[PATCH 0/8 v3] No wait AIO

[PATCH 0/8 v3] No wait AIO

[Goldwyn Rodrigues]

Formerly known as non-blocking AIO.

This series adds nonblocking feature to asynchronous I/O writes.
io_submit() can be delayed because of a number of reason:
 - Block allocation for files
 - Data writebacks for direct I/O
 - Sleeping because of waiting to acquire i_rwsem
 - Congested block device

The goal of the patch series is to return -EAGAIN/-EWOULDBLOCK if
any of these conditions are met. This way userspace can push most
of the write()s to the kernel to the best of its ability to complete
and if it returns -EAGAIN, can defer it to another thread.

In order to enable this, IOCB_RW_FLAG_NOWAIT is introduced in
uapi/linux/aio_abi.h. If set for aio_rw_flags, it translates to
IOCB_NOWAIT for struct iocb, BIO_NOWAIT for bio and IOMAP_NOWAIT for
iomap. aio_rw_flags is a new flag replacing aio_reserved1. We could
not use aio_flags because it is not currently checked for invalidity
in the kernel.

This feature is provided for direct I/O of asynchronous I/O only. I have
tested it against xfs, ext4, and btrfs.

Changes since v1:
 + changed name from _NONBLOCKING to *_NOWAIT
 + filemap_range_has_page call moved to closer to (just before) calling filemap_write_and_wait_range().
 + BIO_NOWAIT limited to get_request()
 + XFS fixes 
	- included reflink 
	- use of xfs_ilock_nowait() instead of a XFS_IOLOCK_NONBLOCKING flag
	- Translate the flag through IOMAP_NOWAIT (iomap) to check for
	  block allocation for the file.
 + ext4 coding style

Changes since v2:
 + Using aio_reserved1 as aio_rw_flags instead of aio_flags
 + blk-mq support
 + xfs uptodate with kernel and reflink changes

-- 
Goldwyn

[PATCH 1/8] nowait aio: Introduce IOCB_RW_FLAG_NOWAIT

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

This flag informs kernel to bail out if an AIO request will block
for reasons such as file allocations, or a writeback triggered,
or would block while allocating requests while performing
direct I/O.

Unfortunately, aio_flags is not checked for validity. If we
add the flags to aio_flags, it would break existing applications
which have it set to anything besides zero or IOCB_FLAG_RESFD.
So, we are using aio_reserved1 and renaming it to aio_rw_flags.

IOCB_RW_FLAG_NOWAIT is translated to IOCB_NOWAIT for
iocb->ki_flags.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/aio.c                     | 10 +++++++++-
 include/linux/fs.h           |  1 +
 include/uapi/linux/aio_abi.h |  9 ++++++++-
 3 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/fs/aio.c b/fs/aio.c
index f52d925..41409ac 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1541,11 +1541,16 @@ static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb,
 	ssize_t ret;
 
 	/* enforce forwards compatibility on users */
-	if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) {
+	if (unlikely(iocb->aio_reserved2)) {
 		pr_debug("EINVAL: reserve field set\n");
 		return -EINVAL;
 	}
 
+	if (unlikely(iocb->aio_rw_flags & ~IOCB_RW_FLAG_NOWAIT)) {
+		pr_debug("EINVAL: aio_rw_flags set with incompatible flags\n");
+		return -EINVAL;
+	}
+
 	/* prevent overflows */
 	if (unlikely(
 	    (iocb->aio_buf != (unsigned long)iocb->aio_buf) ||
@@ -1586,6 +1591,9 @@ static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb,
 		req->common.ki_flags |= IOCB_EVENTFD;
 	}
 
+	if (iocb->aio_rw_flags & IOCB_RW_FLAG_NOWAIT)
+		req->common.ki_flags |= IOCB_NOWAIT;
+
 	ret = put_user(KIOCB_KEY, &user_iocb->aio_key);
 	if (unlikely(ret)) {
 		pr_debug("EFAULT: aio_key\n");
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 7251f7b..e8d9346 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -270,6 +270,7 @@ struct writeback_control;
 #define IOCB_DSYNC		(1 << 4)
 #define IOCB_SYNC		(1 << 5)
 #define IOCB_WRITE		(1 << 6)
+#define IOCB_NOWAIT		(1 << 7)
 
 struct kiocb {
 	struct file		*ki_filp;
diff --git a/include/uapi/linux/aio_abi.h b/include/uapi/linux/aio_abi.h
index bb2554f..6d98cbe 100644
--- a/include/uapi/linux/aio_abi.h
+++ b/include/uapi/linux/aio_abi.h
@@ -54,6 +54,13 @@ enum {
  */
 #define IOCB_FLAG_RESFD		(1 << 0)
 
+/*
+ * Flags for aio_rw_flags member of "struct iocb".
+ * IOCB_RW_FLAG_NOWAIT - Set if the user wants the iocb to fail if it
+ *			would block for operations such as disk allocation.
+ */
+#define IOCB_RW_FLAG_NOWAIT	(1 << 1)
+
 /* read() from /dev/aio returns these structures. */
 struct io_event {
 	__u64		data;		/* the data field from the iocb */
@@ -79,7 +86,7 @@ struct io_event {
 struct iocb {
 	/* these are internal to the kernel/libc. */
 	__u64	aio_data;	/* data to be returned in event's data */
-	__u32	PADDED(aio_key, aio_reserved1);
+	__u32	PADDED(aio_key, aio_rw_flags);
 				/* the kernel sets aio_key to the req # */
 
 	/* common fields */
-- 
2.10.2

[PATCH 2/8] nowait aio: Return if cannot get hold of i_rwsem

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

A failure to lock i_rwsem would mean there is I/O being performed
by another thread. So, let's bail.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 mm/filemap.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index 1694623..e08f3b9 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2982,7 +2982,12 @@ ssize_t generic_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	struct inode *inode = file->f_mapping->host;
 	ssize_t ret;
 
-	inode_lock(inode);
+	if (!inode_trylock(inode)) {
+		/* Don't sleep on inode rwsem */
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			return -EAGAIN;
+		inode_lock(inode);
+	}
 	ret = generic_write_checks(iocb, from);
 	if (ret > 0)
 		ret = __generic_file_write_iter(iocb, from);
-- 
2.10.2

[PATCH 3/8] nowait aio: return if direct write will trigger writeback

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

Find out if the write will trigger a wait due to writeback. If yes,
return -EAGAIN.

This introduces a new function filemap_range_has_page() which
returns true if the file's mapping has a page within the range
mentioned.

Return -EINVAL for buffered AIO: there are multiple causes of
delay such as page locks, dirty throttling logic, page loading
from disk etc. which cannot be taken care of.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 include/linux/fs.h |  2 ++
 mm/filemap.c       | 50 +++++++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 49 insertions(+), 3 deletions(-)

diff --git a/include/linux/fs.h b/include/linux/fs.h
index e8d9346..4a30e8f 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2514,6 +2514,8 @@ extern int filemap_fdatawait(struct address_space *);
 extern void filemap_fdatawait_keep_errors(struct address_space *);
 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
 				   loff_t lend);
+extern int filemap_range_has_page(struct address_space *, loff_t lstart,
+				   loff_t lend);
 extern int filemap_write_and_wait(struct address_space *mapping);
 extern int filemap_write_and_wait_range(struct address_space *mapping,
 				        loff_t lstart, loff_t lend);
diff --git a/mm/filemap.c b/mm/filemap.c
index e08f3b9..c020e23 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -376,6 +376,39 @@ int filemap_flush(struct address_space *mapping)
 }
 EXPORT_SYMBOL(filemap_flush);
 
