Re: [PATCH v7 4/9] block: Add core atomic write support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 05/06/2024 09:32, Christoph Hellwig wrote:
On Mon, Jun 03, 2024 at 02:29:26PM +0100, John Garry wrote:
I think that some of the logic could be re-used.
rq_straddles_atomic_write_boundary() is checked in merging of reqs/bios (to
see if the resultant req straddles a boundary).

So instead of saying: "will the resultant req straddle a boundary",
re-using path like blk_rq_get_max_sectors() -> blk_chunk_sectors_left(), we
check "is there space within the boundary limit to add this req/bio". We
need to take care of front and back merges, though.

Yes, we've used the trick to pass in the relevant limit in explicitly
to reuse infrastructure in other places, e.g. max_hw_sectors vs
max_zone_append_sectors for adding to a bio while respecting hardware
limits.


I assume that you are talking about something like queue_limits_max_zone_append_sectors().

Anyway, below is the prep patch I was considering for this re-use. It's just renaming any infrastructure for "chunk_sectors" to generic "boundary_sectors".

------>8-------

The purpose of the chunk_sectors limit is to ensure that a mergeable request fits within the boundary of the chunck_sector value.

Such a feature will be useful for other request_queue boundary limits, so generalize the chunk_sectors merge code.

This idea was proposed by Hannes Reinecke.

Signed-off-by: John Garry <john.g.garry@xxxxxxxxxx>

diff --git a/block/blk-merge.c b/block/blk-merge.c
index 8957e08e020c..6574c8b64ecc 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -168,11 +168,12 @@ static inline unsigned get_max_io_size(struct bio *bio,
 	unsigned pbs = lim->physical_block_size >> SECTOR_SHIFT;
 	unsigned lbs = lim->logical_block_size >> SECTOR_SHIFT;
 	unsigned max_sectors = lim->max_sectors, start, end;
+	unsigned int boundary_sectors = lim->chunk_sectors;

-	if (lim->chunk_sectors) {
+	if (boundary_sectors) {
 		max_sectors = min(max_sectors,
-			blk_chunk_sectors_left(bio->bi_iter.bi_sector,
-					       lim->chunk_sectors));
+			blk_boundary_sectors_left(bio->bi_iter.bi_sector,
+					      boundary_sectors));
 	}

 	start = bio->bi_iter.bi_sector & (pbs - 1);
@@ -588,19 +589,19 @@ static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
 						  sector_t offset)
 {
 	struct request_queue *q = rq->q;
-	unsigned int max_sectors;
+	unsigned int max_sectors, boundary_sectors = q->limits.chunk_sectors;

 	if (blk_rq_is_passthrough(rq))
 		return q->limits.max_hw_sectors;

 	max_sectors = blk_queue_get_max_sectors(rq);

-	if (!q->limits.chunk_sectors ||
+	if (!boundary_sectors ||
 	    req_op(rq) == REQ_OP_DISCARD ||
 	    req_op(rq) == REQ_OP_SECURE_ERASE)
 		return max_sectors;
 	return min(max_sectors,
-		   blk_chunk_sectors_left(offset, q->limits.chunk_sectors));
+		   blk_boundary_sectors_left(offset, boundary_sectors));
 }

 static inline int ll_new_hw_segment(struct request *req, struct bio *bio,
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 13037d6a6f62..b648253c2300 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1188,7 +1188,7 @@ static sector_t __max_io_len(struct dm_target *ti, sector_t sector,
 		return len;
 	return min_t(sector_t, len,
 		min(max_sectors ? : queue_max_sectors(ti->table->md->queue),
-		    blk_chunk_sectors_left(target_offset, max_granularity)));
+		    blk_boundary_sectors_left(target_offset, max_granularity)));
 }

 static inline sector_t max_io_len(struct dm_target *ti, sector_t sector)
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index ac8e0cb2353a..7657698b47f4 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -866,14 +866,14 @@ static inline bool bio_straddles_zones(struct bio *bio)
 }

 /*
- * Return how much of the chunk is left to be used for I/O at a given offset. + * Return how much within the boundary is left to be used for I/O at a given offset.
  */
-static inline unsigned int blk_chunk_sectors_left(sector_t offset,
-		unsigned int chunk_sectors)
+static inline unsigned int blk_boundary_sectors_left(sector_t offset,
+		unsigned int boundary_sectors)
 {
-	if (unlikely(!is_power_of_2(chunk_sectors)))
-		return chunk_sectors - sector_div(offset, chunk_sectors);
-	return chunk_sectors - (offset & (chunk_sectors - 1));
+	if (unlikely(!is_power_of_2(boundary_sectors)))
+		return boundary_sectors - sector_div(offset, boundary_sectors);
+	return boundary_sectors - (offset & (boundary_sectors - 1));
 }

 /**
--
2.31.1









[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [NTFS 3]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [NTFS 3]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux