[PATCH v3 0/3] block/dm: use BIOSET_PERCPU_CACHE from bio_alloc_bioset

[PATCH v3 0/3] block/dm: use BIOSET_PERCPU_CACHE from bio_alloc_bioset

[Mike Snitzer]

Hi Jens,

This v3 is a rebase of the previous v2 series ontop of the revised v2
patch that Christoph provided.

Linus hasn't pulled the for-5.18/dm-changes branch yet, so the 3rd DM
patch cannot be applied yet.  But feel free to pickup the first 2
block patches for 5.19 and I'll rebase dm-5.19 on block accordingly.

Thanks,
Mike

v3: tweaked some code comments, refined patch headers and folded DM
    patches so only one DM patch now.
v2: add REQ_ALLOC_CACHE and move use of bio_alloc_percpu_cache to
    bio_alloc_bioset

Mike Snitzer (3):
  block: allow using the per-cpu bio cache from bio_alloc_bioset
  block: allow use of per-cpu bio alloc cache by block drivers
  dm: conditionally enable BIOSET_PERCPU_CACHE for dm_io bioset

 block/bio.c               | 88 +++++++++++++++++++++++------------------------
 block/blk.h               |  7 ----
 block/fops.c              | 11 ++++--
 drivers/md/dm-table.c     | 11 ++++--
 drivers/md/dm.c           |  8 ++---
 drivers/md/dm.h           |  4 +--
 include/linux/bio.h       |  8 +++--
 include/linux/blk_types.h |  3 +-
 8 files changed, 73 insertions(+), 67 deletions(-)

-- 
2.15.0

[dm-devel] [PATCH v3 0/3] block/dm: use BIOSET_PERCPU_CACHE from bio_alloc_bioset

[Mike Snitzer]

Hi Jens,

This v3 is a rebase of the previous v2 series ontop of the revised v2
patch that Christoph provided.

Linus hasn't pulled the for-5.18/dm-changes branch yet, so the 3rd DM
patch cannot be applied yet.  But feel free to pickup the first 2
block patches for 5.19 and I'll rebase dm-5.19 on block accordingly.

Thanks,
Mike

v3: tweaked some code comments, refined patch headers and folded DM
    patches so only one DM patch now.
v2: add REQ_ALLOC_CACHE and move use of bio_alloc_percpu_cache to
    bio_alloc_bioset

Mike Snitzer (3):
  block: allow using the per-cpu bio cache from bio_alloc_bioset
  block: allow use of per-cpu bio alloc cache by block drivers
  dm: conditionally enable BIOSET_PERCPU_CACHE for dm_io bioset

 block/bio.c               | 88 +++++++++++++++++++++++------------------------
 block/blk.h               |  7 ----
 block/fops.c              | 11 ++++--
 drivers/md/dm-table.c     | 11 ++++--
 drivers/md/dm.c           |  8 ++---
 drivers/md/dm.h           |  4 +--
 include/linux/bio.h       |  8 +++--
 include/linux/blk_types.h |  3 +-
 8 files changed, 73 insertions(+), 67 deletions(-)

-- 
2.15.0

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel

[PATCH v3 1/3] block: allow using the per-cpu bio cache from bio_alloc_bioset

[Mike Snitzer]

Replace the BIO_PERCPU_CACHE bio-internal flag with a REQ_ALLOC_CACHE
one that can be passed to bio_alloc / bio_alloc_bioset, and implement
the percpu cache allocation logic in a helper called from
bio_alloc_bioset.  This allows any bio_alloc_bioset user to use the
percpu caches instead of having the functionality tied to struct kiocb.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
[hch: refactored a bit]
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 block/bio.c               | 86 ++++++++++++++++++++++-------------------------
 block/blk.h               |  3 +-
 block/fops.c              | 11 ++++--
 include/linux/bio.h       |  2 --
 include/linux/blk_types.h |  3 +-
 5 files changed, 52 insertions(+), 53 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index 33979f306e9e..09b714469b06 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -420,6 +420,28 @@ static void punt_bios_to_rescuer(struct bio_set *bs)
 	queue_work(bs->rescue_workqueue, &bs->rescue_work);
 }
 
+static struct bio *bio_alloc_percpu_cache(struct block_device *bdev,
+		unsigned short nr_vecs, unsigned int opf, gfp_t gfp,
+		struct bio_set *bs)
+{
+	struct bio_alloc_cache *cache;
+	struct bio *bio;
+
+	cache = per_cpu_ptr(bs->cache, get_cpu());
+	if (!cache->free_list) {
+		put_cpu();
+		return NULL;
+	}
+	bio = cache->free_list;
+	cache->free_list = bio->bi_next;
+	cache->nr--;
+	put_cpu();
+
+	bio_init(bio, bdev, nr_vecs ? bio->bi_inline_vecs : NULL, nr_vecs, opf);
+	bio->bi_pool = bs;
+	return bio;
+}
+
 /**
  * bio_alloc_bioset - allocate a bio for I/O
  * @bdev:	block device to allocate the bio for (can be %NULL)
@@ -452,6 +474,9 @@ static void punt_bios_to_rescuer(struct bio_set *bs)
  * submit_bio_noacct() should be avoided - instead, use bio_set's front_pad
  * for per bio allocations.
  *
+ * If REQ_ALLOC_CACHE is set, the final put of the bio MUST be done from process
+ * context, not hard/soft IRQ.
+ *
  * Returns: Pointer to new bio on success, NULL on failure.
  */
 struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
@@ -466,6 +491,21 @@ struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
 	if (WARN_ON_ONCE(!mempool_initialized(&bs->bvec_pool) && nr_vecs > 0))
 		return NULL;
 
+	if (opf & REQ_ALLOC_CACHE) {
+		if (bs->cache && nr_vecs <= BIO_INLINE_VECS) {
+			bio = bio_alloc_percpu_cache(bdev, nr_vecs, opf,
+						     gfp_mask, bs);
+			if (bio)
+				return bio;
+			/*
+			 * No cached bio available, bio returned below marked with
+			 * REQ_ALLOC_CACHE to particpate in per-cpu alloc cache.
+			 */
+		} else {
+			opf &= ~REQ_ALLOC_CACHE;
+		}
+	}
+
 	/*
 	 * submit_bio_noacct() converts recursion to iteration; this means if
 	 * we're running beneath it, any bios we allocate and submit will not be
@@ -712,7 +752,7 @@ void bio_put(struct bio *bio)
 			return;
 	}
 
-	if (bio_flagged(bio, BIO_PERCPU_CACHE)) {
+	if (bio->bi_opf & REQ_ALLOC_CACHE) {
 		struct bio_alloc_cache *cache;
 
 		bio_uninit(bio);
@@ -1734,50 +1774,6 @@ int bioset_init_from_src(struct bio_set *bs, struct bio_set *src)
 }
 EXPORT_SYMBOL(bioset_init_from_src);
 
-/**
- * bio_alloc_kiocb - Allocate a bio from bio_set based on kiocb
- * @kiocb:	kiocb describing the IO
- * @bdev:	block device to allocate the bio for (can be %NULL)
- * @nr_vecs:	number of iovecs to pre-allocate
- * @opf:	operation and flags for bio
- * @bs:		bio_set to allocate from
- *
- * Description:
- *    Like @bio_alloc_bioset, but pass in the kiocb. The kiocb is only
- *    used to check if we should dip into the per-cpu bio_set allocation
- *    cache. The allocation uses GFP_KERNEL internally. On return, the
- *    bio is marked BIO_PERCPU_CACHEABLE, and the final put of the bio
- *    MUST be done from process context, not hard/soft IRQ.
- *
- */
-struct bio *bio_alloc_kiocb(struct kiocb *kiocb, struct block_device *bdev,
-		unsigned short nr_vecs, unsigned int opf, struct bio_set *bs)
-{
-	struct bio_alloc_cache *cache;
-	struct bio *bio;
-
-	if (!(kiocb->ki_flags & IOCB_ALLOC_CACHE) || nr_vecs > BIO_INLINE_VECS)
-		return bio_alloc_bioset(bdev, nr_vecs, opf, GFP_KERNEL, bs);
-
-	cache = per_cpu_ptr(bs->cache, get_cpu());
-	if (cache->free_list) {
-		bio = cache->free_list;
-		cache->free_list = bio->bi_next;
-		cache->nr--;
-		put_cpu();
-		bio_init(bio, bdev, nr_vecs ? bio->bi_inline_vecs : NULL,
-			 nr_vecs, opf);
-		bio->bi_pool = bs;
-		bio_set_flag(bio, BIO_PERCPU_CACHE);
-		return bio;
-	}
-	put_cpu();
-	bio = bio_alloc_bioset(bdev, nr_vecs, opf, GFP_KERNEL, bs);
-	bio_set_flag(bio, BIO_PERCPU_CACHE);
-	return bio;
-}
-EXPORT_SYMBOL_GPL(bio_alloc_kiocb);
-
 static int __init init_bio(void)
 {
 	int i;
diff --git a/block/blk.h b/block/blk.h
index 6f21859c7f0f..9cb04f24ba8a 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -454,8 +454,7 @@ extern struct device_attribute dev_attr_events_poll_msecs;
 static inline void bio_clear_polled(struct bio *bio)
 {
 	/* can't support alloc cache if we turn off polling */
-	bio_clear_flag(bio, BIO_PERCPU_CACHE);
-	bio->bi_opf &= ~REQ_POLLED;
+	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
 }
 
 long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
diff --git a/block/fops.c b/block/fops.c
index 3696665e586a..f8227ef0719f 100644
--- a/block/fops.c
+++ b/block/fops.c
@@ -198,8 +198,10 @@ static ssize_t __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 	    (bdev_logical_block_size(bdev) - 1))
 		return -EINVAL;
 
-	bio = bio_alloc_kiocb(iocb, bdev, nr_pages, opf, &blkdev_dio_pool);
-
+	if (iocb->ki_flags & IOCB_ALLOC_CACHE)
+		opf |= REQ_ALLOC_CACHE;
+	bio = bio_alloc_bioset(bdev, nr_pages, opf, GFP_KERNEL,
+			       &blkdev_dio_pool);
 	dio = container_of(bio, struct blkdev_dio, bio);
 	atomic_set(&dio->ref, 1);
 	/*
@@ -320,7 +322,10 @@ static ssize_t __blkdev_direct_IO_async(struct kiocb *iocb,
 	    (bdev_logical_block_size(bdev) - 1))
 		return -EINVAL;
 
-	bio = bio_alloc_kiocb(iocb, bdev, nr_pages, opf, &blkdev_dio_pool);
+	if (iocb->ki_flags & IOCB_ALLOC_CACHE)
+		opf |= REQ_ALLOC_CACHE;
+	bio = bio_alloc_bioset(bdev, nr_pages, opf, GFP_KERNEL,
+			       &blkdev_dio_pool);
 	dio = container_of(bio, struct blkdev_dio, bio);
 	dio->flags = 0;
 	dio->iocb = iocb;
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 4c21f6e69e18..10406f57d339 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -408,8 +408,6 @@ extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
 struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
 			     unsigned int opf, gfp_t gfp_mask,
 			     struct bio_set *bs);
-struct bio *bio_alloc_kiocb(struct kiocb *kiocb, struct block_device *bdev,
-		unsigned short nr_vecs, unsigned int opf, struct bio_set *bs);
 struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned short nr_iovecs);
 extern void bio_put(struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 0c3563b45fe9..d4ba5251a3a0 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -328,7 +328,6 @@ enum {
 	BIO_QOS_MERGED,		/* but went through rq_qos merge path */
 	BIO_REMAPPED,
 	BIO_ZONE_WRITE_LOCKED,	/* Owns a zoned device zone write lock */
-	BIO_PERCPU_CACHE,	/* can participate in per-cpu alloc cache */
 	BIO_FLAG_LAST
 };
 
@@ -415,6 +414,7 @@ enum req_flag_bits {
 	__REQ_NOUNMAP,		/* do not free blocks when zeroing */
 
 	__REQ_POLLED,		/* caller polls for completion using bio_poll */
+	__REQ_ALLOC_CACHE,	/* allocate IO from cache if available */
 
 	/* for driver use */
 	__REQ_DRV,
@@ -440,6 +440,7 @@ enum req_flag_bits {
 
 #define REQ_NOUNMAP		(1ULL << __REQ_NOUNMAP)
 #define REQ_POLLED		(1ULL << __REQ_POLLED)
+#define REQ_ALLOC_CACHE		(1ULL << __REQ_ALLOC_CACHE)
 
 #define REQ_DRV			(1ULL << __REQ_DRV)
 #define REQ_SWAP		(1ULL << __REQ_SWAP)
-- 
2.15.0

[dm-devel] [PATCH v3 1/3] block: allow using the per-cpu bio cache from bio_alloc_bioset

[Mike Snitzer]

Replace the BIO_PERCPU_CACHE bio-internal flag with a REQ_ALLOC_CACHE
one that can be passed to bio_alloc / bio_alloc_bioset, and implement
the percpu cache allocation logic in a helper called from
bio_alloc_bioset.  This allows any bio_alloc_bioset user to use the
percpu caches instead of having the functionality tied to struct kiocb.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
[hch: refactored a bit]
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 block/bio.c               | 86 ++++++++++++++++++++++-------------------------
 block/blk.h               |  3 +-
 block/fops.c              | 11 ++++--
 include/linux/bio.h       |  2 --
 include/linux/blk_types.h |  3 +-
 5 files changed, 52 insertions(+), 53 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index 33979f306e9e..09b714469b06 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -420,6 +420,28 @@ static void punt_bios_to_rescuer(struct bio_set *bs)
 	queue_work(bs->rescue_workqueue, &bs->rescue_work);
 }
 
+static struct bio *bio_alloc_percpu_cache(struct block_device *bdev,
+		unsigned short nr_vecs, unsigned int opf, gfp_t gfp,
+		struct bio_set *bs)
+{
+	struct bio_alloc_cache *cache;
+	struct bio *bio;
+
+	cache = per_cpu_ptr(bs->cache, get_cpu());
+	if (!cache->free_list) {
+		put_cpu();
+		return NULL;
+	}
+	bio = cache->free_list;
+	cache->free_list = bio->bi_next;
+	cache->nr--;
+	put_cpu();
+
+	bio_init(bio, bdev, nr_vecs ? bio->bi_inline_vecs : NULL, nr_vecs, opf);
+	bio->bi_pool = bs;
+	return bio;
+}
+
 /**
  * bio_alloc_bioset - allocate a bio for I/O
  * @bdev:	block device to allocate the bio for (can be %NULL)
@@ -452,6 +474,9 @@ static void punt_bios_to_rescuer(struct bio_set *bs)
  * submit_bio_noacct() should be avoided - instead, use bio_set's front_pad
  * for per bio allocations.
  *
+ * If REQ_ALLOC_CACHE is set, the final put of the bio MUST be done from process
+ * context, not hard/soft IRQ.
+ *
  * Returns: Pointer to new bio on success, NULL on failure.
  */
 struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
@@ -466,6 +491,21 @@ struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
 	if (WARN_ON_ONCE(!mempool_initialized(&bs->bvec_pool) && nr_vecs > 0))
 		return NULL;
 
+	if (opf & REQ_ALLOC_CACHE) {
+		if (bs->cache && nr_vecs <= BIO_INLINE_VECS) {
+			bio = bio_alloc_percpu_cache(bdev, nr_vecs, opf,
+						     gfp_mask, bs);
+			if (bio)
+				return bio;
+			/*
+			 * No cached bio available, bio returned below marked with
+			 * REQ_ALLOC_CACHE to particpate in per-cpu alloc cache.
+			 */
+		} else {
+			opf &= ~REQ_ALLOC_CACHE;
+		}
+	}
+
 	/*
 	 * submit_bio_noacct() converts recursion to iteration; this means if
 	 * we're running beneath it, any bios we allocate and submit will not be
@@ -712,7 +752,7 @@ void bio_put(struct bio *bio)
 			return;
 	}
 
-	if (bio_flagged(bio, BIO_PERCPU_CACHE)) {
+	if (bio->bi_opf & REQ_ALLOC_CACHE) {
 		struct bio_alloc_cache *cache;
 
 		bio_uninit(bio);
@@ -1734,50 +1774,6 @@ int bioset_init_from_src(struct bio_set *bs, struct bio_set *src)
 }
 EXPORT_SYMBOL(bioset_init_from_src);
 
-/**
- * bio_alloc_kiocb - Allocate a bio from bio_set based on kiocb
- * @kiocb:	kiocb describing the IO
- * @bdev:	block device to allocate the bio for (can be %NULL)
- * @nr_vecs:	number of iovecs to pre-allocate
- * @opf:	operation and flags for bio
- * @bs:		bio_set to allocate from
- *
- * Description:
- *    Like @bio_alloc_bioset, but pass in the kiocb. The kiocb is only
- *    used to check if we should dip into the per-cpu bio_set allocation
- *    cache. The allocation uses GFP_KERNEL internally. On return, the
- *    bio is marked BIO_PERCPU_CACHEABLE, and the final put of the bio
- *    MUST be done from process context, not hard/soft IRQ.
- *
- */
-struct bio *bio_alloc_kiocb(struct kiocb *kiocb, struct block_device *bdev,
-		unsigned short nr_vecs, unsigned int opf, struct bio_set *bs)
-{
-	struct bio_alloc_cache *cache;
-	struct bio *bio;
-
-	if (!(kiocb->ki_flags & IOCB_ALLOC_CACHE) || nr_vecs > BIO_INLINE_VECS)
-		return bio_alloc_bioset(bdev, nr_vecs, opf, GFP_KERNEL, bs);
-
-	cache = per_cpu_ptr(bs->cache, get_cpu());
-	if (cache->free_list) {
-		bio = cache->free_list;
-		cache->free_list = bio->bi_next;
-		cache->nr--;
-		put_cpu();
-		bio_init(bio, bdev, nr_vecs ? bio->bi_inline_vecs : NULL,
-			 nr_vecs, opf);
-		bio->bi_pool = bs;
-		bio_set_flag(bio, BIO_PERCPU_CACHE);
-		return bio;
-	}
-	put_cpu();
-	bio = bio_alloc_bioset(bdev, nr_vecs, opf, GFP_KERNEL, bs);
-	bio_set_flag(bio, BIO_PERCPU_CACHE);
-	return bio;
-}
-EXPORT_SYMBOL_GPL(bio_alloc_kiocb);
-
 static int __init init_bio(void)
 {
 	int i;
diff --git a/block/blk.h b/block/blk.h
index 6f21859c7f0f..9cb04f24ba8a 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -454,8 +454,7 @@ extern struct device_attribute dev_attr_events_poll_msecs;
 static inline void bio_clear_polled(struct bio *bio)
 {
 	/* can't support alloc cache if we turn off polling */
-	bio_clear_flag(bio, BIO_PERCPU_CACHE);
-	bio->bi_opf &= ~REQ_POLLED;
+	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
 }
 
 long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
diff --git a/block/fops.c b/block/fops.c
index 3696665e586a..f8227ef0719f 100644
--- a/block/fops.c
+++ b/block/fops.c
@@ -198,8 +198,10 @@ static ssize_t __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 	    (bdev_logical_block_size(bdev) - 1))
 		return -EINVAL;
 
-	bio = bio_alloc_kiocb(iocb, bdev, nr_pages, opf, &blkdev_dio_pool);
-
+	if (iocb->ki_flags & IOCB_ALLOC_CACHE)
+		opf |= REQ_ALLOC_CACHE;
+	bio = bio_alloc_bioset(bdev, nr_pages, opf, GFP_KERNEL,
+			       &blkdev_dio_pool);
 	dio = container_of(bio, struct blkdev_dio, bio);
 	atomic_set(&dio->ref, 1);
 	/*
@@ -320,7 +322,10 @@ static ssize_t __blkdev_direct_IO_async(struct kiocb *iocb,
 	    (bdev_logical_block_size(bdev) - 1))
 		return -EINVAL;
 
-	bio = bio_alloc_kiocb(iocb, bdev, nr_pages, opf, &blkdev_dio_pool);
+	if (iocb->ki_flags & IOCB_ALLOC_CACHE)
+		opf |= REQ_ALLOC_CACHE;
+	bio = bio_alloc_bioset(bdev, nr_pages, opf, GFP_KERNEL,
+			       &blkdev_dio_pool);
 	dio = container_of(bio, struct blkdev_dio, bio);
 	dio->flags = 0;
 	dio->iocb = iocb;
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 4c21f6e69e18..10406f57d339 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -408,8 +408,6 @@ extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
 struct bio *bio_alloc_bioset(struct block_device *bdev, unsigned short nr_vecs,
 			     unsigned int opf, gfp_t gfp_mask,
 			     struct bio_set *bs);
-struct bio *bio_alloc_kiocb(struct kiocb *kiocb, struct block_device *bdev,
-		unsigned short nr_vecs, unsigned int opf, struct bio_set *bs);
 struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned short nr_iovecs);
 extern void bio_put(struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 0c3563b45fe9..d4ba5251a3a0 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -328,7 +328,6 @@ enum {
 	BIO_QOS_MERGED,		/* but went through rq_qos merge path */
 	BIO_REMAPPED,
 	BIO_ZONE_WRITE_LOCKED,	/* Owns a zoned device zone write lock */
-	BIO_PERCPU_CACHE,	/* can participate in per-cpu alloc cache */
 	BIO_FLAG_LAST
 };
 
@@ -415,6 +414,7 @@ enum req_flag_bits {
 	__REQ_NOUNMAP,		/* do not free blocks when zeroing */
 
 	__REQ_POLLED,		/* caller polls for completion using bio_poll */
+	__REQ_ALLOC_CACHE,	/* allocate IO from cache if available */
 
 	/* for driver use */
 	__REQ_DRV,
@@ -440,6 +440,7 @@ enum req_flag_bits {
 
 #define REQ_NOUNMAP		(1ULL << __REQ_NOUNMAP)
 #define REQ_POLLED		(1ULL << __REQ_POLLED)
+#define REQ_ALLOC_CACHE		(1ULL << __REQ_ALLOC_CACHE)
 
 #define REQ_DRV			(1ULL << __REQ_DRV)
 #define REQ_SWAP		(1ULL << __REQ_SWAP)
-- 
2.15.0

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel

[PATCH v3 2/3] block: allow use of per-cpu bio alloc cache by block drivers

[Mike Snitzer]

Refine per-cpu bio alloc cache interfaces so that DM and other block
drivers can properly create and use the cache:

DM uses bioset_init_from_src() to do its final bioset initialization,
so must update bioset_init_from_src() to set BIOSET_PERCPU_CACHE if
%src bioset has a cache.

Also move bio_clear_polled() to include/linux/bio.h to allow users
outside of block core.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
---
 block/bio.c         | 2 ++
 block/blk.h         | 6 ------
 include/linux/bio.h | 6 ++++++
 3 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index 09b714469b06..859f728e42dc 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1769,6 +1769,8 @@ int bioset_init_from_src(struct bio_set *bs, struct bio_set *src)
 		flags |= BIOSET_NEED_BVECS;
 	if (src->rescue_workqueue)
 		flags |= BIOSET_NEED_RESCUER;
+	if (src->cache)
+		flags |= BIOSET_PERCPU_CACHE;
 
 	return bioset_init(bs, src->bio_pool.min_nr, src->front_pad, flags);
 }
diff --git a/block/blk.h b/block/blk.h
index 9cb04f24ba8a..4f6b172c3342 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -451,12 +451,6 @@ extern struct device_attribute dev_attr_events;
 extern struct device_attribute dev_attr_events_async;
 extern struct device_attribute dev_attr_events_poll_msecs;
 
-static inline void bio_clear_polled(struct bio *bio)
-{
-	/* can't support alloc cache if we turn off polling */
-	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
-}
-
 long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 10406f57d339..a40a4ba2771f 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -783,6 +783,12 @@ static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
 		bio->bi_opf |= REQ_NOWAIT;
 }
 
+static inline void bio_clear_polled(struct bio *bio)
+{
+	/* can't support alloc cache if we turn off polling */
+	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
+}
+
 struct bio *blk_next_bio(struct bio *bio, struct block_device *bdev,
 		unsigned int nr_pages, unsigned int opf, gfp_t gfp);
 
-- 
2.15.0

[dm-devel] [PATCH v3 2/3] block: allow use of per-cpu bio alloc cache by block drivers

[Mike Snitzer]

Refine per-cpu bio alloc cache interfaces so that DM and other block
drivers can properly create and use the cache:

DM uses bioset_init_from_src() to do its final bioset initialization,
so must update bioset_init_from_src() to set BIOSET_PERCPU_CACHE if
%src bioset has a cache.

Also move bio_clear_polled() to include/linux/bio.h to allow users
outside of block core.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
---
 block/bio.c         | 2 ++
 block/blk.h         | 6 ------
 include/linux/bio.h | 6 ++++++
 3 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/block/bio.c b/block/bio.c
index 09b714469b06..859f728e42dc 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1769,6 +1769,8 @@ int bioset_init_from_src(struct bio_set *bs, struct bio_set *src)
 		flags |= BIOSET_NEED_BVECS;
 	if (src->rescue_workqueue)
 		flags |= BIOSET_NEED_RESCUER;
+	if (src->cache)
+		flags |= BIOSET_PERCPU_CACHE;
 
 	return bioset_init(bs, src->bio_pool.min_nr, src->front_pad, flags);
 }
diff --git a/block/blk.h b/block/blk.h
index 9cb04f24ba8a..4f6b172c3342 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -451,12 +451,6 @@ extern struct device_attribute dev_attr_events;
 extern struct device_attribute dev_attr_events_async;
 extern struct device_attribute dev_attr_events_poll_msecs;
 
-static inline void bio_clear_polled(struct bio *bio)
-{
-	/* can't support alloc cache if we turn off polling */
-	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
-}
-
 long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 10406f57d339..a40a4ba2771f 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -783,6 +783,12 @@ static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
 		bio->bi_opf |= REQ_NOWAIT;
 }
 
+static inline void bio_clear_polled(struct bio *bio)
+{
+	/* can't support alloc cache if we turn off polling */
+	bio->bi_opf &= ~(REQ_POLLED | REQ_ALLOC_CACHE);
+}
+
 struct bio *blk_next_bio(struct bio *bio, struct block_device *bdev,
 		unsigned int nr_pages, unsigned int opf, gfp_t gfp);
 
-- 
2.15.0

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel

Re: [PATCH v3 2/3] block: allow use of per-cpu bio alloc cache by block drivers

[Christoph Hellwig]

On Thu, Mar 24, 2022 at 04:35:25PM -0400, Mike Snitzer wrote:
> Refine per-cpu bio alloc cache interfaces so that DM and other block
> drivers can properly create and use the cache:
> 
> DM uses bioset_init_from_src() to do its final bioset initialization,
> so must update bioset_init_from_src() to set BIOSET_PERCPU_CACHE if
> %src bioset has a cache.
> 
> Also move bio_clear_polled() to include/linux/bio.h to allow users
> outside of block core.
> 
> Signed-off-by: Mike Snitzer <snitzer@kernel.org>

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [dm-devel] [PATCH v3 2/3] block: allow use of per-cpu bio alloc cache by block drivers

[Christoph Hellwig]

On Thu, Mar 24, 2022 at 04:35:25PM -0400, Mike Snitzer wrote:
> Refine per-cpu bio alloc cache interfaces so that DM and other block
> drivers can properly create and use the cache:
> 
> DM uses bioset_init_from_src() to do its final bioset initialization,
> so must update bioset_init_from_src() to set BIOSET_PERCPU_CACHE if
> %src bioset has a cache.
> 
> Also move bio_clear_polled() to include/linux/bio.h to allow users
> outside of block core.
> 
> Signed-off-by: Mike Snitzer <snitzer@kernel.org>

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel

[PATCH v3 3/3] dm: conditionally enable BIOSET_PERCPU_CACHE for dm_io bioset

[Mike Snitzer]

A bioset's per-cpu alloc cache may have broader utility in the future
but for now constrain it to being tightly coupled to QUEUE_FLAG_POLL.

Also change dm_io_complete() to use bio_clear_polled() so that it
properly clears all associated bio state on requeue.

This commit improves DM's hipri bio polling (REQ_POLLED) perf by
7 - 20% depending on the system.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
---
 drivers/md/dm-table.c | 11 ++++++++---
 drivers/md/dm.c       |  8 ++++----
 drivers/md/dm.h       |  4 ++--
 3 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index c0be4f60b427..7ebc70e3eb2f 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1002,6 +1002,8 @@ bool dm_table_request_based(struct dm_table *t)
 	return __table_type_request_based(dm_table_get_type(t));
 }
 
+static int dm_table_supports_poll(struct dm_table *t);
+
 static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *md)
 {
 	enum dm_queue_mode type = dm_table_get_type(t);
@@ -1009,21 +1011,24 @@ static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *
 	unsigned min_pool_size = 0;
 	struct dm_target *ti;
 	unsigned i;
+	bool poll_supported = false;
 
 	if (unlikely(type == DM_TYPE_NONE)) {
 		DMWARN("no table type is set, can't allocate mempools");
 		return -EINVAL;
 	}
 
-	if (__table_type_bio_based(type))
+	if (__table_type_bio_based(type)) {
 		for (i = 0; i < t->num_targets; i++) {
 			ti = t->targets + i;
 			per_io_data_size = max(per_io_data_size, ti->per_io_data_size);
 			min_pool_size = max(min_pool_size, ti->num_flush_bios);
 		}
+		poll_supported = !!dm_table_supports_poll(t);
+	}
 
-	t->mempools = dm_alloc_md_mempools(md, type, t->integrity_supported,
-					   per_io_data_size, min_pool_size);
+	t->mempools = dm_alloc_md_mempools(md, type, per_io_data_size, min_pool_size,
+					   t->integrity_supported, poll_supported);
 	if (!t->mempools)
 		return -ENOMEM;
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index b762a48d3fdf..b3e32116c31f 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -898,7 +898,7 @@ static void dm_io_complete(struct dm_io *io)
 		 * may only reflect a subset of the pre-split original,
 		 * so clear REQ_POLLED in case of requeue
 		 */
-		bio->bi_opf &= ~REQ_POLLED;
+		bio_clear_polled(bio);
 		return;
 	}
 
@@ -2915,8 +2915,8 @@ int dm_noflush_suspending(struct dm_target *ti)
 EXPORT_SYMBOL_GPL(dm_noflush_suspending);
 
 struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_queue_mode type,
-					    unsigned integrity, unsigned per_io_data_size,
-					    unsigned min_pool_size)
+					    unsigned per_io_data_size, unsigned min_pool_size,
+					    bool integrity, bool poll)
 {
 	struct dm_md_mempools *pools = kzalloc_node(sizeof(*pools), GFP_KERNEL, md->numa_node_id);
 	unsigned int pool_size = 0;
@@ -2932,7 +2932,7 @@ struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_qu
 		pool_size = max(dm_get_reserved_bio_based_ios(), min_pool_size);
 		front_pad = roundup(per_io_data_size, __alignof__(struct dm_target_io)) + DM_TARGET_IO_BIO_OFFSET;
 		io_front_pad = roundup(per_io_data_size,  __alignof__(struct dm_io)) + DM_IO_BIO_OFFSET;
-		ret = bioset_init(&pools->io_bs, pool_size, io_front_pad, 0);
+		ret = bioset_init(&pools->io_bs, pool_size, io_front_pad, poll ? BIOSET_PERCPU_CACHE : 0);
 		if (ret)
 			goto out;
 		if (integrity && bioset_integrity_create(&pools->io_bs, pool_size))
diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index 9013dc1a7b00..3f89664fea01 100644
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -221,8 +221,8 @@ void dm_kcopyd_exit(void);
  * Mempool operations
  */
 struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_queue_mode type,
-					    unsigned integrity, unsigned per_bio_data_size,
-					    unsigned min_pool_size);
+					    unsigned per_io_data_size, unsigned min_pool_size,
+					    bool integrity, bool poll);
 void dm_free_md_mempools(struct dm_md_mempools *pools);
 
 /*
-- 
2.15.0

[dm-devel] [PATCH v3 3/3] dm: conditionally enable BIOSET_PERCPU_CACHE for dm_io bioset

[Mike Snitzer]

A bioset's per-cpu alloc cache may have broader utility in the future
but for now constrain it to being tightly coupled to QUEUE_FLAG_POLL.

Also change dm_io_complete() to use bio_clear_polled() so that it
properly clears all associated bio state on requeue.

This commit improves DM's hipri bio polling (REQ_POLLED) perf by
7 - 20% depending on the system.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
---
 drivers/md/dm-table.c | 11 ++++++++---
 drivers/md/dm.c       |  8 ++++----
 drivers/md/dm.h       |  4 ++--
 3 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index c0be4f60b427..7ebc70e3eb2f 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1002,6 +1002,8 @@ bool dm_table_request_based(struct dm_table *t)
 	return __table_type_request_based(dm_table_get_type(t));
 }
 
+static int dm_table_supports_poll(struct dm_table *t);
+
 static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *md)
 {
 	enum dm_queue_mode type = dm_table_get_type(t);
@@ -1009,21 +1011,24 @@ static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *
 	unsigned min_pool_size = 0;
 	struct dm_target *ti;
 	unsigned i;
+	bool poll_supported = false;
 
 	if (unlikely(type == DM_TYPE_NONE)) {
 		DMWARN("no table type is set, can't allocate mempools");
 		return -EINVAL;
 	}
 
-	if (__table_type_bio_based(type))
+	if (__table_type_bio_based(type)) {
 		for (i = 0; i < t->num_targets; i++) {
 			ti = t->targets + i;
 			per_io_data_size = max(per_io_data_size, ti->per_io_data_size);
 			min_pool_size = max(min_pool_size, ti->num_flush_bios);
 		}
+		poll_supported = !!dm_table_supports_poll(t);
+	}
 
-	t->mempools = dm_alloc_md_mempools(md, type, t->integrity_supported,
-					   per_io_data_size, min_pool_size);
+	t->mempools = dm_alloc_md_mempools(md, type, per_io_data_size, min_pool_size,
+					   t->integrity_supported, poll_supported);
 	if (!t->mempools)
 		return -ENOMEM;
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index b762a48d3fdf..b3e32116c31f 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -898,7 +898,7 @@ static void dm_io_complete(struct dm_io *io)
 		 * may only reflect a subset of the pre-split original,
 		 * so clear REQ_POLLED in case of requeue
 		 */
-		bio->bi_opf &= ~REQ_POLLED;
+		bio_clear_polled(bio);
 		return;
 	}
 
@@ -2915,8 +2915,8 @@ int dm_noflush_suspending(struct dm_target *ti)
 EXPORT_SYMBOL_GPL(dm_noflush_suspending);
 
 struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_queue_mode type,
-					    unsigned integrity, unsigned per_io_data_size,
-					    unsigned min_pool_size)
+					    unsigned per_io_data_size, unsigned min_pool_size,
+					    bool integrity, bool poll)
 {
 	struct dm_md_mempools *pools = kzalloc_node(sizeof(*pools), GFP_KERNEL, md->numa_node_id);
 	unsigned int pool_size = 0;
@@ -2932,7 +2932,7 @@ struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_qu
 		pool_size = max(dm_get_reserved_bio_based_ios(), min_pool_size);
 		front_pad = roundup(per_io_data_size, __alignof__(struct dm_target_io)) + DM_TARGET_IO_BIO_OFFSET;
 		io_front_pad = roundup(per_io_data_size,  __alignof__(struct dm_io)) + DM_IO_BIO_OFFSET;
-		ret = bioset_init(&pools->io_bs, pool_size, io_front_pad, 0);
+		ret = bioset_init(&pools->io_bs, pool_size, io_front_pad, poll ? BIOSET_PERCPU_CACHE : 0);
 		if (ret)
 			goto out;
 		if (integrity && bioset_integrity_create(&pools->io_bs, pool_size))
diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index 9013dc1a7b00..3f89664fea01 100644
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -221,8 +221,8 @@ void dm_kcopyd_exit(void);
  * Mempool operations
  */
 struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_queue_mode type,
-					    unsigned integrity, unsigned per_bio_data_size,
-					    unsigned min_pool_size);
+					    unsigned per_io_data_size, unsigned min_pool_size,
+					    bool integrity, bool poll);
 void dm_free_md_mempools(struct dm_md_mempools *pools);
 
 /*
-- 
2.15.0

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel

Re: (subset) [PATCH v3 0/3] block/dm: use BIOSET_PERCPU_CACHE from bio_alloc_bioset

[Jens Axboe]

On Thu, 24 Mar 2022 16:35:23 -0400, Mike Snitzer wrote:
> This v3 is a rebase of the previous v2 series ontop of the revised v2
> patch that Christoph provided.
> 
> Linus hasn't pulled the for-5.18/dm-changes branch yet, so the 3rd DM
> patch cannot be applied yet.  But feel free to pickup the first 2
> block patches for 5.19 and I'll rebase dm-5.19 on block accordingly.
> 
> [...]

Applied, thanks!

[1/3] block: allow using the per-cpu bio cache from bio_alloc_bioset
      commit: a147e4805855e34f8e1027b88baf59a7f7c8b8d3
[2/3] block: allow use of per-cpu bio alloc cache by block drivers
      commit: e866e4dbad251b4dd1e134c295afd862333864bc

Best regards,
-- 
Jens Axboe

Re: [dm-devel] (subset) [PATCH v3 0/3] block/dm: use BIOSET_PERCPU_CACHE from bio_alloc_bioset

[Jens Axboe]

On Thu, 24 Mar 2022 16:35:23 -0400, Mike Snitzer wrote:
> This v3 is a rebase of the previous v2 series ontop of the revised v2
> patch that Christoph provided.
> 
> Linus hasn't pulled the for-5.18/dm-changes branch yet, so the 3rd DM
> patch cannot be applied yet.  But feel free to pickup the first 2
> block patches for 5.19 and I'll rebase dm-5.19 on block accordingly.
> 
> [...]

Applied, thanks!

[1/3] block: allow using the per-cpu bio cache from bio_alloc_bioset
      commit: a147e4805855e34f8e1027b88baf59a7f7c8b8d3
[2/3] block: allow use of per-cpu bio alloc cache by block drivers
      commit: e866e4dbad251b4dd1e134c295afd862333864bc

Best regards,
-- 
Jens Axboe


--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel