If all subdevices support the same protection format the DM device is
flagged as capable.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/md/dm-table.c         |   35 ++++++++++++++++++++++++++++++++++-
 drivers/md/dm.c               |   26 ++++++++++++++++++++++++--
 drivers/md/dm.h               |    2 +-
 include/linux/device-mapper.h |    1 +
 4 files changed, 60 insertions(+), 4 deletions(-)

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 61f4414..006a10e 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -841,8 +841,12 @@ struct dm_target *dm_table_find_target(struct dm_table *t, sector_t sector)
 	return &t->targets[(KEYS_PER_NODE * n) + k];
 }
 
-void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q)
+void dm_table_set_restrictions(struct dm_table *t, struct mapped_device *md)
 {
+	struct request_queue *q = dm_queue(md);
+	struct list_head *devices = dm_table_get_devices(t);
+	struct dm_dev *prev, *cur;
+
 	/*
 	 * Make sure we obey the optimistic sub devices
 	 * restrictions.
@@ -861,6 +865,35 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q)
 	else
 		queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, q);
 
+	/*
+	 * Run through all devices to ensure they have matching
+	 * integrity profile
+	 */
+	cur = prev = NULL;
+
+	list_for_each_entry(cur, devices, list) {
+
+		if (prev && blk_integrity_compare(prev->bdev, cur->bdev) < 0) {
+			printk(KERN_ERR "%s: %s %s Integrity mismatch!\n",
+			       __func__, prev->bdev->bd_disk->disk_name,
+			       cur->bdev->bd_disk->disk_name);
+			return;
+		}
+		prev = cur;
+	}
+
+	/* Register dm device as being integrity capable */
+	if (prev && bdev_get_integrity(prev->bdev)) {
+		struct gendisk *disk = dm_disk(md);
+
+		if (blk_integrity_register(dm_disk(md), 
+					   bdev_get_integrity(prev->bdev)))
+			printk(KERN_ERR "%s: %s Could not register integrity!\n",
+			       __func__, ...
[ message continues ]

Please use checkpatch.  Always.  There's just no reason not to.  Sure,
you can make a decision to ignore some of its reports, but at least

For better or for worse, the developers of dm.c have chosen to keep the
whole file presentable in an 80-col display.  This patch breaks that,

This will provide no more information than the oops which will happen


This (unused, undocumented) function should be inlined.


--

[ message continues ]

Sorry about the delayed response.  I'm still on the road after LPC.

Andrew> For better or for worse, the developers of dm.c have chosen to
Andrew> keep the whole file presentable in an 80-col display.  This
Andrew> patch breaks that, and they might not like this.

I've broken the lines.


[Excessive BUG_ONs]

Andrew> This will provide no more information than the oops which will
Andrew> happen two lines later.


Andrew> This (unused, undocumented) function should be inlined.

 - It's not unused.

 - return md->queue is more descriptive than anything I could possibly
   put in a comment.

 - None of the other accessor functions in DM are inlined.  I'll let
   Alasdair decide.

Updated patches follow.

-- 
Martin K. Petersen      Oracle Linux Engineering
--

[ message continues ]

If all subdevices support the same protection format the DM device is
flagged as capable.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/md/dm-table.c         |   35 ++++++++++++++++++++++++++++++++++-
 drivers/md/dm.c               |   24 ++++++++++++++++++++++--
 drivers/md/dm.h               |    2 +-
 include/linux/device-mapper.h |    1 +
 4 files changed, 58 insertions(+), 4 deletions(-)

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 61f4414..7290a90 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -841,8 +841,12 @@ struct dm_target *dm_table_find_target(struct dm_table *t, sector_t sector)
 	return &t->targets[(KEYS_PER_NODE * n) + k];
 }
 
-void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q)
+void dm_table_set_restrictions(struct dm_table *t, struct mapped_device *md)
 {
+	struct request_queue *q = dm_queue(md);
+	struct list_head *devices = dm_table_get_devices(t);
+	struct dm_dev *prev, *cur;
+
 	/*
 	 * Make sure we obey the optimistic sub devices
 	 * restrictions.
@@ -861,6 +865,35 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q)
 	else
 		queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, q);
 
+	/*
+	 * Run through all devices to ensure they have matching
+	 * integrity profile
+	 */
+	cur = prev = NULL;
+
+	list_for_each_entry(cur, devices, list) {
+
+		if (prev && blk_integrity_compare(prev->bdev, cur->bdev) < 0) {
+			printk(KERN_ERR "%s: %s %s Integrity mismatch!\n",
+			       __func__, prev->bdev->bd_disk->disk_name,
+			       cur->bdev->bd_disk->disk_name);
+			return;
+		}
+		prev = cur;
+	}
+
+	/* Register dm device as being integrity capable */
+	if (prev && bdev_get_integrity(prev->bdev)) {
+		struct gendisk *disk = dm_disk(md);
+
+		if (blk_integrity_register(dm_disk(md),
+					   bdev_get_integrity(prev->bdev)))
+			printk(KERN_ERR "%s: %s Could not register integrity!\n",
+			       __func__, ...
[ message continues ]

Hi Martin,


Creating another function (e.g. dm_table_set_integrity(t, md)) may
be better, since then you wouldn't need dm_queue() nor need to change

If you need dm_queue() anyway, this part should be in drivers/md/dm.h
because other kernel modules like dm target drivers (e.g. dm-mirror)
don't use it.

Thanks,
Kiyoshi Ueda
--

[ message continues ]

Hi Martin,


Or you can also get md from t->md or q->queuedata, if you don't mind
to pass q as an argument of the function which sets integrity feature.
(I thought you mind that, since the integrity feature is set to
 md (md->disk), not q.)

Thanks,
Kiyoshi Ueda
--

[ message continues ]

>>>>> "Kiyoshi" == Kiyoshi Ueda <k-ueda@ct.jp.nec.com> writes:

Kiyoshi> Creating another function (e.g. dm_table_set_integrity(t,
Kiyoshi> md)) may be better, since then you wouldn't need dm_queue()
Kiyoshi> nor need to change the function interface of
Kiyoshi> dm_table_set_restrictions() I think.

You are right.  Originally I had more reasons for having that calling
interface change but that's no longer the case.  I'll fan the
integrity bits out to a separate function.

-- 
Martin K. Petersen	Oracle Linux Engineering

--

[ message continues ]

Helper function to find the sector offset in a bio given bvec index
and page offset.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 fs/bio.c            |   34 ++++++++++++++++++++++++++++++++++
 include/linux/bio.h |    1 +
 2 files changed, 35 insertions(+), 0 deletions(-)

diff --git a/fs/bio.c b/fs/bio.c
index e56e768..55ee2fd 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -1300,6 +1300,40 @@ struct bio_pair *bio_split(struct bio *bi, mempool_t *pool, int first_sectors)
 	return bp;
 }
 
+/**
+ *	bio_sector_offset - Find hardware sector offset in bio
+ *	@bio:		bio to inspect
+ *	@index:		bio_vec index
+ *	@offset:	offset in bv_page
+ *
+ *	Return the number of hardware sectors between beginning of bio
+ *	and an end point indicated by a bio_vec index and an offset
+ *	within that vector's page.
+ */
+sector_t bio_sector_offset(struct bio *bio, unsigned short index,
+			   unsigned int offset)
+{
+	struct bio_vec *bv;
+	unsigned int sector_sz = bio->bi_bdev->bd_disk->queue->hardsect_size;
+	sector_t sectors;
+	int i;
+
+	sectors = 0;
+
+	BUG_ON(index >= bio->bi_vcnt);
+
+	bio_for_each_segment(bv, bio, i) {
+		if (i == index) {
+			if (offset > bv->bv_offset)
+				sectors += (offset - bv->bv_offset) / sector_sz;
+			return sectors;
+		}
+
+		sectors += bv->bv_len / sector_sz;
+	}
+
+	BUG();
+}
 
 /*
  * create memory pools for biovec's in a bio_set.
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 6aba97d..386a1df 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -327,6 +327,7 @@ extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
 			   unsigned int, unsigned int);
 extern int bio_get_nr_vecs(struct block_device *);
+extern sector_t bio_sector_offset(struct bio *, unsigned short, unsigned int);
 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
 				unsigned long, ...
[ message continues ]