAST).  FAILFAST flags are identical for req and bio.
 	 */
-	if (bio_rw_ahead(bio))
+	if (bio_rw_flagged(bio, BIO_RW_AHEAD))
 		req->cmd_flags |= REQ_FAILFAST_MASK;
 	else
 		req->cmd_flags |= bio->bi_rw & REQ_FAILFAST_MASK;
 
-	if (unlikely(bio_discard(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_DISCARD))) {
 		req->cmd_flags |= REQ_DISCARD;
-		if (bio_barrier(bio))
+		if (bio_rw_flagged(bio, BIO_RW_BARRIER))
 			req->cmd_flags |= REQ_SOFTBARRIER;
 		req->q->prepare_discard_fn(req->q, req);
-	} else if (unlikely(bio_barrier(bio)))
+	} else if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER)))
 		req->cmd_flags |= REQ_HARDBARRIER;
 
-	if (bio_sync(bio))
+	if (bio_rw_flagged(bio, BIO_RW_SYNCIO))
 		req->cmd_flags |= REQ_RW_SYNC;
-	if (bio_rw_meta(bio))
+	if (bio_rw_flagged(bio, BIO_RW_META))
 		req->cmd_flags |= REQ_RW_META;
-	if (bio_noidle(bio))
+	if (bio_rw_flagged(bio, BIO_RW_NOIDLE))
 		req->cmd_flags |= REQ_NOIDLE;
 
 	req->errors = 0;
@@ -1155,12 +1155,12 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 	int el_ret;
 	unsigned int bytes = bio->bi_size;
 	const unsigned short prio = bio_prio(bio);
-	const int sync = bio_sync(bio);
-	const int unplug = bio_unplug(bio);
+	const bool sync = bio_rw_flagged(bio, BIO_RW_SYNCIO);
+	const bool unplug = bio_rw_flagged(bio, BIO_RW_UNPLUG);
 	const unsigned int ff = bio->bi_rw & REQ_FAILFAST_MASK;
 	int rw_flags;
 
-	if (bio_barrier(bio) && bio_has_data(bio) &&
+	if (bio_rw_flagged(bio, BIO_RW_BARRIER) && bio_has_data(bio) &&
 	    (q->next_ordered == QUEUE_ORDERED_NONE)) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
@@ -1174,7 +1174,7 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 
 	spin_lock_irq(q->queue_lock);
 
-	if (unlikely(bio_barrier(bio)) || elv_queue_empty(q))
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER)) || elv_queue_empty(q))
 		goto get_rq;
 
 	el_ret = elv_merge(q, &req, bio);
@@ -1470,7 +1470,8 @@ static inline void __generic_make_request(struct bio *bio)
 		if (bio_check_eod(bio, nr_sectors))
 			goto end_io;
 
-		if (bio_discard(bio) && !q->prepare_discard_fn) {
+		if (bio_rw_flagged(bio, BIO_RW_DISCARD) &&
+		    !q->prepare_discard_fn) {
 			err = -EOPNOTSUPP;
 			goto end_io;
 		}
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index ca0d7e71324b..9e6d0af6c990 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -257,7 +257,7 @@ static inline void cic_set_cfqq(struct cfq_io_context *cic,
  */
 static inline int cfq_bio_sync(struct bio *bio)
 {
-	if (bio_data_dir(bio) == READ || bio_sync(bio))
+	if (bio_data_dir(bio) == READ || bio_rw_flagged(bio, BIO_RW_SYNCIO))
 		return 1;
 
 	return 0;
diff --git a/block/elevator.c b/block/elevator.c
index ca861927ba41..1975b619c86d 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -79,7 +79,8 @@ int elv_rq_merge_ok(struct request *rq, struct bio *bio)
 	/*
 	 * Don't merge file system requests and discard requests
 	 */
-	if (bio_discard(bio) != bio_discard(rq->bio))
+	if (bio_rw_flagged(bio, BIO_RW_DISCARD) !=
+	    bio_rw_flagged(rq->bio, BIO_RW_DISCARD))
 		return 0;
 
 	/*
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 5757188cd1fb..bbb79441d895 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -475,7 +475,7 @@ static int do_bio_filebacked(struct loop_device *lo, struct bio *bio)
 	pos = ((loff_t) bio->bi_sector << 9) + lo->lo_offset;
 
 	if (bio_rw(bio) == WRITE) {
-		int barrier = bio_barrier(bio);
+		bool barrier = bio_rw_flagged(bio, BIO_RW_BARRIER);
 		struct file *file = lo->lo_backing_file;
 
 		if (barrier) {
diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 33f179e66bf5..cc9dc79b0784 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -1129,7 +1129,7 @@ static int mirror_end_io(struct dm_target *ti, struct bio *bio,
 	if (error == -EOPNOTSUPP)
 		goto out;
 
-	if ((error == -EWOULDBLOCK) && bio_rw_ahead(bio))
+	if ((error == -EWOULDBLOCK) && bio_rw_flagged(bio, BIO_RW_AHEAD))
 		goto out;
 
 	if (unlikely(error)) {
diff --git a/drivers/md/dm-stripe.c b/drivers/md/dm-stripe.c
index 3e563d251733..fde658ccbcec 100644
--- a/drivers/md/dm-stripe.c
+++ b/drivers/md/dm-stripe.c
@@ -285,7 +285,7 @@ static int stripe_end_io(struct dm_target *ti, struct bio *bio,
 	if (!error)
 		return 0; /* I/O complete */
 
-	if ((error == -EWOULDBLOCK) && bio_rw_ahead(bio))
+	if ((error == -EWOULDBLOCK) && bio_rw_flagged(bio, BIO_RW_AHEAD))
 		return error;
 
 	if (error == -EOPNOTSUPP)
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index b4845b14740d..ec012f030240 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -586,7 +586,7 @@ static void dec_pending(struct dm_io *io, int error)
 			 */
 			spin_lock_irqsave(&md->deferred_lock, flags);
 			if (__noflush_suspending(md)) {
-				if (!bio_barrier(io->bio))
+				if (!bio_rw_flagged(io->bio, BIO_RW_BARRIER))
 					bio_list_add_head(&md->deferred,
 							  io->bio);
 			} else
@@ -598,7 +598,7 @@ static void dec_pending(struct dm_io *io, int error)
 		io_error = io->error;
 		bio = io->bio;
 
-		if (bio_barrier(bio)) {
+		if (bio_rw_flagged(bio, BIO_RW_BARRIER)) {
 			/*
 			 * There can be just one barrier request so we use
 			 * a per-device variable for error reporting.
@@ -1209,7 +1209,7 @@ static void __split_and_process_bio(struct mapped_device *md, struct bio *bio)
 
 	ci.map = dm_get_table(md);
 	if (unlikely(!ci.map)) {
-		if (!bio_barrier(bio))
+		if (!bio_rw_flagged(bio, BIO_RW_BARRIER))
 			bio_io_error(bio);
 		else
 			if (!md->barrier_error)
@@ -1321,7 +1321,7 @@ static int _dm_request(struct request_queue *q, struct bio *bio)
 	 * we have to queue this io for later.
 	 */
 	if (unlikely(test_bit(DMF_QUEUE_IO_TO_THREAD, &md->flags)) ||
-	    unlikely(bio_barrier(bio))) {
+	    unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		up_read(&md->io_lock);
 
 		if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) &&
@@ -1344,7 +1344,7 @@ static int dm_make_request(struct request_queue *q, struct bio *bio)
 {
 	struct mapped_device *md = q->queuedata;
 
-	if (unlikely(bio_barrier(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
 	}
@@ -2164,7 +2164,7 @@ static void dm_wq_work(struct work_struct *work)
 		if (dm_request_based(md))
 			generic_make_request(c);
 		else {
-			if (bio_barrier(c))
+			if (bio_rw_flagged(c, BIO_RW_BARRIER))
 				process_barrier(md, c);
 			else
 				__split_and_process_bio(md, c);
diff --git a/drivers/md/linear.c b/drivers/md/linear.c
index 5fe39c2a3d2b..ea4842905444 100644
--- a/drivers/md/linear.c
+++ b/drivers/md/linear.c
@@ -288,7 +288,7 @@ static int linear_make_request (struct request_queue *q, struct bio *bio)
 	sector_t start_sector;
 	int cpu;
 
-	if (unlikely(bio_barrier(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
 	}
diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index 7140909f6662..89e76819f61f 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -90,7 +90,7 @@ static void multipath_end_request(struct bio *bio, int error)
 
 	if (uptodate)
 		multipath_end_bh_io(mp_bh, 0);
-	else if (!bio_rw_ahead(bio)) {
+	else if (!bio_rw_flagged(bio, BIO_RW_AHEAD)) {
 		/*
 		 * oops, IO error:
 		 */
@@ -144,7 +144,7 @@ static int multipath_make_request (struct request_queue *q, struct bio * bio)
 	const int rw = bio_data_dir(bio);
 	int cpu;
 
-	if (unlikely(bio_barrier(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
 	}
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index 898e2bdfee47..f845ed98fec9 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -448,7 +448,7 @@ static int raid0_make_request(struct request_queue *q, struct bio *bio)
 	const int rw = bio_data_dir(bio);
 	int cpu;
 
-	if (unlikely(bio_barrier(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
 	}
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 8726fd7ebce5..ff7ed3335995 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -782,8 +782,9 @@ static int make_request(struct request_queue *q, struct bio * bio)
 	struct bio_list bl;
 	struct page **behind_pages = NULL;
 	const int rw = bio_data_dir(bio);
-	const int do_sync = bio_sync(bio);
-	int cpu, do_barriers;
+	const bool do_sync = bio_rw_flagged(bio, BIO_RW_SYNCIO);
+	int cpu;
+	bool do_barriers;
 	mdk_rdev_t *blocked_rdev;
 
 	/*
@@ -797,7 +798,8 @@ static int make_request(struct request_queue *q, struct bio * bio)
 
 	md_write_start(mddev, bio); /* wait on superblock update early */
 
-	if (unlikely(!mddev->barriers_work && bio_barrier(bio))) {
+	if (unlikely(!mddev->barriers_work &&
+		     bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		if (rw == WRITE)
 			md_write_end(mddev);
 		bio_endio(bio, -EOPNOTSUPP);
@@ -925,7 +927,7 @@ static int make_request(struct request_queue *q, struct bio * bio)
 	atomic_set(&r1_bio->remaining, 0);
 	atomic_set(&r1_bio->behind_remaining, 0);
 
-	do_barriers = bio_barrier(bio);
+	do_barriers = bio_rw_flagged(bio, BIO_RW_BARRIER);
 	if (do_barriers)
 		set_bit(R1BIO_Barrier, &r1_bio->state);
 
@@ -1600,7 +1602,7 @@ static void raid1d(mddev_t *mddev)
 			 * We already have a nr_pending reference on these rdevs.
 			 */
 			int i;
-			const int do_sync = bio_sync(r1_bio->master_bio);
+			const bool do_sync = bio_rw_flagged(r1_bio->master_bio, BIO_RW_SYNCIO);
 			clear_bit(R1BIO_BarrierRetry, &r1_bio->state);
 			clear_bit(R1BIO_Barrier, &r1_bio->state);
 			for (i=0; i < conf->raid_disks; i++)
@@ -1654,7 +1656,7 @@ static void raid1d(mddev_t *mddev)
 				       (unsigned long long)r1_bio->sector);
 				raid_end_bio_io(r1_bio);
 			} else {
-				const int do_sync = bio_sync(r1_bio->master_bio);
+				const bool do_sync = bio_rw_flagged(r1_bio->master_bio, BIO_RW_SYNCIO);
 				r1_bio->bios[r1_bio->read_disk] =
 					mddev->ro ? IO_BLOCKED : NULL;
 				r1_bio->read_disk = disk;
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 3d9020cf6f6e..d0a2152e064f 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -796,12 +796,12 @@ static int make_request(struct request_queue *q, struct bio * bio)
 	int i;
 	int chunk_sects = conf->chunk_mask + 1;
 	const int rw = bio_data_dir(bio);
-	const int do_sync = bio_sync(bio);
+	const bool do_sync = bio_rw_flagged(bio, BIO_RW_SYNCIO);
 	struct bio_list bl;
 	unsigned long flags;
 	mdk_rdev_t *blocked_rdev;
 
-	if (unlikely(bio_barrier(bio))) {
+	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
 		bio_endio(bio, -EOPNOTSUPP);
 		return 0;
 	}
@@ -1610,7 +1610,7 @@ static void raid10d(mddev_t *mddev)
 				raid_end_bio_io(r10_bio);
 				bio_put(bio);
 			} else {
-				const int do_sync = bio_sync(r10_bio->master_bio);
+				const bool do_sync = bio_rw_flagged(r10_bio->master_bio, BIO_RW_SYNCIO);
 				bio_put(bio);
 				rdev = conf->mirrors[mirror].rdev;
 				if (printk_ratelimit())
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index b8a2c5dc67ba..826eb3467357 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -3606,7 +3606,7 @@ static int make_request(struct request_queue *q, struct bio * bi)
 	const int rw = bio_data_dir(bi);
 	int cpu, remaining;
 
-	if (unlikely(bio_barrier(bi))) {
+	if (unlikely(bio_rw_flagged(bi, BIO_RW_BARRIER))) {
 		bio_endio(bi, -EOPNOTSUPP);
 		return 0;
 	}
diff --git a/drivers/staging/dst/dcore.c b/drivers/staging/dst/dcore.c
index fad25b753042..b1c258ca2102 100644
--- a/drivers/staging/dst/dcore.c
+++ b/drivers/staging/dst/dcore.c
@@ -112,8 +112,9 @@ static int dst_request(struct request_queue *q, struct bio *bio)
 		 * I worked with.
 		 *
 		 * Empty barriers are not allowed anyway, see 51fd77bd9f512
-		 * for example, although later it was changed to bio_discard()
-		 * only, which does not work in this case.
+		 * for example, although later it was changed to
+		 * bio_rw_flagged(bio, BIO_RW_DISCARD) only, which does not
+		 * work in this case.
 		 */
 		//err = -EOPNOTSUPP;
 		err = 0;
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 5dbefd11b4af..5cf405b0828d 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -260,7 +260,7 @@ loop_lock:
 		num_run++;
 		batch_run++;
 
-		if (bio_sync(cur))
+		if (bio_rw_flagged(cur, BIO_RW_SYNCIO))
 			num_sync_run++;
 
 		if (need_resched()) {
@@ -2903,7 +2903,7 @@ static noinline int schedule_bio(struct btrfs_root *root,
 	bio->bi_rw |= rw;
 
 	spin_lock(&device->io_lock);
-	if (bio_sync(bio))
+	if (bio_rw_flagged(bio, BIO_RW_SYNCIO))
 		pending_bios = &device->pending_sync_bios;
 	else
 		pending_bios = &device->pending_bios;
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 4f8fd0221cd2..5be93f18d842 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -177,28 +177,17 @@ enum bio_rw_flags {
 	BIO_RW_NOIDLE,
 };
 
+/*
+ * First four bits must match between bio->bi_rw and rq->cmd_flags, make
+ * that explicit here.
+ */
+#define BIO_RW_RQ_MASK		0xf
+
 static inline bool bio_rw_flagged(struct bio *bio, enum bio_rw_flags flag)
 {
 	return (bio->bi_rw & (1 << flag)) != 0;
 }
 
-/*
- * Old defines, these should eventually be replaced by direct usage of
- * bio_rw_flagged()
- */
-#define bio_barrier(bio)	bio_rw_flagged(bio, BIO_RW_BARRIER)
-#define bio_sync(bio)		bio_rw_flagged(bio, BIO_RW_SYNCIO)
-#define bio_unplug(bio)		bio_rw_flagged(bio, BIO_RW_UNPLUG)
-#define bio_failfast_dev(bio)	bio_rw_flagged(bio, BIO_RW_FAILFAST_DEV)
-#define bio_failfast_transport(bio)	\
-		bio_rw_flagged(bio, BIO_RW_FAILFAST_TRANSPORT)
-#define bio_failfast_driver(bio) 	\
-		bio_rw_flagged(bio, BIO_RW_FAILFAST_DRIVER)
-#define bio_rw_ahead(bio)	bio_rw_flagged(bio, BIO_RW_AHEAD)
-#define bio_rw_meta(bio)	bio_rw_flagged(bio, BIO_RW_META)
-#define bio_discard(bio)	bio_rw_flagged(bio, BIO_RW_DISCARD)
-#define bio_noidle(bio)		bio_rw_flagged(bio, BIO_RW_NOIDLE)
-
 /*
  * upper 16 bits of bi_rw define the io priority of this bio
  */
@@ -222,7 +211,7 @@ static inline bool bio_rw_flagged(struct bio *bio, enum bio_rw_flags flag)
 #define bio_offset(bio)		bio_iovec((bio))->bv_offset
 #define bio_segments(bio)	((bio)->bi_vcnt - (bio)->bi_idx)
 #define bio_sectors(bio)	((bio)->bi_size >> 9)
-#define bio_empty_barrier(bio)	(bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
+#define bio_empty_barrier(bio)	(bio_rw_flagged(bio, BIO_RW_BARRIER) && !bio_has_data(bio) && !bio_rw_flagged(bio, BIO_RW_DISCARD))
 
 static inline unsigned int bio_cur_bytes(struct bio *bio)
 {
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 650b6a9cb679..88edb62def82 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -86,7 +86,7 @@ enum {
 };
 
 /*
- * request type modified bits. first two bits match BIO_RW* bits, important
+ * request type modified bits. first four bits match BIO_RW* bits, important
  */
 enum rq_flag_bits {
 	__REQ_RW,		/* not set, read. set, write */
-- 
cgit v1.2.3


From fb1e75389bd06fd5987e9cda1b4e0305c782f854 Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Thu, 30 Jul 2009 08:18:24 +0200
Subject: block: improve queue_should_plug() by looking at IO depths

Instead of just checking whether this device uses block layer
tagging, we can improve the detection by looking at the maximum
queue depth it has reached. If that crosses 4, then deem it a
queuing device.

This is important on high IOPS devices, since plugging hurts
the performance there (it can be as much as 10-15% of the sys
time).

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c       | 11 +++++++++--
 include/linux/blkdev.h |  2 ++
 2 files changed, 11 insertions(+), 2 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-core.c b/block/blk-core.c
index 52559715cb90..93051d151635 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1146,7 +1146,7 @@ void init_request_from_bio(struct request *req, struct bio *bio)
  */
 static inline bool queue_should_plug(struct request_queue *q)
 {
-	return !(blk_queue_nonrot(q) && blk_queue_tagged(q));
+	return !(blk_queue_nonrot(q) && blk_queue_queuing(q));
 }
 
 static int __make_request(struct request_queue *q, struct bio *bio)
@@ -1857,8 +1857,15 @@ void blk_dequeue_request(struct request *rq)
 	 * and to it is freed is accounted as io that is in progress at
 	 * the driver side.
 	 */
-	if (blk_account_rq(rq))
+	if (blk_account_rq(rq)) {
 		q->in_flight[rq_is_sync(rq)]++;
+		/*
+		 * Mark this device as supporting hardware queuing, if
+		 * we have more IOs in flight than 4.
+		 */
+		if (!blk_queue_queuing(q) && queue_in_flight(q) > 4)
+			set_bit(QUEUE_FLAG_CQ, &q->queue_flags);
+	}
 }
 
 /**
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 88edb62def82..98b45633a27e 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -459,6 +459,7 @@ struct request_queue
 #define QUEUE_FLAG_NONROT      14	/* non-rotational device (SSD) */
 #define QUEUE_FLAG_VIRT        QUEUE_FLAG_NONROT /* paravirt device */
 #define QUEUE_FLAG_IO_STAT     15	/* do IO stats */
+#define QUEUE_FLAG_CQ	       16	/* hardware does queuing */
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_CLUSTER) |		\
@@ -581,6 +582,7 @@ enum {
 
 #define blk_queue_plugged(q)	test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
 #define blk_queue_tagged(q)	test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
+#define blk_queue_queuing(q)	test_bit(QUEUE_FLAG_CQ, &(q)->queue_flags)
 #define blk_queue_stopped(q)	test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
 #define blk_queue_nomerges(q)	test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
 #define blk_queue_nonrot(q)	test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
-- 
cgit v1.2.3


From 01e97f6b897bf06ec83375d691f2f4d57f5b3a09 Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Thu, 3 Sep 2009 20:06:47 +0200
Subject: block: enable rq CPU completion affinity by default

Test results here look good, and on big OLTP runs it has also shown
to significantly increase cycles attributed to the database and
cause a performance boost.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 include/linux/blkdev.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'include/linux/blkdev.h')

diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 98b45633a27e..8bf1a10e4d88 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -463,7 +463,8 @@ struct request_queue
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_CLUSTER) |		\
-				 (1 << QUEUE_FLAG_STACKABLE))
+				 (1 << QUEUE_FLAG_STACKABLE)	|	\
+				 (1 << QUEUE_FLAG_SAME_COMP))
 
 static inline int queue_is_locked(struct request_queue *q)
 {
-- 
cgit v1.2.3


From 3c5820c743479285ce2678fd3c12b1fd39fe998f Mon Sep 17 00:00:00 2001
From: "Martin K. Petersen" <martin.petersen@oracle.com>
Date: Fri, 11 Sep 2009 21:54:52 +0200
Subject: block: Optimal I/O limit wrapper

Implement blk_limits_io_opt() and make blk_queue_io_opt() a wrapper
around it. DM needs this to avoid poking at the queue_limits directly.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-settings.c   | 21 ++++++++++++++++++++-
 drivers/md/dm-stripe.c |  2 +-
 include/linux/blkdev.h |  1 +
 3 files changed, 22 insertions(+), 2 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-settings.c b/block/blk-settings.c
index 476d87065073..83413ff83739 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -427,6 +427,25 @@ void blk_queue_io_min(struct request_queue *q, unsigned int min)
 }
 EXPORT_SYMBOL(blk_queue_io_min);
 
+/**
+ * blk_limits_io_opt - set optimal request size for a device
+ * @limits: the queue limits
+ * @opt:  smallest I/O size in bytes
+ *
+ * Description:
+ *   Storage devices may report an optimal I/O size, which is the
+ *   device's preferred unit for sustained I/O.  This is rarely reported
+ *   for disk drives.  For RAID arrays it is usually the stripe width or
+ *   the internal track size.  A properly aligned multiple of
+ *   optimal_io_size is the preferred request size for workloads where
+ *   sustained throughput is desired.
+ */
+void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt)
+{
+	limits->io_opt = opt;
+}
+EXPORT_SYMBOL(blk_limits_io_opt);
+
 /**
  * blk_queue_io_opt - set optimal request size for the queue
  * @q:	the request queue for the device
@@ -442,7 +461,7 @@ EXPORT_SYMBOL(blk_queue_io_min);
  */
 void blk_queue_io_opt(struct request_queue *q, unsigned int opt)
 {
-	q->limits.io_opt = opt;
+	blk_limits_io_opt(&q->limits, opt);
 }
 EXPORT_SYMBOL(blk_queue_io_opt);
 
diff --git a/drivers/md/dm-stripe.c b/drivers/md/dm-stripe.c
index fde658ccbcec..e0efc1adcaff 100644
--- a/drivers/md/dm-stripe.c
+++ b/drivers/md/dm-stripe.c
@@ -336,7 +336,7 @@ static void stripe_io_hints(struct dm_target *ti,
 	unsigned chunk_size = (sc->chunk_mask + 1) << 9;
 
 	blk_limits_io_min(limits, chunk_size);
-	limits->io_opt = chunk_size * sc->stripes;
+	blk_limits_io_opt(limits, chunk_size * sc->stripes);
 }
 
 static struct target_type stripe_target = {
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 8bf1a10e4d88..86c2bdff3b89 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -935,6 +935,7 @@ extern void blk_queue_alignment_offset(struct request_queue *q,
 				       unsigned int alignment);
 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
+extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
 extern void blk_set_default_limits(struct queue_limits *lim);
 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
-- 
cgit v1.2.3


From 746cd1e7e4a555ddaee53b19a46e05c9c61eaf09 Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <hch@infradead.org>
Date: Sat, 12 Sep 2009 07:35:43 +0200
Subject: block: use blkdev_issue_discard in blk_ioctl_discard

blk_ioctl_discard duplicates large amounts of code from blkdev_issue_discard,
the only difference between the two is that blkdev_issue_discard needs to
send a barrier discard request and blk_ioctl_discard a non-barrier one,
and blk_ioctl_discard needs to wait on the request.  To facilitates this
add a flags argument to blkdev_issue_discard to control both aspects of the
behaviour.  This will be very useful later on for using the waiting
funcitonality for other callers.

Based on an earlier patch from Matthew Wilcox <matthew@wil.cx>.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-barrier.c    | 31 +++++++++++++++++++------------
 block/ioctl.c          | 49 ++-----------------------------------------------
 fs/btrfs/extent-tree.c |  3 ++-
 fs/gfs2/rgrp.c         |  6 ++++--
 include/linux/blkdev.h |  9 ++++++---
 mm/swapfile.c          |  6 ++++--
 6 files changed, 37 insertions(+), 67 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-barrier.c b/block/blk-barrier.c
index 30022b4e2f63..6593ab39cfe9 100644
--- a/block/blk-barrier.c
+++ b/block/blk-barrier.c
@@ -348,6 +348,9 @@ static void blkdev_discard_end_io(struct bio *bio, int err)
 		clear_bit(BIO_UPTODATE, &bio->bi_flags);
 	}
 
+	if (bio->bi_private)
+		complete(bio->bi_private);
+
 	bio_put(bio);
 }
 
@@ -357,21 +360,20 @@ static void blkdev_discard_end_io(struct bio *bio, int err)
  * @sector:	start sector
  * @nr_sects:	number of sectors to discard
  * @gfp_mask:	memory allocation flags (for bio_alloc)
+ * @flags:	DISCARD_FL_* flags to control behaviour
  *
  * Description:
- *    Issue a discard request for the sectors in question. Does not wait.
+ *    Issue a discard request for the sectors in question.
  */
-int blkdev_issue_discard(struct block_device *bdev,
-			 sector_t sector, sector_t nr_sects, gfp_t gfp_mask)
+int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
+		sector_t nr_sects, gfp_t gfp_mask, int flags)
 {
-	struct request_queue *q;
-	struct bio *bio;
+	DECLARE_COMPLETION_ONSTACK(wait);
+	struct request_queue *q = bdev_get_queue(bdev);
+	int type = flags & DISCARD_FL_BARRIER ?
+		DISCARD_BARRIER : DISCARD_NOBARRIER;
 	int ret = 0;
 
-	if (bdev->bd_disk == NULL)
-		return -ENXIO;
-
-	q = bdev_get_queue(bdev);
 	if (!q)
 		return -ENXIO;
 
@@ -379,12 +381,14 @@ int blkdev_issue_discard(struct block_device *bdev,
 		return -EOPNOTSUPP;
 
 	while (nr_sects && !ret) {
-		bio = bio_alloc(gfp_mask, 0);
+		struct bio *bio = bio_alloc(gfp_mask, 0);
 		if (!bio)
 			return -ENOMEM;
 
 		bio->bi_end_io = blkdev_discard_end_io;
 		bio->bi_bdev = bdev;
+		if (flags & DISCARD_FL_WAIT)
+			bio->bi_private = &wait;
 
 		bio->bi_sector = sector;
 
@@ -396,10 +400,13 @@ int blkdev_issue_discard(struct block_device *bdev,
 			bio->bi_size = nr_sects << 9;
 			nr_sects = 0;
 		}
+
 		bio_get(bio);
-		submit_bio(DISCARD_BARRIER, bio);
+		submit_bio(type, bio);
+
+		if (flags & DISCARD_FL_WAIT)
+			wait_for_completion(&wait);
 
-		/* Check if it failed immediately */
 		if (bio_flagged(bio, BIO_EOPNOTSUPP))
 			ret = -EOPNOTSUPP;
 		else if (!bio_flagged(bio, BIO_UPTODATE))
diff --git a/block/ioctl.c b/block/ioctl.c
index 500e4c73cc52..d3e6b5827a34 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -112,22 +112,9 @@ static int blkdev_reread_part(struct block_device *bdev)
 	return res;
 }
 
-static void blk_ioc_discard_endio(struct bio *bio, int err)
-{
-	if (err) {
-		if (err == -EOPNOTSUPP)
-			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
-		clear_bit(BIO_UPTODATE, &bio->bi_flags);
-	}
-	complete(bio->bi_private);
-}
-
 static int blk_ioctl_discard(struct block_device *bdev, uint64_t start,
 			     uint64_t len)
 {
-	struct request_queue *q = bdev_get_queue(bdev);
-	int ret = 0;
-
 	if (start & 511)
 		return -EINVAL;
 	if (len & 511)
@@ -137,40 +124,8 @@ static int blk_ioctl_discard(struct block_device *bdev, uint64_t start,
 
 	if (start + len > (bdev->bd_inode->i_size >> 9))
 		return -EINVAL;
-
-	if (!q->prepare_discard_fn)
-		return -EOPNOTSUPP;
-
-	while (len && !ret) {
-		DECLARE_COMPLETION_ONSTACK(wait);
-		struct bio *bio;
-
-		bio = bio_alloc(GFP_KERNEL, 0);
-
-		bio->bi_end_io = blk_ioc_discard_endio;
-		bio->bi_bdev = bdev;
-		bio->bi_private = &wait;
-		bio->bi_sector = start;
-
-		if (len > queue_max_hw_sectors(q)) {
-			bio->bi_size = queue_max_hw_sectors(q) << 9;
-			len -= queue_max_hw_sectors(q);
-			start += queue_max_hw_sectors(q);
-		} else {
-			bio->bi_size = len << 9;
-			len = 0;
-		}
-		submit_bio(DISCARD_NOBARRIER, bio);
-
-		wait_for_completion(&wait);
-
-		if (bio_flagged(bio, BIO_EOPNOTSUPP))
-			ret = -EOPNOTSUPP;
-		else if (!bio_flagged(bio, BIO_UPTODATE))
-			ret = -EIO;
-		bio_put(bio);
-	}
-	return ret;
+	return blkdev_issue_discard(bdev, start, len, GFP_KERNEL,
+				    DISCARD_FL_WAIT);
 }
 
 static int put_ushort(unsigned long arg, unsigned short val)
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 72a2b9c28e9f..535f85ba104f 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -1511,7 +1511,8 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
 static void btrfs_issue_discard(struct block_device *bdev,
 				u64 start, u64 len)
 {
-	blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL);
+	blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL,
+			     DISCARD_FL_BARRIER);
 }
 #endif
 
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index fba795798d3a..fbc43241f2ef 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -857,7 +857,8 @@ static void gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset,
 					goto start_new_extent;
 				if ((start + nr_sects) != blk) {
 					rv = blkdev_issue_discard(bdev, start,
-							    nr_sects, GFP_NOFS);
+							    nr_sects, GFP_NOFS,
+							    DISCARD_FL_BARRIER);
 					if (rv)
 						goto fail;
 					nr_sects = 0;
@@ -871,7 +872,8 @@ start_new_extent:
 		}
 	}
 	if (nr_sects) {
-		rv = blkdev_issue_discard(bdev, start, nr_sects, GFP_NOFS);
+		rv = blkdev_issue_discard(bdev, start, nr_sects, GFP_NOFS,
+					 DISCARD_FL_BARRIER);
 		if (rv)
 			goto fail;
 	}
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 86c2bdff3b89..e23a86cae5ac 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -998,15 +998,18 @@ static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
 }
 
 extern int blkdev_issue_flush(struct block_device *, sector_t *);
-extern int blkdev_issue_discard(struct block_device *,
-				sector_t sector, sector_t nr_sects, gfp_t);
+#define DISCARD_FL_WAIT		0x01	/* wait for completion */
+#define DISCARD_FL_BARRIER	0x02	/* issue DISCARD_BARRIER request */
+extern int blkdev_issue_discard(struct block_device *, sector_t sector,
+		sector_t nr_sects, gfp_t, int flags);
 
 static inline int sb_issue_discard(struct super_block *sb,
 				   sector_t block, sector_t nr_blocks)
 {
 	block <<= (sb->s_blocksize_bits - 9);
 	nr_blocks <<= (sb->s_blocksize_bits - 9);
-	return blkdev_issue_discard(sb->s_bdev, block, nr_blocks, GFP_KERNEL);
+	return blkdev_issue_discard(sb->s_bdev, block, nr_blocks, GFP_KERNEL,
+				    DISCARD_FL_BARRIER);
 }
 
 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 8ffdc0d23c53..74f1102e8749 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -161,7 +161,8 @@ static int discard_swap(struct swap_info_struct *si)
 		}
 
 		err = blkdev_issue_discard(si->bdev, start_block,
-						nr_blocks, GFP_KERNEL);
+						nr_blocks, GFP_KERNEL,
+						DISCARD_FL_BARRIER);
 		if (err)
 			break;
 
@@ -200,7 +201,8 @@ static void discard_swap_cluster(struct swap_info_struct *si,
 			start_block <<= PAGE_SHIFT - 9;
 			nr_blocks <<= PAGE_SHIFT - 9;
 			if (blkdev_issue_discard(si->bdev, start_block,
-							nr_blocks, GFP_NOIO))
+							nr_blocks, GFP_NOIO,
+							DISCARD_FL_BARRIER))
 				break;
 		}
 
-- 
cgit v1.2.3


From c15227de132f1295f3db6b7df9079956b1020fd8 Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <hch@infradead.org>
Date: Wed, 30 Sep 2009 13:52:12 +0200
Subject: block: use normal I/O path for discard requests

prepare_discard_fn() was being called in a place where memory allocation
was effectively impossible.  This makes it inappropriate for all but
the most trivial translations of Linux's DISCARD operation to the block
command set.  Additionally adding a payload there makes the ownership
of the bio backing unclear as it's now allocated by the device driver
and not the submitter as usual.

It is replaced with QUEUE_FLAG_DISCARD which is used to indicate whether
the queue supports discard operations or not.  blkdev_issue_discard now
allocates a one-page, sector-length payload which is the right thing
for the common ATA and SCSI implementations.

The mtd implementation of prepare_discard_fn() is replaced with simply
checking for the request being a discard.

Largely based on a previous patch from Matthew Wilcox <matthew@wil.cx>
which did the prepare_discard_fn but not the different payload allocation
yet.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-barrier.c         | 35 ++++++++++++++++++++++++++++++-----
 block/blk-core.c            |  3 +--
 block/blk-settings.c        | 17 -----------------
 drivers/mtd/mtd_blkdevs.c   | 19 +++++--------------
 drivers/staging/dst/dcore.c |  2 +-
 include/linux/blkdev.h      |  6 ++----
 6 files changed, 39 insertions(+), 43 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-barrier.c b/block/blk-barrier.c
index 6593ab39cfe9..21f5025c3945 100644
--- a/block/blk-barrier.c
+++ b/block/blk-barrier.c
@@ -350,6 +350,7 @@ static void blkdev_discard_end_io(struct bio *bio, int err)
 
 	if (bio->bi_private)
 		complete(bio->bi_private);
+	__free_page(bio_page(bio));
 
 	bio_put(bio);
 }
@@ -372,26 +373,44 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 	struct request_queue *q = bdev_get_queue(bdev);
 	int type = flags & DISCARD_FL_BARRIER ?
 		DISCARD_BARRIER : DISCARD_NOBARRIER;
+	struct bio *bio;
+	struct page *page;
 	int ret = 0;
 
 	if (!q)
 		return -ENXIO;
 
-	if (!q->prepare_discard_fn)
+	if (!blk_queue_discard(q))
 		return -EOPNOTSUPP;
 
 	while (nr_sects && !ret) {
-		struct bio *bio = bio_alloc(gfp_mask, 0);
-		if (!bio)
-			return -ENOMEM;
+		unsigned int sector_size = q->limits.logical_block_size;
 
+		bio = bio_alloc(gfp_mask, 1);
+		if (!bio)
+			goto out;
+		bio->bi_sector = sector;
 		bio->bi_end_io = blkdev_discard_end_io;
 		bio->bi_bdev = bdev;
 		if (flags & DISCARD_FL_WAIT)
 			bio->bi_private = &wait;
 
-		bio->bi_sector = sector;
+		/*
+		 * Add a zeroed one-sector payload as that's what
+		 * our current implementations need.  If we'll ever need
+		 * more the interface will need revisiting.
+		 */
+		page = alloc_page(GFP_KERNEL | __GFP_ZERO);
+		if (!page)
+			goto out_free_bio;
+		if (bio_add_pc_page(q, bio, page, sector_size, 0) < sector_size)
+			goto out_free_page;
 
+		/*
+		 * And override the bio size - the way discard works we
+		 * touch many more blocks on disk than the actual payload
+		 * length.
+		 */
 		if (nr_sects > queue_max_hw_sectors(q)) {
 			bio->bi_size = queue_max_hw_sectors(q) << 9;
 			nr_sects -= queue_max_hw_sectors(q);
@@ -414,5 +433,11 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 		bio_put(bio);
 	}
 	return ret;
+out_free_page:
+	__free_page(page);
+out_free_bio:
+	bio_put(bio);
+out:
+	return -ENOMEM;
 }
 EXPORT_SYMBOL(blkdev_issue_discard);
diff --git a/block/blk-core.c b/block/blk-core.c
index 8135228e4b29..80a020dd1580 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1124,7 +1124,6 @@ void init_request_from_bio(struct request *req, struct bio *bio)
 		req->cmd_flags |= REQ_DISCARD;
 		if (bio_rw_flagged(bio, BIO_RW_BARRIER))
 			req->cmd_flags |= REQ_SOFTBARRIER;
-		req->q->prepare_discard_fn(req->q, req);
 	} else if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER)))
 		req->cmd_flags |= REQ_HARDBARRIER;
 
@@ -1470,7 +1469,7 @@ static inline void __generic_make_request(struct bio *bio)
 			goto end_io;
 
 		if (bio_rw_flagged(bio, BIO_RW_DISCARD) &&
-		    !q->prepare_discard_fn) {
+		    !blk_queue_discard(q)) {
 			err = -EOPNOTSUPP;
 			goto end_io;
 		}
diff --git a/block/blk-settings.c b/block/blk-settings.c
index eaf122ff5f16..d29498ef1eb5 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -33,23 +33,6 @@ void blk_queue_prep_rq(struct request_queue *q, prep_rq_fn *pfn)
 }
 EXPORT_SYMBOL(blk_queue_prep_rq);
 
-/**
- * blk_queue_set_discard - set a discard_sectors function for queue
- * @q:		queue
- * @dfn:	prepare_discard function
- *
- * It's possible for a queue to register a discard callback which is used
- * to transform a discard request into the appropriate type for the
- * hardware. If none is registered, then discard requests are failed
- * with %EOPNOTSUPP.
- *
- */
-void blk_queue_set_discard(struct request_queue *q, prepare_discard_fn *dfn)
-{
-	q->prepare_discard_fn = dfn;
-}
-EXPORT_SYMBOL(blk_queue_set_discard);
-
 /**
  * blk_queue_merge_bvec - set a merge_bvec function for queue
  * @q:		queue
diff --git a/drivers/mtd/mtd_blkdevs.c b/drivers/mtd/mtd_blkdevs.c
index 0acbf4f5be50..8ca17a3e96ea 100644
--- a/drivers/mtd/mtd_blkdevs.c
+++ b/drivers/mtd/mtd_blkdevs.c
@@ -32,14 +32,6 @@ struct mtd_blkcore_priv {
 	spinlock_t queue_lock;
 };
 
-static int blktrans_discard_request(struct request_queue *q,
-				    struct request *req)
-{
-	req->cmd_type = REQ_TYPE_LINUX_BLOCK;
-	req->cmd[0] = REQ_LB_OP_DISCARD;
-	return 0;
-}
-
 static int do_blktrans_request(struct mtd_blktrans_ops *tr,
 			       struct mtd_blktrans_dev *dev,
 			       struct request *req)
@@ -52,10 +44,6 @@ static int do_blktrans_request(struct mtd_blktrans_ops *tr,
 
 	buf = req->buffer;
 
-	if (req->cmd_type == REQ_TYPE_LINUX_BLOCK &&
-	    req->cmd[0] == REQ_LB_OP_DISCARD)
-		return tr->discard(dev, block, nsect);
-
 	if (!blk_fs_request(req))
 		return -EIO;
 
@@ -63,6 +51,9 @@ static int do_blktrans_request(struct mtd_blktrans_ops *tr,
 	    get_capacity(req->rq_disk))
 		return -EIO;
 
+	if (blk_discard_rq(req))
+		return tr->discard(dev, block, nsect);
+
 	switch(rq_data_dir(req)) {
 	case READ:
 		for (; nsect > 0; nsect--, block++, buf += tr->blksize)
@@ -380,8 +371,8 @@ int register_mtd_blktrans(struct mtd_blktrans_ops *tr)
 	tr->blkcore_priv->rq->queuedata = tr;
 	blk_queue_logical_block_size(tr->blkcore_priv->rq, tr->blksize);
 	if (tr->discard)
-		blk_queue_set_discard(tr->blkcore_priv->rq,
-				      blktrans_discard_request);
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD,
+					tr->blkcore_priv->rq);
 
 	tr->blkshift = ffs(tr->blksize) - 1;
 
diff --git a/drivers/staging/dst/dcore.c b/drivers/staging/dst/dcore.c
index ac8577358ba0..5e8db0677582 100644
--- a/drivers/staging/dst/dcore.c
+++ b/drivers/staging/dst/dcore.c
@@ -102,7 +102,7 @@ static int dst_request(struct request_queue *q, struct bio *bio)
 	struct dst_node *n = q->queuedata;
 	int err = -EIO;
 
-	if (bio_empty_barrier(bio) && !q->prepare_discard_fn) {
+	if (bio_empty_barrier(bio) && !blk_queue_discard(q)) {
 		/*
 		 * This is a dirty^Wnice hack, but if we complete this
 		 * operation with -EOPNOTSUPP like intended, XFS
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index e23a86cae5ac..f62d45e87618 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -82,7 +82,6 @@ enum rq_cmd_type_bits {
 enum {
 	REQ_LB_OP_EJECT	= 0x40,		/* eject request */
 	REQ_LB_OP_FLUSH = 0x41,		/* flush request */
-	REQ_LB_OP_DISCARD = 0x42,	/* discard sectors */
 };
 
 /*
@@ -261,7 +260,6 @@ typedef void (request_fn_proc) (struct request_queue *q);
 typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
 typedef void (unplug_fn) (struct request_queue *);
-typedef int (prepare_discard_fn) (struct request_queue *, struct request *);
 
 struct bio_vec;
 struct bvec_merge_data {
@@ -340,7 +338,6 @@ struct request_queue
 	make_request_fn		*make_request_fn;
 	prep_rq_fn		*prep_rq_fn;
 	unplug_fn		*unplug_fn;
-	prepare_discard_fn	*prepare_discard_fn;
 	merge_bvec_fn		*merge_bvec_fn;
 	prepare_flush_fn	*prepare_flush_fn;
 	softirq_done_fn		*softirq_done_fn;
@@ -460,6 +457,7 @@ struct request_queue
 #define QUEUE_FLAG_VIRT        QUEUE_FLAG_NONROT /* paravirt device */
 #define QUEUE_FLAG_IO_STAT     15	/* do IO stats */
 #define QUEUE_FLAG_CQ	       16	/* hardware does queuing */
+#define QUEUE_FLAG_DISCARD     17	/* supports DISCARD */
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_CLUSTER) |		\
@@ -591,6 +589,7 @@ enum {
 #define blk_queue_flushing(q)	((q)->ordseq)
 #define blk_queue_stackable(q)	\
 	test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
+#define blk_queue_discard(q)	test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
 
 #define blk_fs_request(rq)	((rq)->cmd_type == REQ_TYPE_FS)
 #define blk_pc_request(rq)	((rq)->cmd_type == REQ_TYPE_BLOCK_PC)
@@ -955,7 +954,6 @@ extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
 extern void blk_queue_dma_alignment(struct request_queue *, int);
 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
-extern void blk_queue_set_discard(struct request_queue *, prepare_discard_fn *);
 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
-- 
cgit v1.2.3


From 67efc9258010da35b27b3854d0880c7e193004ed Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <hch@infradead.org>
Date: Wed, 30 Sep 2009 13:54:20 +0200
Subject: block: allow large discard requests

Currently we set the bio size to the byte equivalent of the blocks to
be trimmed when submitting the initial DISCARD ioctl.  That means it
is subject to the max_hw_sectors limitation of the HBA which is
much lower than the size of a DISCARD request we can support.
Add a separate max_discard_sectors tunable to limit the size for discard
requests.

We limit the max discard request size in bytes to 32bit as that is the
limit for bio->bi_size.  This could be much larger if we had a way to pass
that information through the block layer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-barrier.c    | 10 ++++++----
 block/blk-core.c       |  3 ++-
 block/blk-settings.c   | 13 +++++++++++++
 include/linux/blkdev.h |  3 +++
 4 files changed, 24 insertions(+), 5 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-barrier.c b/block/blk-barrier.c
index 21f5025c3945..8873b9b439ff 100644
--- a/block/blk-barrier.c
+++ b/block/blk-barrier.c
@@ -385,6 +385,8 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 
 	while (nr_sects && !ret) {
 		unsigned int sector_size = q->limits.logical_block_size;
+		unsigned int max_discard_sectors =
+			min(q->limits.max_discard_sectors, UINT_MAX >> 9);
 
 		bio = bio_alloc(gfp_mask, 1);
 		if (!bio)
@@ -411,10 +413,10 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 		 * touch many more blocks on disk than the actual payload
 		 * length.
 		 */
-		if (nr_sects > queue_max_hw_sectors(q)) {
-			bio->bi_size = queue_max_hw_sectors(q) << 9;
-			nr_sects -= queue_max_hw_sectors(q);
-			sector += queue_max_hw_sectors(q);
+		if (nr_sects > max_discard_sectors) {
+			bio->bi_size = max_discard_sectors << 9;
+			nr_sects -= max_discard_sectors;
+			sector += max_discard_sectors;
 		} else {
 			bio->bi_size = nr_sects << 9;
 			nr_sects = 0;
diff --git a/block/blk-core.c b/block/blk-core.c
index 80a020dd1580..34504f309728 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1436,7 +1436,8 @@ static inline void __generic_make_request(struct bio *bio)
 			goto end_io;
 		}
 
-		if (unlikely(nr_sectors > queue_max_hw_sectors(q))) {
+		if (unlikely(!bio_rw_flagged(bio, BIO_RW_DISCARD) &&
+			     nr_sectors > queue_max_hw_sectors(q))) {
 			printk(KERN_ERR "bio too big device %s (%u > %u)\n",
 			       bdevname(bio->bi_bdev, b),
 			       bio_sectors(bio),
diff --git a/block/blk-settings.c b/block/blk-settings.c
index d29498ef1eb5..e0695bca7027 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -96,6 +96,7 @@ void blk_set_default_limits(struct queue_limits *lim)
 	lim->max_segment_size = MAX_SEGMENT_SIZE;
 	lim->max_sectors = BLK_DEF_MAX_SECTORS;
 	lim->max_hw_sectors = INT_MAX;
+	lim->max_discard_sectors = SAFE_MAX_SECTORS;
 	lim->logical_block_size = lim->physical_block_size = lim->io_min = 512;
 	lim->bounce_pfn = (unsigned long)(BLK_BOUNCE_ANY >> PAGE_SHIFT);
 	lim->alignment_offset = 0;
@@ -238,6 +239,18 @@ void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_sectors)
 }
 EXPORT_SYMBOL(blk_queue_max_hw_sectors);
 
+/**
+ * blk_queue_max_discard_sectors - set max sectors for a single discard
+ * @q:  the request queue for the device
+ * @max_discard: maximum number of sectors to discard
+ **/
+void blk_queue_max_discard_sectors(struct request_queue *q,
+		unsigned int max_discard_sectors)
+{
+	q->limits.max_discard_sectors = max_discard_sectors;
+}
+EXPORT_SYMBOL(blk_queue_max_discard_sectors);
+
 /**
  * blk_queue_max_phys_segments - set max phys segments for a request for this queue
  * @q:  the request queue for the device
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index f62d45e87618..1a03b715dfad 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -311,6 +311,7 @@ struct queue_limits {
 	unsigned int		alignment_offset;
 	unsigned int		io_min;
 	unsigned int		io_opt;
+	unsigned int		max_discard_sectors;
 
 	unsigned short		logical_block_size;
 	unsigned short		max_hw_segments;
@@ -928,6 +929,8 @@ extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
 extern void blk_queue_max_phys_segments(struct request_queue *, unsigned short);
 extern void blk_queue_max_hw_segments(struct request_queue *, unsigned short);
 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
+extern void blk_queue_max_discard_sectors(struct request_queue *q,
+		unsigned int max_discard_sectors);
 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
 extern void blk_queue_physical_block_size(struct request_queue *, unsigned short);
 extern void blk_queue_alignment_offset(struct request_queue *q,
-- 
cgit v1.2.3


From 8e2967555571659d2c8a70dd120710110ed7bba4 Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Sat, 3 Oct 2009 16:26:03 +0200
Subject: cfq-iosched: implement slower async initiate and queue ramp up

This slowly ramps up the async queue depth based on the time
passed since the sync IO, and doesn't allow async at all until
a sync slice period has passed.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c       |  8 ++++++++
 block/cfq-iosched.c    | 56 ++++++++++++++++++++++++++++++++------------------
 include/linux/blkdev.h |  4 ++++
 3 files changed, 48 insertions(+), 20 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/blk-core.c b/block/blk-core.c
index ddaaea4fdffc..a8c7fbe52e24 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -2492,6 +2492,14 @@ int kblockd_schedule_work(struct request_queue *q, struct work_struct *work)
 }
 EXPORT_SYMBOL(kblockd_schedule_work);
 
+int kblockd_schedule_delayed_work(struct request_queue *q,
+				  struct delayed_work *work,
+				  unsigned long delay)
+{
+	return queue_delayed_work(kblockd_workqueue, work, delay);
+}
+EXPORT_SYMBOL(kblockd_schedule_delayed_work);
+
 int __init blk_dev_init(void)
 {
 	BUILD_BUG_ON(__REQ_NR_BITS > 8 *
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 70b48ea0e3e9..fce8a749f4be 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -150,7 +150,7 @@ struct cfq_data {
 	 * idle window management
 	 */
 	struct timer_list idle_slice_timer;
-	struct work_struct unplug_work;
+	struct delayed_work unplug_work;
 
 	struct cfq_queue *active_queue;
 	struct cfq_io_context *active_cic;
@@ -268,11 +268,13 @@ static inline int cfq_bio_sync(struct bio *bio)
  * scheduler run of queue, if there are requests pending and no one in the
  * driver that will restart queueing
  */
-static inline void cfq_schedule_dispatch(struct cfq_data *cfqd)
+static inline void cfq_schedule_dispatch(struct cfq_data *cfqd,
+					 unsigned long delay)
 {
 	if (cfqd->busy_queues) {
 		cfq_log(cfqd, "schedule dispatch");
-		kblockd_schedule_work(cfqd->queue, &cfqd->unplug_work);
+		kblockd_schedule_delayed_work(cfqd->queue, &cfqd->unplug_work,
+						delay);
 	}
 }
 
@@ -1316,8 +1318,6 @@ static int cfq_dispatch_requests(struct request_queue *q, int force)
 	 * Does this cfqq already have too much IO in flight?
 	 */
 	if (cfqq->dispatched >= max_dispatch) {
-		unsigned long load_at = cfqd->last_end_sync_rq + cfq_slice_sync;
-
 		/*
 		 * idle queue must always only have a single IO in flight
 		 */
@@ -1331,20 +1331,36 @@ static int cfq_dispatch_requests(struct request_queue *q, int force)
 			return 0;
 
 		/*
-		 * If a sync request has completed recently, don't overload
-		 * the dispatch queue yet with async requests.
+		 * Sole queue user, allow bigger slice
 		 */
-		if (cfqd->cfq_desktop && !cfq_cfqq_sync(cfqq)
-		    && time_before(jiffies, load_at))
-			return 0;
+		max_dispatch *= 4;
+	}
+
+	/*
+	 * Async queues must wait a bit before being allowed dispatch.
+	 * We also ramp up the dispatch depth gradually for async IO,
+	 * based on the last sync IO we serviced
+	 */
+	if (!cfq_cfqq_sync(cfqq) && cfqd->cfq_desktop) {
+		unsigned long last_sync = jiffies - cfqd->last_end_sync_rq;
+		unsigned int depth;
 
 		/*
-		 * we are the only queue, allow up to 4 times of 'quantum'
+		 * must wait a bit longer
 		 */
-		if (cfqq->dispatched >= 4 * max_dispatch)
+		if (last_sync < cfq_slice_sync) {
+			cfq_schedule_dispatch(cfqd, cfq_slice_sync - last_sync);
 			return 0;
+		}
+
+		depth = last_sync / cfq_slice_sync;
+		if (depth < max_dispatch)
+			max_dispatch = depth;
 	}
 
+	if (cfqq->dispatched >= max_dispatch)
+		return 0;
+
 	/*
 	 * Dispatch a request from this cfqq
 	 */
@@ -1389,7 +1405,7 @@ static void cfq_put_queue(struct cfq_queue *cfqq)
 
 	if (unlikely(cfqd->active_queue == cfqq)) {
 		__cfq_slice_expired(cfqd, cfqq, 0);
-		cfq_schedule_dispatch(cfqd);
+		cfq_schedule_dispatch(cfqd, 0);
 	}
 
 	kmem_cache_free(cfq_pool, cfqq);
@@ -1484,7 +1500,7 @@ static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 {
 	if (unlikely(cfqq == cfqd->active_queue)) {
 		__cfq_slice_expired(cfqd, cfqq, 0);
-		cfq_schedule_dispatch(cfqd);
+		cfq_schedule_dispatch(cfqd, 0);
 	}
 
 	cfq_put_queue(cfqq);
@@ -2201,7 +2217,7 @@ static void cfq_completed_request(struct request_queue *q, struct request *rq)
 	}
 
 	if (!rq_in_driver(cfqd))
-		cfq_schedule_dispatch(cfqd);
+		cfq_schedule_dispatch(cfqd, 0);
 }
 
 /*
@@ -2331,7 +2347,7 @@ queue_fail:
 	if (cic)
 		put_io_context(cic->ioc);
 
-	cfq_schedule_dispatch(cfqd);
+	cfq_schedule_dispatch(cfqd, 0);
 	spin_unlock_irqrestore(q->queue_lock, flags);
 	cfq_log(cfqd, "set_request fail");
 	return 1;
@@ -2340,7 +2356,7 @@ queue_fail:
 static void cfq_kick_queue(struct work_struct *work)
 {
 	struct cfq_data *cfqd =
-		container_of(work, struct cfq_data, unplug_work);
+		container_of(work, struct cfq_data, unplug_work.work);
 	struct request_queue *q = cfqd->queue;
 
 	spin_lock_irq(q->queue_lock);
@@ -2394,7 +2410,7 @@ static void cfq_idle_slice_timer(unsigned long data)
 expire:
 	cfq_slice_expired(cfqd, timed_out);
 out_kick:
-	cfq_schedule_dispatch(cfqd);
+	cfq_schedule_dispatch(cfqd, 0);
 out_cont:
 	spin_unlock_irqrestore(cfqd->queue->queue_lock, flags);
 }
@@ -2402,7 +2418,7 @@ out_cont:
 static void cfq_shutdown_timer_wq(struct cfq_data *cfqd)
 {
 	del_timer_sync(&cfqd->idle_slice_timer);
-	cancel_work_sync(&cfqd->unplug_work);
+	cancel_delayed_work_sync(&cfqd->unplug_work);
 }
 
 static void cfq_put_async_queues(struct cfq_data *cfqd)
@@ -2484,7 +2500,7 @@ static void *cfq_init_queue(struct request_queue *q)
 	cfqd->idle_slice_timer.function = cfq_idle_slice_timer;
 	cfqd->idle_slice_timer.data = (unsigned long) cfqd;
 
-	INIT_WORK(&cfqd->unplug_work, cfq_kick_queue);
+	INIT_DELAYED_WORK(&cfqd->unplug_work, cfq_kick_queue);
 
 	cfqd->cfq_quantum = cfq_quantum;
 	cfqd->cfq_fifo_expire[0] = cfq_fifo_expire[0];
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 1a03b715dfad..a7323930d2ba 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1147,7 +1147,11 @@ static inline void put_dev_sector(Sector p)
 }
 
 struct work_struct;
+struct delayed_work;
 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
+int kblockd_schedule_delayed_work(struct request_queue *q,
+					struct delayed_work *work,
+				 	unsigned long delay);
 
 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
 	MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
-- 
cgit v1.2.3


From ac481c20ef8f6c6f2be75d581863f40c43874ef7 Mon Sep 17 00:00:00 2001
From: "Martin K. Petersen" <martin.petersen@oracle.com>
Date: Sat, 3 Oct 2009 20:52:01 +0200
Subject: block: Topology ioctls

Not all users of the topology information want to use libblkid.  Provide
the topology information through bdev ioctls.

Also clarify sector size comments for existing BLK ioctls.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/compat_ioctl.c   | 13 +++++++++++++
 block/ioctl.c          | 17 +++++++++++++++--
 include/linux/blkdev.h | 35 ++++++++++++++++++++++++++++++-----
 include/linux/fs.h     |  4 ++++
 4 files changed, 62 insertions(+), 7 deletions(-)

(limited to 'include/linux/blkdev.h')

diff --git a/block/compat_ioctl.c b/block/compat_ioctl.c
index 7865a34e0faa..9bd086c1a4d5 100644
--- a/block/compat_ioctl.c
+++ b/block/compat_ioctl.c
@@ -21,6 +21,11 @@ static int compat_put_int(unsigned long arg, int val)
 	return put_user(val, (compat_int_t __user *)compat_ptr(arg));
 }
 
+static int compat_put_uint(unsigned long arg, unsigned int val)
+{
+	return put_user(val, (compat_uint_t __user *)compat_ptr(arg));
+}
+
 static int compat_put_long(unsigned long arg, long val)
 {
 	return put_user(val, (compat_long_t __user *)compat_ptr(arg));
@@ -734,6 +739,14 @@ long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg)
 	switch (cmd) {
 	case HDIO_GETGEO:
 		return compat_hdio_getgeo(disk, bdev, compat_ptr(arg));
+	case BLKPBSZGET:
+		return compat_put_uint(arg, bdev_physical_block_size(bdev));
+	case BLKIOMIN:
+		return compat_put_uint(arg, bdev_io_min(bdev));
+	case BLKIOOPT:
+		return compat_put_uint(arg, bdev_io_opt(bdev));
+	case BLKALIGNOFF:
+		return compat_put_int(arg, bdev_alignment_offset(bdev));
 	case BLKFLSBUF:
 	case BLKROSET:
 	case BLKDISCARD:
diff --git a/block/ioctl.c b/block/ioctl.c
index d3e6b5827a34..1f4d1de12b09 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -138,6 +138,11 @@ static int put_int(unsigned long arg, int val)
 	return put_user(val, (int __user *)arg);
 }
 
+static int put_uint(unsigned long arg, unsigned int val)
+{
+	return put_user(val, (unsigned int __user *)arg);
+}
+
 static int put_long(unsigned long arg, long val)
 {
 	return put_user(val, (long __user *)arg);
@@ -263,10 +268,18 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
 		return put_long(arg, (bdi->ra_pages * PAGE_CACHE_SIZE) / 512);
 	case BLKROGET:
 		return put_int(arg, bdev_read_only(bdev) != 0);
-	case BLKBSZGET: /* get the logical block size (cf. BLKSSZGET) */
+	case BLKBSZGET: /* get block device soft block size (cf. BLKSSZGET) */
 		return put_int(arg, block_size(bdev));
-	case BLKSSZGET: /* get block device hardware sector size */
+	case BLKSSZGET: /* get block device logical block size */
 		return put_int(arg, bdev_logical_block_size(bdev));
+	case BLKPBSZGET: /* get block device physical block size */
+		return put_uint(arg, bdev_physical_block_size(bdev));
+	case BLKIOMIN:
+		return put_uint(arg, bdev_io_min(bdev));
+	case BLKIOOPT:
+		return put_uint(arg, bdev_io_opt(bdev));
+	case BLKALIGNOFF:
+		return put_int(arg, bdev_alignment_offset(bdev));
 	case BLKSECTGET:
 		return put_ushort(arg, queue_max_sectors(bdev_get_queue(bdev)));
 	case BLKRASET:
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index a7323930d2ba..25119041e034 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1081,25 +1081,37 @@ static inline unsigned int queue_physical_block_size(struct request_queue *q)
 	return q->limits.physical_block_size;
 }
 
+static inline int bdev_physical_block_size(struct block_device *bdev)
+{
+	return queue_physical_block_size(bdev_get_queue(bdev));
+}
+
 static inline unsigned int queue_io_min(struct request_queue *q)
 {
 	return q->limits.io_min;
 }
 
+static inline int bdev_io_min(struct block_device *bdev)
+{
+	return queue_io_min(bdev_get_queue(bdev));
+}
+
 static inline unsigned int queue_io_opt(struct request_queue *q)
 {
 	return q->limits.io_opt;
 }
 
+static inline int bdev_io_opt(struct block_device *bdev)
+{
+	return queue_io_opt(bdev_get_queue(bdev));
+}
+
 static inline int queue_alignment_offset(struct request_queue *q)
 {
-	if (q && q->limits.misaligned)
+	if (q->limits.misaligned)
 		return -1;
 
-	if (q && q->limits.alignment_offset)
-		return q->limits.alignment_offset;
-
-	return 0;
+	return q->limits.alignment_offset;
 }
 
 static inline int queue_sector_alignment_offset(struct request_queue *q,
@@ -1109,6 +1121,19 @@ static inline int queue_sector_alignment_offset(struct request_queue *q,
 		& (q->limits.io_min - 1);
 }
 
+static inline int bdev_alignment_offset(struct block_device *bdev)
+{
+	struct request_queue *q = bdev_get_queue(bdev);
+
+	if (q->limits.misaligned)
+		return -1;
+
+	if (bdev != bdev->bd_contains)
+		return bdev->bd_part->alignment_offset;
+
+	return q->limits.alignment_offset;
+}
+
 static inline int queue_dma_alignment(struct request_queue *q)
 {
 	return q ? q->dma_alignment : 511;
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 2adaa2529f18..883eaacfd924 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -300,6 +300,10 @@ struct inodes_stat_t {
 #define BLKTRACESTOP _IO(0x12,117)
 #define BLKTRACETEARDOWN _IO(0x12,118)
 #define BLKDISCARD _IO(0x12,119)
+#define BLKIOMIN _IO(0x12,120)
+#define BLKIOOPT _IO(0x12,121)
+#define BLKALIGNOFF _IO(0x12,122)
+#define BLKPBSZGET _IO(0x12,123)
 
 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
-- 
cgit v1.2.3