+/**
+ * filemap_range_has_page - check if a page exists in range.
+ * @mapping:		address space structure to wait for
+ * @start_byte:		offset in bytes where the range starts
+ * @end_byte:		offset in bytes where the range ends (inclusive)
+ *
+ * Find at least one page in the range supplied, usually used to check if
+ * direct writing in this range will trigger a writeback.
+ */
+int filemap_range_has_page(struct address_space *mapping,
+				loff_t start_byte, loff_t end_byte)
+{
+	pgoff_t index = start_byte >> PAGE_SHIFT;
+	pgoff_t end = end_byte >> PAGE_SHIFT;
+	struct pagevec pvec;
+	int ret;
+
+	if (end_byte < start_byte)
+		return 0;
+
+	if (mapping->nrpages == 0)
+		return 0;
+
+	pagevec_init(&pvec, 0);
+	ret = pagevec_lookup(&pvec, mapping, index, 1);
+	if (!ret)
+		return 0;
+	ret = (pvec.pages[0]->index <= end);
+	pagevec_release(&pvec);
+	return ret;
+}
+EXPORT_SYMBOL(filemap_range_has_page);
+
 static int __filemap_fdatawait_range(struct address_space *mapping,
 				     loff_t start_byte, loff_t end_byte)
 {
@@ -2640,6 +2673,9 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
 
 	pos = iocb->ki_pos;
 
+	if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT))
+		return -EINVAL;
+
 	if (limit != RLIM_INFINITY) {
 		if (iocb->ki_pos >= limit) {
 			send_sig(SIGXFSZ, current, 0);
@@ -2709,9 +2745,17 @@ generic_file_direct_write(struct kiocb *iocb, struct iov_iter *from)
 	write_len = iov_iter_count(from);
 	end = (pos + write_len - 1) >> PAGE_SHIFT;
 
-	written = filemap_write_and_wait_range(mapping, pos, pos + write_len - 1);
-	if (written)
-		goto out;
+	if (iocb->ki_flags & IOCB_NOWAIT) {
+		/* If there are pages to writeback, return */
+		if (filemap_range_has_page(inode->i_mapping, pos,
+					   pos + iov_iter_count(from)))
+			return -EAGAIN;
+	} else {
+		written = filemap_write_and_wait_range(mapping, pos,
+							pos + write_len - 1);
+		if (written)
+			goto out;
+	}
 
 	/*
 	 * After a write we want buffered reads to be sure to go to disk to get
-- 
2.10.2

[PATCH 4/8] nowait-aio: Introduce IOMAP_NOWAIT

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

IOCB_NOWAIT translates to IOMAP_NOWAIT for iomaps.
This is used by XFS in the XFS patch.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/iomap.c            | 2 ++
 include/linux/iomap.h | 1 +
 2 files changed, 3 insertions(+)

diff --git a/fs/iomap.c b/fs/iomap.c
index 141c3cd..d1c8175 100644
--- a/fs/iomap.c
+++ b/fs/iomap.c
@@ -885,6 +885,8 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 	} else {
 		dio->flags |= IOMAP_DIO_WRITE;
 		flags |= IOMAP_WRITE;
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			flags |= IOMAP_NOWAIT;
 	}
 
 	if (mapping->nrpages) {
diff --git a/include/linux/iomap.h b/include/linux/iomap.h
index 7291810..53f6af8 100644
--- a/include/linux/iomap.h
+++ b/include/linux/iomap.h
@@ -51,6 +51,7 @@ struct iomap {
 #define IOMAP_REPORT		(1 << 2) /* report extent status, e.g. FIEMAP */
 #define IOMAP_FAULT		(1 << 3) /* mapping for page fault */
 #define IOMAP_DIRECT		(1 << 4) /* direct I/O */
+#define IOMAP_NOWAIT		(1 << 5) /* Don't wait for writeback */
 
 struct iomap_ops {
 	/*
-- 
2.10.2

[PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

A new flag BIO_NOWAIT is introduced to identify bio's
orignating from iocb with IOCB_NOWAIT. This flag indicates
to return immediately if a request cannot be made instead
of retrying.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 block/blk-core.c          | 12 ++++++++++--
 block/blk-mq-sched.c      |  3 +++
 block/blk-mq.c            |  4 ++++
 fs/direct-io.c            | 11 +++++++++--
 include/linux/bio.h       |  6 ++++++
 include/linux/blk_types.h |  1 +
 6 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 0eeb99e..2e5cba2 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1232,6 +1232,11 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
 	if (!IS_ERR(rq))
 		return rq;
 
+	if (bio && bio_flagged(bio, BIO_NOWAIT)) {
+		blk_put_rl(rl);
+		return ERR_PTR(-EAGAIN);
+	}
+
 	if (!gfpflags_allow_blocking(gfp_mask) || unlikely(blk_queue_dying(q))) {
 		blk_put_rl(rl);
 		return rq;
@@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 	do {
 		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 
-		if (likely(blk_queue_enter(q, false) == 0)) {
+		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
 			struct bio_list hold;
 			struct bio_list lower, same;
 
@@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
 			bio_list_merge(&bio_list_on_stack, &same);
 			bio_list_merge(&bio_list_on_stack, &hold);
 		} else {
-			bio_io_error(bio);
+			if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
+				bio_wouldblock_error(bio);
+			else
+				bio_io_error(bio);
 		}
 		bio = bio_list_pop(current->bio_list);
 	} while (bio);
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 09af8ff..40e78b5 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
 	if (likely(!data->hctx))
 		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
 
+	if (likely(bio) && bio_flagged(bio, BIO_NOWAIT))
+		data->flags |= BLK_MQ_REQ_NOWAIT;
+
 	if (e) {
 		data->flags |= BLK_MQ_REQ_INTERNAL;
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 159187a..942ce8c 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1518,6 +1518,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && bio_flagged(bio, BIO_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
@@ -1642,6 +1644,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && bio_flagged(bio, BIO_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
diff --git a/fs/direct-io.c b/fs/direct-io.c
index a04ebea..f6835d3 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -386,6 +386,9 @@ dio_bio_alloc(struct dio *dio, struct dio_submit *sdio,
 	else
 		bio->bi_end_io = dio_bio_end_io;
 
+	if (dio->iocb->ki_flags & IOCB_NOWAIT)
+		bio_set_flag(bio, BIO_NOWAIT);
+
 	sdio->bio = bio;
 	sdio->logical_offset_in_bio = sdio->cur_page_fs_offset;
 }
@@ -480,8 +483,12 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
 	unsigned i;
 	int err;
 
-	if (bio->bi_error)
-		dio->io_error = -EIO;
+	if (bio->bi_error) {
+		if (bio_flagged(bio, BIO_NOWAIT))
+			dio->io_error = -EAGAIN;
+		else
+			dio->io_error = -EIO;
+	}
 
 	if (dio->is_async && dio->op == REQ_OP_READ && dio->should_dirty) {
 		err = bio->bi_error;
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 8e52119..1a92707 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -425,6 +425,12 @@ static inline void bio_io_error(struct bio *bio)
 	bio_endio(bio);
 }
 
+static inline void bio_wouldblock_error(struct bio *bio)
+{
+	bio->bi_error = -EAGAIN;
+	bio_endio(bio);
+}
+
 struct request_queue;
 extern int bio_phys_segments(struct request_queue *, struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index d703acb..514c08e 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -102,6 +102,7 @@ struct bio {
 #define BIO_REFFED	8	/* bio has elevated ->bi_cnt */
 #define BIO_THROTTLED	9	/* This bio has already been subjected to
 				 * throttling rules. Don't do it again. */
+#define BIO_NOWAIT	10	/* don't block over blk device congestion */
 
 /*
  * Flags starting here get preserved by bio_reset() - this includes
-- 
2.10.2

[PATCH 6/8] nowait aio: ext4

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

Return EAGAIN if any of the following checks fail for direct I/O:
 + i_rwsem is lockable
 + Writing beyond end of file (will trigger allocation)
 + Blocks are not allocated at the write location

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/ext4/file.c | 48 +++++++++++++++++++++++++++++++-----------------
 1 file changed, 31 insertions(+), 17 deletions(-)

diff --git a/fs/ext4/file.c b/fs/ext4/file.c
index 8210c1f..e223b9f 100644
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -127,27 +127,22 @@ ext4_unaligned_aio(struct inode *inode, struct iov_iter *from, loff_t pos)
 	return 0;
 }
 
-/* Is IO overwriting allocated and initialized blocks? */
-static bool ext4_overwrite_io(struct inode *inode, loff_t pos, loff_t len)
+/* Are IO blocks allocated */
+static bool ext4_blocks_mapped(struct inode *inode, loff_t pos, loff_t len,
+				struct ext4_map_blocks *map)
 {
-	struct ext4_map_blocks map;
 	unsigned int blkbits = inode->i_blkbits;
 	int err, blklen;
 
 	if (pos + len > i_size_read(inode))
 		return false;
 
-	map.m_lblk = pos >> blkbits;
-	map.m_len = EXT4_MAX_BLOCKS(len, pos, blkbits);
-	blklen = map.m_len;
+	map->m_lblk = pos >> blkbits;
+	map->m_len = EXT4_MAX_BLOCKS(len, pos, blkbits);
+	blklen = map->m_len;
 
-	err = ext4_map_blocks(NULL, inode, &map, 0);
-	/*
-	 * 'err==len' means that all of the blocks have been preallocated,
-	 * regardless of whether they have been initialized or not. To exclude
-	 * unwritten extents, we need to check m_flags.
-	 */
-	return err == blklen && (map.m_flags & EXT4_MAP_MAPPED);
+	err = ext4_map_blocks(NULL, inode, map, 0);
+	return err == blklen;
 }
 
 static ssize_t ext4_write_checks(struct kiocb *iocb, struct iov_iter *from)
@@ -204,6 +199,7 @@ ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 {
 	struct inode *inode = file_inode(iocb->ki_filp);
 	int o_direct = iocb->ki_flags & IOCB_DIRECT;
+	int nowait = iocb->ki_flags & IOCB_NOWAIT;
 	int unaligned_aio = 0;
 	int overwrite = 0;
 	ssize_t ret;
@@ -216,7 +212,13 @@ ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 		return ext4_dax_write_iter(iocb, from);
 #endif
 
-	inode_lock(inode);
+	if (o_direct && nowait) {
+		if (!inode_trylock(inode))
+			return -EAGAIN;
+	} else {
+		inode_lock(inode);
+	}
+
 	ret = ext4_write_checks(iocb, from);
 	if (ret <= 0)
 		goto out;
@@ -235,9 +237,21 @@ ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 
 	iocb->private = &overwrite;
 	/* Check whether we do a DIO overwrite or not */
-	if (o_direct && ext4_should_dioread_nolock(inode) && !unaligned_aio &&
-	    ext4_overwrite_io(inode, iocb->ki_pos, iov_iter_count(from)))
-		overwrite = 1;
+	if (o_direct && !unaligned_aio) {
+		struct ext4_map_blocks map;
+		if (ext4_blocks_mapped(inode, iocb->ki_pos,
+				      iov_iter_count(from), &map)) {
+	 		/* To exclude unwritten extents, we need to check
+			 * m_flags.
+			 */
+			if (ext4_should_dioread_nolock(inode) &&
+			    (map.m_flags & EXT4_MAP_MAPPED))
+				overwrite = 1;
+		} else if (iocb->ki_flags & IOCB_NOWAIT) {
+			ret = -EAGAIN;
+			goto out;
+		}
+	}
 
 	ret = __generic_file_write_iter(iocb, from);
 	inode_unlock(inode);
-- 
2.10.2

[PATCH 7/8] nowait aio: xfs

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

If IOCB_NOWAIT is set, bail if the i_rwsem is not lockable
immediately.

IF IOMAP_NOWAIT is set, return EAGAIN in xfs_file_iomap_begin
if it needs allocation either due to file extension, writing to a hole,
or COW or waiting for other DIOs to finish.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/xfs/xfs_file.c  | 15 +++++++++++----
 fs/xfs/xfs_iomap.c | 13 +++++++++++++
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 35703a8..08a5eef 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -541,8 +541,11 @@ xfs_file_dio_aio_write(
 		iolock = XFS_IOLOCK_SHARED;
 	}
 
-	xfs_ilock(ip, iolock);
-
+	if (!xfs_ilock_nowait(ip, iolock)) {
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			return -EAGAIN;
+		xfs_ilock(ip, iolock);
+	}
 	ret = xfs_file_aio_write_checks(iocb, from, &iolock);
 	if (ret)
 		goto out;
@@ -553,9 +556,13 @@ xfs_file_dio_aio_write(
 	 * otherwise demote the lock if we had to take the exclusive lock
 	 * for other reasons in xfs_file_aio_write_checks.
 	 */
-	if (unaligned_io)
+	if (unaligned_io) {
+		/* If we are going to wait for other DIO to finish, bail */
+		if ((iocb->ki_flags & IOCB_NOWAIT) &&
+		     atomic_read(&inode->i_dio_count))
+			return -EAGAIN;
 		inode_dio_wait(inode);
-	else if (iolock == XFS_IOLOCK_EXCL) {
+	} else if (iolock == XFS_IOLOCK_EXCL) {
 		xfs_ilock_demote(ip, XFS_IOLOCK_EXCL);
 		iolock = XFS_IOLOCK_SHARED;
 	}
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 288ee5b..6843725 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -1015,6 +1015,11 @@ xfs_file_iomap_begin(
 
 	if ((flags & (IOMAP_WRITE | IOMAP_ZERO)) && xfs_is_reflink_inode(ip)) {
 		if (flags & IOMAP_DIRECT) {
+			/* A reflinked inode will result in CoW alloc */
+			if (flags & IOMAP_NOWAIT) {
+				error = -EAGAIN;
+				goto out_unlock;
+			}
 			/* may drop and re-acquire the ilock */
 			error = xfs_reflink_allocate_cow(ip, &imap, &shared,
 					&lockmode);
@@ -1032,6 +1037,14 @@ xfs_file_iomap_begin(
 
 	if ((flags & IOMAP_WRITE) && imap_needs_alloc(inode, &imap, nimaps)) {
 		/*
+		 * If nowait is set bail since we are going to make
+		 * allocations.
+		 */
+		if (flags & IOMAP_NOWAIT) {
+			error = -EAGAIN;
+			goto out_unlock;
+		}
+		/*
 		 * We cap the maximum length we map here to MAX_WRITEBACK_PAGES
 		 * pages to keep the chunks of work done where somewhat symmetric
 		 * with the work writeback does. This is a completely arbitrary
-- 
2.10.2

[PATCH 8/8] nowait aio: btrfs

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

Return EAGAIN if any of the following checks fail
 + i_rwsem is not lockable
 + NODATACOW or PREALLOC is not set
 + Cannot nocow at the desired location
 + Writing beyond end of file which is not allocated

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/btrfs/file.c  | 25 ++++++++++++++++++++-----
 fs/btrfs/inode.c |  3 +++
 2 files changed, 23 insertions(+), 5 deletions(-)

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 520cb72..a870e5d 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1823,12 +1823,29 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	ssize_t num_written = 0;
 	bool sync = (file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host);
 	ssize_t err;
-	loff_t pos;
-	size_t count;
+	loff_t pos = iocb->ki_pos;
+	size_t count = iov_iter_count(from);
 	loff_t oldsize;
 	int clean_page = 0;
 
-	inode_lock(inode);
+	if ((iocb->ki_flags & IOCB_NOWAIT) &&
+			(iocb->ki_flags & IOCB_DIRECT)) {
+		/* Don't sleep on inode rwsem */
+		if (!inode_trylock(inode))
+			return -EAGAIN;
+		/*
+		 * We will allocate space in case nodatacow is not set,
+		 * so bail
+		 */
+		if (!(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
+					      BTRFS_INODE_PREALLOC)) ||
+		    check_can_nocow(BTRFS_I(inode), pos, &count) <= 0) {
+			inode_unlock(inode);
+			return -EAGAIN;
+		}
+	} else
+		inode_lock(inode);
+
 	err = generic_write_checks(iocb, from);
 	if (err <= 0) {
 		inode_unlock(inode);
@@ -1862,8 +1879,6 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	 */
 	update_time_for_write(inode);
 
