From bf0f97025c7306870b86ccd63669aa278e7ec1c2 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:09:39 +0200
Subject: block: sysfs store function needs to grab queue_lock and use
 queue_flag_*()

Concurrency isn't a big deal here since we have requests in flight
at this point, but do the locked variant to set a better example.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-sysfs.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index e85c4013e8a2..304ec73ab821 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -146,11 +146,13 @@ static ssize_t queue_nomerges_store(struct request_queue *q, const char *page,
 	unsigned long nm;
 	ssize_t ret = queue_var_store(&nm, page, count);
 
+	spin_lock_irq(q->queue_lock);
 	if (nm)
-	       set_bit(QUEUE_FLAG_NOMERGES, &q->queue_flags);
+		queue_flag_set(QUEUE_FLAG_NOMERGES, q);
 	else
-	       clear_bit(QUEUE_FLAG_NOMERGES, &q->queue_flags);
+		queue_flag_clear(QUEUE_FLAG_NOMERGES, q);
 
+	spin_unlock_irq(q->queue_lock);
 	return ret;
 }
 
-- 
cgit v1.2.3


From aa94b5371f6f898558d9fa5690cc6e4bf917a572 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:27:43 +0200
Subject: block: adjust tagging function queue bit locking

For most initialization purposes, calling blk_queue_init_tags() without
the queue lock held is OK. Only if called for resizing an existing map
must the lock be held. Ditto for tag cleanup, the maps are reference
counted.

So switch the general queue flag setting to the unlocked variant, but
retain the locked variant for resizing.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-tag.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/block/blk-tag.c b/block/blk-tag.c
index de64e0429977..32667beb03ee 100644
--- a/block/blk-tag.c
+++ b/block/blk-tag.c
@@ -70,7 +70,7 @@ void __blk_queue_free_tags(struct request_queue *q)
 	__blk_free_tags(bqt);
 
 	q->queue_tags = NULL;
-	queue_flag_clear(QUEUE_FLAG_QUEUED, q);
+	queue_flag_clear_unlocked(QUEUE_FLAG_QUEUED, q);
 }
 
 /**
@@ -98,7 +98,7 @@ EXPORT_SYMBOL(blk_free_tags);
  **/
 void blk_queue_free_tags(struct request_queue *q)
 {
-	queue_flag_clear(QUEUE_FLAG_QUEUED, q);
+	queue_flag_clear_unlocked(QUEUE_FLAG_QUEUED, q);
 }
 EXPORT_SYMBOL(blk_queue_free_tags);
 
@@ -171,6 +171,9 @@ EXPORT_SYMBOL(blk_init_tags);
  * @q:  the request queue for the device
  * @depth:  the maximum queue depth supported
  * @tags: the tag to use
+ *
+ * Queue lock must be held here if the function is called to resize an
+ * existing map.
  **/
 int blk_queue_init_tags(struct request_queue *q, int depth,
 			struct blk_queue_tag *tags)
@@ -197,7 +200,7 @@ int blk_queue_init_tags(struct request_queue *q, int depth,
 	 * assign it, all done
 	 */
 	q->queue_tags = tags;
-	queue_flag_set(QUEUE_FLAG_QUEUED, q);
+	queue_flag_set_unlocked(QUEUE_FLAG_QUEUED, q);
 	INIT_LIST_HEAD(&q->tag_busy_list);
 	return 0;
 fail:
-- 
cgit v1.2.3


From 07416d29bcf608257f1e5280642dcbe0021518a3 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:17:12 +0200
Subject: cfq-iosched: fix RCU race in the cfq io_context destructor handling

put_io_context() drops the RCU read lock before calling into cfq_dtor(),
however we need to hold off freeing there before grabbing and
dereferencing the first object on the list.

So extend the rcu_read_lock() scope to cover the calling of cfq_dtor(),
and optimize cfq_free_io_context() to use a new variant for
call_for_each_cic() that assumes the RCU read lock is already held.

