[34-longterm 194/209] block: Deprecate QUEUE_FLAG_CLUSTER and use queue_limits instead

From: Paul Gortmaker
Date: Thu Apr 14 2011 - 14:04:03 EST


From: Martin K. Petersen <martin.petersen@xxxxxxxxxx>

=====================================================================
| This is a commit scheduled for the next v2.6.34 longterm release. |
| If you see a problem with using this for longterm, please comment.|
=====================================================================

commit e692cb668fdd5a712c6ed2a2d6f2a36ee83997b4 upstream.

When stacking devices, a request_queue is not always available. This
forced us to have a no_cluster flag in the queue_limits that could be
used as a carrier until the request_queue had been set up for a
metadevice.

There were several problems with that approach. First of all it was up
to the stacking device to remember to set queue flag after stacking had
completed. Also, the queue flag and the queue limits had to be kept in
sync at all times. We got that wrong, which could lead to us issuing
commands that went beyond the max scatterlist limit set by the driver.

The proper fix is to avoid having two flags for tracking the same thing.
We deprecate QUEUE_FLAG_CLUSTER and use the queue limit directly in the
block layer merging functions. The queue_limit 'no_cluster' is turned
into 'cluster' to avoid double negatives and to ease stacking.
Clustering defaults to being enabled as before. The queue flag logic is
removed from the stacking function, and explicitly setting the cluster
flag is no longer necessary in DM and MD.

Reported-by: Ed Lin <ed.lin@xxxxxxxxxxx>
Signed-off-by: Martin K. Petersen <martin.petersen@xxxxxxxxxx>
Acked-by: Mike Snitzer <snitzer@xxxxxxxxxx>
Signed-off-by: Jens Axboe <jaxboe@xxxxxxxxxxxx>
Signed-off-by: Paul Gortmaker <paul.gortmaker@xxxxxxxxxxxxx>
---
block/blk-merge.c | 6 +++---
block/blk-settings.c | 25 ++-----------------------
block/blk-sysfs.c | 2 +-
drivers/md/dm-table.c | 5 -----
drivers/md/md.c | 3 ---
drivers/scsi/scsi_lib.c | 3 +--
include/linux/blkdev.h | 9 ++++++---
7 files changed, 13 insertions(+), 40 deletions(-)

diff --git a/block/blk-merge.c b/block/blk-merge.c
index 5e7dc99..01d85b5 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -22,7 +22,7 @@ static unsigned int __blk_recalc_rq_segments(struct request_queue *q,
return 0;

fbio = bio;
- cluster = test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags);
+ cluster = blk_queue_cluster(q);
seg_size = 0;
phys_size = nr_phys_segs = 0;
for_each_bio(bio) {
@@ -88,7 +88,7 @@ EXPORT_SYMBOL(blk_recount_segments);
static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
struct bio *nxt)
{
- if (!test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags))
+ if (!blk_queue_cluster(q))
return 0;

if (bio->bi_seg_back_size + nxt->bi_seg_front_size >
@@ -124,7 +124,7 @@ int blk_rq_map_sg(struct request_queue *q, struct request *rq,
int nsegs, cluster;

nsegs = 0;
- cluster = test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags);
+ cluster = blk_queue_cluster(q);

/*
* for each bio in rq
diff --git a/block/blk-settings.c b/block/blk-settings.c
index 3430c1f..331763a 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -108,7 +108,7 @@ void blk_set_default_limits(struct queue_limits *lim)
lim->alignment_offset = 0;
lim->io_opt = 0;
lim->misaligned = 0;
- lim->no_cluster = 0;
+ lim->cluster = 1;
}
EXPORT_SYMBOL(blk_set_default_limits);

@@ -451,15 +451,6 @@ EXPORT_SYMBOL(blk_queue_io_opt);
void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b)
{
blk_stack_limits(&t->limits, &b->limits, 0);
-
- if (!t->queue_lock)
- WARN_ON_ONCE(1);
- else if (!test_bit(QUEUE_FLAG_CLUSTER, &b->queue_flags)) {
- unsigned long flags;
- spin_lock_irqsave(t->queue_lock, flags);
- queue_flag_clear(QUEUE_FLAG_CLUSTER, t);
- spin_unlock_irqrestore(t->queue_lock, flags);
- }
}
EXPORT_SYMBOL(blk_queue_stack_limits);

@@ -530,7 +521,7 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
t->io_min = max(t->io_min, b->io_min);
t->io_opt = lcm(t->io_opt, b->io_opt);

- t->no_cluster |= b->no_cluster;
+ t->cluster &= b->cluster;
t->discard_zeroes_data &= b->discard_zeroes_data;

/* Physical block size a multiple of the logical block size? */
@@ -626,7 +617,6 @@ void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
sector_t offset)
{
struct request_queue *t = disk->queue;
- struct request_queue *b = bdev_get_queue(bdev);

if (bdev_stack_limits(&t->limits, bdev, offset >> 9) < 0) {
char top[BDEVNAME_SIZE], bottom[BDEVNAME_SIZE];
@@ -637,17 +627,6 @@ void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
printk(KERN_NOTICE "%s: Warning: Device %s is misaligned\n",
top, bottom);
}
-
- if (!t->queue_lock)
- WARN_ON_ONCE(1);
- else if (!test_bit(QUEUE_FLAG_CLUSTER, &b->queue_flags)) {
- unsigned long flags;
-
- spin_lock_irqsave(t->queue_lock, flags);
- if (!test_bit(QUEUE_FLAG_CLUSTER, &b->queue_flags))
- queue_flag_clear(QUEUE_FLAG_CLUSTER, t);
- spin_unlock_irqrestore(t->queue_lock, flags);
- }
}
EXPORT_SYMBOL(disk_stack_limits);

diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 306759b..d246654 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -114,7 +114,7 @@ static ssize_t queue_max_segments_show(struct request_queue *q, char *page)

static ssize_t queue_max_segment_size_show(struct request_queue *q, char *page)
{
- if (test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags))
+ if (blk_queue_cluster(q))
return queue_var_show(queue_max_segment_size(q), (page));

return queue_var_show(PAGE_CACHE_SIZE, (page));
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 9924ea2..4a83321 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1081,11 +1081,6 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
*/
q->limits = *limits;

- if (limits->no_cluster)
- queue_flag_clear_unlocked(QUEUE_FLAG_CLUSTER, q);
- else
- queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, q);
-
dm_table_set_integrity(t);

/*
diff --git a/drivers/md/md.c b/drivers/md/md.c
index cf511de..e56e35e 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -4185,9 +4185,6 @@ static int md_alloc(dev_t dev, char *name)
goto abort;
mddev->queue->queuedata = mddev;

- /* Can be unlocked because the queue is new: no concurrency */
- queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, mddev->queue);
-
blk_queue_make_request(mddev->queue, md_make_request);

disk = alloc_disk(1 << shift);
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index 9cc450b..c2a9e12 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -1640,9 +1640,8 @@ struct request_queue *__scsi_alloc_queue(struct Scsi_Host *shost,

blk_queue_max_segment_size(q, dma_get_max_seg_size(dev));

- /* New queue, no concurrency on queue_flags */
if (!shost->use_clustering)
- queue_flag_clear_unlocked(QUEUE_FLAG_CLUSTER, q);
+ q->limits.cluster = 0;

/*
* set a reasonable default alignment on word boundaries: the
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 68eef76..cda62da 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -321,7 +321,7 @@ struct queue_limits {

unsigned char misaligned;
unsigned char discard_misaligned;
- unsigned char no_cluster;
+ unsigned char cluster;
signed char discard_zeroes_data;
};

@@ -444,7 +444,6 @@ struct request_queue
#endif
};

-#define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
#define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
#define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
#define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
@@ -465,7 +464,6 @@ struct request_queue
#define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */

#define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
- (1 << QUEUE_FLAG_CLUSTER) | \
(1 << QUEUE_FLAG_STACKABLE) | \
(1 << QUEUE_FLAG_SAME_COMP))

@@ -632,6 +630,11 @@ enum {

#define rq_data_dir(rq) ((rq)->cmd_flags & 1)

+static inline unsigned int blk_queue_cluster(struct request_queue *q)
+{
+ return q->limits.cluster;
+}
+
/*
* We regard a request as sync, if either a read or a sync write
*/
--
1.7.4.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/