Hit in the wild by Alexey Dobriyan <adobriyan@gmail.com>

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-ioc.c     |  2 +-
 block/cfq-iosched.c | 19 +++++++++++++------
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/block/blk-ioc.c b/block/blk-ioc.c
index e34df7c9fc36..012f065ac8e2 100644
--- a/block/blk-ioc.c
+++ b/block/blk-ioc.c
@@ -41,8 +41,8 @@ int put_io_context(struct io_context *ioc)
 		rcu_read_lock();
 		if (ioc->aic && ioc->aic->dtor)
 			ioc->aic->dtor(ioc->aic);
-		rcu_read_unlock();
 		cfq_dtor(ioc);
+		rcu_read_unlock();
 
 		kmem_cache_free(iocontext_cachep, ioc);
 		return 1;
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index f4e1006c253d..7f909d2f4886 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1142,6 +1142,17 @@ static void cfq_put_queue(struct cfq_queue *cfqq)
 	kmem_cache_free(cfq_pool, cfqq);
 }
 
+static void
+__call_for_each_cic(struct io_context *ioc,
+		    void (*func)(struct io_context *, struct cfq_io_context *))
+{
+	struct cfq_io_context *cic;
+	struct hlist_node *n;
+
+	hlist_for_each_entry_rcu(cic, n, &ioc->cic_list, cic_list)
+		func(ioc, cic);
+}
+
 /*
  * Call func for each cic attached to this ioc.
  */
@@ -1149,12 +1160,8 @@ static void
 call_for_each_cic(struct io_context *ioc,
 		  void (*func)(struct io_context *, struct cfq_io_context *))
 {
-	struct cfq_io_context *cic;
-	struct hlist_node *n;
-
 	rcu_read_lock();
-	hlist_for_each_entry_rcu(cic, n, &ioc->cic_list, cic_list)
-		func(ioc, cic);
+	__call_for_each_cic(ioc, func);
 	rcu_read_unlock();
 }
 
@@ -1198,7 +1205,7 @@ static void cfq_free_io_context(struct io_context *ioc)
 	 * should be ok to iterate over the known list, we will see all cic's
 	 * since no new ones are added.
 	 */
-	call_for_each_cic(ioc, cic_free_func);
+	__call_for_each_cic(ioc, cic_free_func);
 }
 
 static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
-- 
cgit v1.2.3


From 7f3d4ee108c184ab215036051087aaaaa8de7661 Mon Sep 17 00:00:00 2001
From: Miklos Szeredi
Date: Wed, 7 May 2008 09:22:39 +0200
Subject: vfs: splice remove_suid() cleanup

generic_file_splice_write() duplicates remove_suid() just because it
doesn't hold i_mutex.  But it grabs i_mutex inside splice_from_pipe()
anyway, so this is rather pointless.

Move locking to generic_file_splice_write() and call remove_suid() and
__splice_from_pipe() instead.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 fs/splice.c        | 29 +++++++++++++----------------
 include/linux/fs.h |  1 -
 mm/filemap.c       |  2 +-
 3 files changed, 14 insertions(+), 18 deletions(-)

diff --git a/fs/splice.c b/fs/splice.c
index 633f58ebfb72..cece15b4ef72 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -811,24 +811,19 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
 {
 	struct address_space *mapping = out->f_mapping;
 	struct inode *inode = mapping->host;
-	int killsuid, killpriv;
+	struct splice_desc sd = {
+		.total_len = len,
+		.flags = flags,
+		.pos = *ppos,
+		.u.file = out,
+	};
 	ssize_t ret;
-	int err = 0;
-
-	killpriv = security_inode_need_killpriv(out->f_path.dentry);
-	killsuid = should_remove_suid(out->f_path.dentry);
-	if (unlikely(killsuid || killpriv)) {
-		mutex_lock(&inode->i_mutex);
-		if (killpriv)
-			err = security_inode_killpriv(out->f_path.dentry);
-		if (!err && killsuid)
-			err = __remove_suid(out->f_path.dentry, killsuid);
-		mutex_unlock(&inode->i_mutex);
-		if (err)
-			return err;
-	}
 
-	ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
+	inode_double_lock(inode, pipe->inode);
+	ret = remove_suid(out->f_path.dentry);
+	if (likely(!ret))
+		ret = __splice_from_pipe(pipe, &sd, pipe_to_file);
+	inode_double_unlock(inode, pipe->inode);
 	if (ret > 0) {
 		unsigned long nr_pages;
 
@@ -840,6 +835,8 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
 		 * sync it.
 		 */
 		if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) {
+			int err;
+
 			mutex_lock(&inode->i_mutex);
 			err = generic_osync_inode(inode, mapping,
 						  OSYNC_METADATA|OSYNC_DATA);
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 7e0fa9e64479..f413085f748e 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1816,7 +1816,6 @@ extern void iget_failed(struct inode *);
 extern void clear_inode(struct inode *);
 extern void destroy_inode(struct inode *);
 extern struct inode *new_inode(struct super_block *);
-extern int __remove_suid(struct dentry *, int);
 extern int should_remove_suid(struct dentry *);
 extern int remove_suid(struct dentry *);
 
diff --git a/mm/filemap.c b/mm/filemap.c
index 239d36163bbe..2dead9adf8b7 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1655,7 +1655,7 @@ int should_remove_suid(struct dentry *dentry)
 }
 EXPORT_SYMBOL(should_remove_suid);
 
-int __remove_suid(struct dentry *dentry, int kill)
+static int __remove_suid(struct dentry *dentry, int kill)
 {
 	struct iattr newattrs;
 
-- 
cgit v1.2.3


From 2cdf79cafbd11580f5b63cd4993b45c1c4952415 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:33:55 +0200
Subject: block: get rid of likely/unlikely predictions in merge logic

They tend to depend a lot on the workload, so not a clear-cut
likely or unlikely fit.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-merge.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/block/blk-merge.c b/block/blk-merge.c
index 73b23562af20..651136aae76e 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -149,9 +149,9 @@ static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
 static int blk_hw_contig_segment(struct request_queue *q, struct bio *bio,
 				 struct bio *nxt)
 {
-	if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
+	if (!bio_flagged(bio, BIO_SEG_VALID))
 		blk_recount_segments(q, bio);
-	if (unlikely(!bio_flagged(nxt, BIO_SEG_VALID)))
+	if (!bio_flagged(nxt, BIO_SEG_VALID))
 		blk_recount_segments(q, nxt);
 	if (!BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)) ||
 	    BIOVEC_VIRT_OVERSIZE(bio->bi_hw_back_size + nxt->bi_hw_front_size))
@@ -312,9 +312,9 @@ int ll_back_merge_fn(struct request_queue *q, struct request *req,
 			q->last_merge = NULL;
 		return 0;
 	}
-	if (unlikely(!bio_flagged(req->biotail, BIO_SEG_VALID)))
+	if (!bio_flagged(req->biotail, BIO_SEG_VALID))
 		blk_recount_segments(q, req->biotail);
-	if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
+	if (!bio_flagged(bio, BIO_SEG_VALID))
 		blk_recount_segments(q, bio);
 	len = req->biotail->bi_hw_back_size + bio->bi_hw_front_size;
 	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio))
@@ -352,9 +352,9 @@ int ll_front_merge_fn(struct request_queue *q, struct request *req,
 		return 0;
 	}
 	len = bio->bi_hw_back_size + req->bio->bi_hw_front_size;
-	if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
+	if (!bio_flagged(bio, BIO_SEG_VALID))
 		blk_recount_segments(q, bio);
-	if (unlikely(!bio_flagged(req->bio, BIO_SEG_VALID)))
+	if (!bio_flagged(req->bio, BIO_SEG_VALID))
 		blk_recount_segments(q, req->bio);
 	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)) &&
 	    !BIOVEC_VIRT_OVERSIZE(len)) {
-- 
cgit v1.2.3


From dbaf2c003e151ad9231778819b0977f95e20e06f Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:48:17 +0200
Subject: block: optimize generic_unplug_device()

Original patch from Mikulas Patocka <mpatocka@redhat.com>

Mike Anderson was doing an OLTP benchmark on a computer with 48 physical
disks mapped to one logical device via device mapper.

He found that there was a slowdown on request_queue->lock in function
generic_unplug_device. The slowdown is caused by the fact that when some
code calls unplug on the device mapper, device mapper calls unplug on all
physical disks. These unplug calls take the lock, find that the queue is
already unplugged, release the lock and exit.

With the below patch, performance of the benchmark was increased by 18%
(the whole OLTP application, not just block layer microbenchmarks).

So I'm submitting this patch for upstream. I think the patch is correct,
because when more threads call simultaneously plug and unplug, it is
unspecified, if the queue is or isn't plugged (so the patch can't make
this worse). And the caller that plugged the queue should unplug it
anyway. (if it doesn't, there's 3ms timeout).

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index b754a4a2f9bd..1b7dddf94f4f 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -253,9 +253,11 @@ EXPORT_SYMBOL(__generic_unplug_device);
  **/
 void generic_unplug_device(struct request_queue *q)
 {
-	spin_lock_irq(q->queue_lock);
-	__generic_unplug_device(q);
-	spin_unlock_irq(q->queue_lock);
+	if (blk_queue_plugged(q)) {
+		spin_lock_irq(q->queue_lock);
+		__generic_unplug_device(q);
+		spin_unlock_irq(q->queue_lock);
+	}
 }
 EXPORT_SYMBOL(generic_unplug_device);
 
-- 
cgit v1.2.3


From 6d63c275572d1e6f00d4fa154f16fbb0d8c2d2bf Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 09:51:23 +0200
Subject: cfq-iosched: make io priorities inherit CPU scheduling class as well
 as nice

We currently set all processes to the best-effort scheduling class,
regardless of what CPU scheduling class they belong to. Improve that
so that we correctly track idle and rt scheduling classes as well.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/cfq-iosched.c    |  4 ++--
 include/linux/ioprio.h | 14 ++++++++++++++
 2 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 7f909d2f4886..b399c62936e0 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1303,10 +1303,10 @@ static void cfq_init_prio_data(struct cfq_queue *cfqq, struct io_context *ioc)
 		printk(KERN_ERR "cfq: bad prio %x\n", ioprio_class);
 	case IOPRIO_CLASS_NONE:
 		/*
-		 * no prio set, place us in the middle of the BE classes
+		 * no prio set, inherit CPU scheduling settings
 		 */
 		cfqq->ioprio = task_nice_ioprio(tsk);
-		cfqq->ioprio_class = IOPRIO_CLASS_BE;
+		cfqq->ioprio_class = task_nice_ioclass(tsk);
 		break;
 	case IOPRIO_CLASS_RT:
 		cfqq->ioprio = task_ioprio(ioc);
diff --git a/include/linux/ioprio.h b/include/linux/ioprio.h
index 2a3bb1bb7433..f98a656b17e5 100644
--- a/include/linux/ioprio.h
+++ b/include/linux/ioprio.h
@@ -67,6 +67,20 @@ static inline int task_nice_ioprio(struct task_struct *task)
 	return (task_nice(task) + 20) / 5;
 }
 
+/*
+ * This is for the case where the task hasn't asked for a specific IO class.
+ * Check for idle and rt task process, and return appropriate IO class.
+ */
+static inline int task_nice_ioclass(struct task_struct *task)
+{
+	if (task->policy == SCHED_IDLE)
+		return IOPRIO_CLASS_IDLE;
+	else if (task->policy == SCHED_FIFO || task->policy == SCHED_RR)
+		return IOPRIO_CLASS_RT;
+	else
+		return IOPRIO_CLASS_BE;
+}
+
 /*
  * For inheritance, return the highest of the two given priorities
  */
-- 
cgit v1.2.3


From 28f13702f03e527fcb979747a882cf366c489c50 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 10:15:46 +0200
Subject: block: avoid duplicate calls to get_part() in disk stat code

get_part() is fairly expensive, as it O(N) loops over partitions
to find the right one. In lots of normal IO paths we end up looking
up the partition twice, to make matters even worse. Change the
stat add code to accept a passed in partition instead.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/blk-core.c           | 18 ++++++++++--------
 drivers/block/aoe/aoecmd.c | 10 ++++++----
 include/linux/genhd.h      | 35 ++++++++++++++++++-----------------
 3 files changed, 34 insertions(+), 29 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 1b7dddf94f4f..2987fe47b5ee 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -54,15 +54,16 @@ static DEFINE_PER_CPU(struct list_head, blk_cpu_done);
 
 static void drive_stat_acct(struct request *rq, int new_io)
 {
+	struct hd_struct *part;
 	int rw = rq_data_dir(rq);
 
 	if (!blk_fs_request(rq) || !rq->rq_disk)
 		return;
 
-	if (!new_io) {
-		__all_stat_inc(rq->rq_disk, merges[rw], rq->sector);
-	} else {
-		struct hd_struct *part = get_part(rq->rq_disk, rq->sector);
+	part = get_part(rq->rq_disk, rq->sector);
+	if (!new_io)
+		__all_stat_inc(rq->rq_disk, part, merges[rw], rq->sector);
+	else {
 		disk_round_stats(rq->rq_disk);
 		rq->rq_disk->in_flight++;
 		if (part) {
@@ -1538,10 +1539,11 @@ static int __end_that_request_first(struct request *req, int error,
 	}
 
 	if (blk_fs_request(req) && req->rq_disk) {
+		struct hd_struct *part = get_part(req->rq_disk, req->sector);
 		const int rw = rq_data_dir(req);
 
-		all_stat_add(req->rq_disk, sectors[rw],
-			     nr_bytes >> 9, req->sector);
+		all_stat_add(req->rq_disk, part, sectors[rw],
+				nr_bytes >> 9, req->sector);
 	}
 
 	total_bytes = bio_nbytes = 0;
@@ -1727,8 +1729,8 @@ static void end_that_request_last(struct request *req, int error)
 		const int rw = rq_data_dir(req);
 		struct hd_struct *part = get_part(disk, req->sector);
 
-		__all_stat_inc(disk, ios[rw], req->sector);
-		__all_stat_add(disk, ticks[rw], duration, req->sector);
+		__all_stat_inc(disk, part, ios[rw], req->sector);
+		__all_stat_add(disk, part, ticks[rw], duration, req->sector);
 		disk_round_stats(disk);
 		disk->in_flight--;
 		if (part) {
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
index 8fc429cf82b6..41f818be2f7e 100644
--- a/drivers/block/aoe/aoecmd.c
+++ b/drivers/block/aoe/aoecmd.c
@@ -755,11 +755,13 @@ diskstats(struct gendisk *disk, struct bio *bio, ulong duration, sector_t sector
 {
 	unsigned long n_sect = bio->bi_size >> 9;
 	const int rw = bio_data_dir(bio);
+	struct hd_struct *part;
 
-	all_stat_inc(disk, ios[rw], sector);
-	all_stat_add(disk, ticks[rw], duration, sector);
-	all_stat_add(disk, sectors[rw], n_sect, sector);
-	all_stat_add(disk, io_ticks, duration, sector);
+	part = get_part(disk, sector);
+	all_stat_inc(disk, part, ios[rw], sector);
+	all_stat_add(disk, part, ticks[rw], duration, sector);
+	all_stat_add(disk, part, sectors[rw], n_sect, sector);
+	all_stat_add(disk, part, io_ticks, duration, sector);
 }
 
 void
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index ecd2bf63fc84..e9874e7fcdf9 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -178,17 +178,17 @@ static inline struct hd_struct *get_part(struct gendisk *gendiskp,
 
 static inline void disk_stat_set_all(struct gendisk *gendiskp, int value)	{
 	int i;
+
 	for_each_possible_cpu(i)
 		memset(per_cpu_ptr(gendiskp->dkstats, i), value,
-				sizeof (struct disk_stats));
+				sizeof(struct disk_stats));
 }		
 
 #define __part_stat_add(part, field, addnd)				\
 	(per_cpu_ptr(part->dkstats, smp_processor_id())->field += addnd)
 
-#define __all_stat_add(gendiskp, field, addnd, sector)		\
+#define __all_stat_add(gendiskp, part, field, addnd, sector)	\
 ({								\
-	struct hd_struct *part = get_part(gendiskp, sector);	\
 	if (part)						\
 		__part_stat_add(part, field, addnd);		\
 	__disk_stat_add(gendiskp, field, addnd);		\
@@ -203,11 +203,13 @@ static inline void disk_stat_set_all(struct gendisk *gendiskp, int value)	{
 	res;								\
 })
 
-static inline void part_stat_set_all(struct hd_struct *part, int value)	{
+static inline void part_stat_set_all(struct hd_struct *part, int value)
+{
 	int i;
+
 	for_each_possible_cpu(i)
 		memset(per_cpu_ptr(part->dkstats, i), value,
-		       sizeof(struct disk_stats));
+				sizeof(struct disk_stats));
 }
 				
 #else /* !CONFIG_SMP */
@@ -223,9 +225,8 @@ static inline void disk_stat_set_all(struct gendisk *gendiskp, int value)
 #define __part_stat_add(part, field, addnd) \
 	(part->dkstats.field += addnd)
 
-#define __all_stat_add(gendiskp, field, addnd, sector)		\
+#define __all_stat_add(gendiskp, part, field, addnd, sector)	\
 ({								\
-	struct hd_struct *part = get_part(gendiskp, sector);	\
 	if (part)						\
 		part->dkstats.field += addnd;			\
 	__disk_stat_add(gendiskp, field, addnd);		\
@@ -276,10 +277,10 @@ static inline void part_stat_set_all(struct hd_struct *part, int value)
 #define part_stat_sub(gendiskp, field, subnd) \
 		part_stat_add(gendiskp, field, -subnd)
 
-#define all_stat_add(gendiskp, field, addnd, sector)		\
+#define all_stat_add(gendiskp, part, field, addnd, sector)	\
 	do {							\
 		preempt_disable();				\
-		__all_stat_add(gendiskp, field, addnd, sector);	\
+		__all_stat_add(gendiskp, part, field, addnd, sector);	\
 		preempt_enable();				\
 	} while (0)
 
@@ -288,15 +289,15 @@ static inline void part_stat_set_all(struct hd_struct *part, int value)
 #define all_stat_dec(gendiskp, field, sector) \
 		all_stat_add(gendiskp, field, -1, sector)
 
-#define __all_stat_inc(gendiskp, field, sector) \
-		__all_stat_add(gendiskp, field, 1, sector)
-#define all_stat_inc(gendiskp, field, sector) \
-		all_stat_add(gendiskp, field, 1, sector)
+#define __all_stat_inc(gendiskp, part, field, sector) \
+		__all_stat_add(gendiskp, part, field, 1, sector)
+#define all_stat_inc(gendiskp, part, field, sector) \
+		all_stat_add(gendiskp, part, field, 1, sector)
 
-#define __all_stat_sub(gendiskp, field, subnd, sector) \
-		__all_stat_add(gendiskp, field, -subnd, sector)
-#define all_stat_sub(gendiskp, field, subnd, sector) \
-		all_stat_add(gendiskp, field, -subnd, sector)
+#define __all_stat_sub(gendiskp, part, field, subnd, sector) \
+		__all_stat_add(gendiskp, part, field, -subnd, sector)
+#define all_stat_sub(gendiskp, part, field, subnd, sector) \
+		all_stat_add(gendiskp, part, field, -subnd, sector)
 
 /* Inlines to alloc and free disk stats in struct gendisk */
 #ifdef  CONFIG_SMP
-- 
cgit v1.2.3


From eeae1d48c011839d9e1cdc1e8aacf0193c9d8197 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Wed, 7 May 2008 13:26:27 +0200
Subject: block: use unitialized_var() in bio_alloc_bioset()

Better than setting idx to some random value and it silences the
same bogus gcc warning.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 fs/bio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/bio.c b/fs/bio.c
index 799f86deff24..2fa04ff86703 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -158,7 +158,7 @@ struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
 
 		bio_init(bio);
 		if (likely(nr_iovecs)) {
-			unsigned long idx = 0; /* shut up gcc */
+			unsigned long uninitialized_var(idx);
 
 			bvl = bvec_alloc_bs(gfp_mask, nr_iovecs, &idx, bs);
 			if (unlikely(!bvl)) {
-- 
cgit v1.2.3


From ffee0259c9edcc4c0f06b60df51c461eeecad4c0 Mon Sep 17 00:00:00 2001
From: Randy Dunlap
Date: Wed, 30 Apr 2008 09:08:54 +0200
Subject: docbook: fix bio missing parameter

Fix fs/bio.c kernel-doc parameter warning:
Warning(linux-2.6.25-git14//fs/bio.c:972): No description found for parameter 'reading'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 fs/bio.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/bio.c b/fs/bio.c
index 2fa04ff86703..78562574cb52 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -963,6 +963,7 @@ static void bio_copy_kern_endio(struct bio *bio, int err)
  *	@data: pointer to buffer to copy
  *	@len: length in bytes
  *	@gfp_mask: allocation flags for bio and page allocation
+ *	@reading: data direction is READ
  *
  *	copy the kernel address into a bio suitable for io to a block
  *	device. Returns an error pointer in case of error.
-- 
cgit v1.2.3


From 75065ff619e42fe35178eda863cbcddd57776794 Mon Sep 17 00:00:00 2001
From: Jens Axboe
Date: Thu, 8 May 2008 14:06:19 +0200
Subject: Revert "relay: fix splice problem"

This reverts commit c3270e577c18b3d0e984c3371493205a4807db9d.
---
 fs/splice.c    | 2 +-
 kernel/relay.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/splice.c b/fs/splice.c
index cece15b4ef72..78150038b584 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -1072,7 +1072,7 @@ long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
 
 	ret = splice_direct_to_actor(in, &sd, direct_splice_actor);
 	if (ret > 0)
-		*ppos = sd.pos;
+		*ppos += ret;
 
 	return ret;
 }
diff --git a/kernel/relay.c b/kernel/relay.c
index 7de644cdec43..bc24dcdc570f 100644
--- a/kernel/relay.c
+++ b/kernel/relay.c
@@ -1191,7 +1191,7 @@ static ssize_t relay_file_splice_read(struct file *in,
 	ret = 0;
 	spliced = 0;
 
-	while (len && !spliced) {
+	while (len) {
 		ret = subbuf_splice_actor(in, ppos, pipe, len, flags, &nonpad_ret);
 		if (ret < 0)
 			break;
-- 
cgit v1.2.3