-	pos = iocb->ki_pos;
-	count = iov_iter_count(from);
 	start_pos = round_down(pos, fs_info->sectorsize);
 	oldsize = i_size_read(inode);
 	if (start_pos > oldsize) {
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index c40060c..788bb93 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -8613,6 +8613,9 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 			dio_data.overwrite = 1;
 			inode_unlock(inode);
 			relock = true;
+		} else if (iocb->ki_flags & IOCB_NOWAIT) {
+			ret = -EAGAIN;
+			goto out;
 		}
 		ret = btrfs_delalloc_reserve_space(inode, offset, count);
 		if (ret)
-- 
2.10.2

Re: [PATCH 3/8] nowait aio: return if direct write will trigger writeback

[Matthew Wilcox]

On Wed, Mar 15, 2017 at 04:51:02PM -0500, Goldwyn Rodrigues wrote:
> This introduces a new function filemap_range_has_page() which
> returns true if the file's mapping has a page within the range
> mentioned.

I thought you were going to replace this patch with one that starts
writeback for these pages but does not wait for them?

Re: [PATCH 3/8] nowait aio: return if direct write will trigger writeback

[Matthew Wilcox]

On Wed, Mar 15, 2017 at 04:51:02PM -0500, Goldwyn Rodrigues wrote:
> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
> 
> Find out if the write will trigger a wait due to writeback. If yes,
> return -EAGAIN.
> 
> This introduces a new function filemap_range_has_page() which
> returns true if the file's mapping has a page within the range
> mentioned.
> 
> Return -EINVAL for buffered AIO: there are multiple causes of
> delay such as page locks, dirty throttling logic, page loading
> from disk etc. which cannot be taken care of.

Also, this patch only touches the write path; we have a similar call to
write_and_wait_range() in generic_file_read_iter().

Actually, why do we even have that?  Why can't we satisfy an O_DIRECT
read from the cache?

Re: [PATCH 3/8] nowait aio: return if direct write will trigger writeback

[Goldwyn Rodrigues]

On 03/16/2017 08:08 AM, Matthew Wilcox wrote:
> On Wed, Mar 15, 2017 at 04:51:02PM -0500, Goldwyn Rodrigues wrote:
>> This introduces a new function filemap_range_has_page() which
>> returns true if the file's mapping has a page within the range
>> mentioned.
> 
> I thought you were going to replace this patch with one that starts
> writeback for these pages but does not wait for them?
> 

As mentioned by Jan, Flags to filemap_write_and_wait_range are
unnecessarily complicated. The AIO-DIO API users who eye for performance
usually are careful with page writes/evictions. As a fallback, they can
(and should) go the wait route (without  IOCB_RW_FLAG_NOWAIT).

Finally, my take on this is that we don't want to perform tasks for a
following system call, which may or may not immediately follow the
current one. May not, because an application (DB) will offload the task
from the CPU thread to the I/O thread in case of -EAGAIN. A system call
should be complete in itself (and do the minimum, what is asked).


-- 
Goldwyn

Re: [PATCH 3/8] nowait aio: return if direct write will trigger writeback

[Goldwyn Rodrigues]

On 03/16/2017 08:20 AM, Matthew Wilcox wrote:
> On Wed, Mar 15, 2017 at 04:51:02PM -0500, Goldwyn Rodrigues wrote:
>> From: Goldwyn Rodrigues <rgoldwyn-IBi9RG/b67k@public.gmane.org>
>>
>> Find out if the write will trigger a wait due to writeback. If yes,
>> return -EAGAIN.
>>
>> This introduces a new function filemap_range_has_page() which
>> returns true if the file's mapping has a page within the range
>> mentioned.
>>
>> Return -EINVAL for buffered AIO: there are multiple causes of
>> delay such as page locks, dirty throttling logic, page loading
>> from disk etc. which cannot be taken care of.
> 
> Also, this patch only touches the write path; we have a similar call to
> write_and_wait_range() in generic_file_read_iter().
> 

This patch series is concerned with direct-write AIO paths only.

-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Jens Axboe]

On 03/15/2017 03:51 PM, Goldwyn Rodrigues wrote:
> diff --git a/block/blk-core.c b/block/blk-core.c
> index 0eeb99e..2e5cba2 100644
> --- a/block/blk-core.c
> +++ b/block/blk-core.c
> @@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
>  	do {
>  		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
>  
> -		if (likely(blk_queue_enter(q, false) == 0)) {
> +		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
>  			struct bio_list hold;
>  			struct bio_list lower, same;
>  
> @@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
>  			bio_list_merge(&bio_list_on_stack, &same);
>  			bio_list_merge(&bio_list_on_stack, &hold);
>  		} else {
> -			bio_io_error(bio);
> +			if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
> +				bio_wouldblock_error(bio);
> +			else
> +				bio_io_error(bio);

This doesn't look right. What if the queue is dying, and BIO_NOWAIT just
happened to be set?

And you're missing wbt_wait() as well as a blocking point. Ditto in
blk-mq.

> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index 159187a..942ce8c 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -1518,6 +1518,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>  	if (unlikely(!rq)) {
>  		__wbt_done(q->rq_wb, wb_acct);
> +		if (bio && bio_flagged(bio, BIO_NOWAIT))
> +			bio_wouldblock_error(bio);
>  		return BLK_QC_T_NONE;
>  	}
>  

This seems a little fragile now, since not both paths free the bio.

-- 
Jens Axboe

Re: [PATCH 5/8] nowait aio: return on congested block device

[Dave Chinner]

On Wed, Mar 15, 2017 at 04:51:04PM -0500, Goldwyn Rodrigues wrote:
> From: Goldwyn Rodrigues <rgoldwyn-IBi9RG/b67k@public.gmane.org>
> 
> A new flag BIO_NOWAIT is introduced to identify bio's
> orignating from iocb with IOCB_NOWAIT. This flag indicates
> to return immediately if a request cannot be made instead
> of retrying.

So this makes a congested block device run the bio IO completion
callback with an -EAGAIN error present? Are all the filesystem
direct IO submission and completion routines OK with that? i.e. does
such a congestion case cause filesystems to temporarily expose stale
data to unprivileged users when the IO is requeued in this way?

e.g. filesystem does allocation without blocking, submits bio,
device is congested, runs IO completion with error, so nothing
written to allocated blocks, write gets queued, so other read
comes in while the write is queued, reads data from uninitialised
blocks that were allocated during the write....

Seems kinda problematic to me to have a undocumented design
constraint (i.e a landmine) where we submit the AIO only to have it
error out and then expect the filesystem to do something special and
different /without blocking/ on EAGAIN.

Why isn't the congestion check at a higher layer like we do for page
cache readahead? i.e. using the bdi*congested() API at the time we
are doing all the other filesystem blocking checks.

-Dave.
-- 
Dave Chinner
david-FqsqvQoI3Ljby3iVrkZq2A@public.gmane.org

Re: [PATCH 5/8] nowait aio: return on congested block device

[Ming Lei]

On Thu, Mar 16, 2017 at 10:33 PM, Jens Axboe <axboe@kernel.dk> wrote:
> On 03/15/2017 03:51 PM, Goldwyn Rodrigues wrote:
>> diff --git a/block/blk-core.c b/block/blk-core.c
>> index 0eeb99e..2e5cba2 100644
>> --- a/block/blk-core.c
>> +++ b/block/blk-core.c
>> @@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
>>       do {
>>               struct request_queue *q = bdev_get_queue(bio->bi_bdev);
>>
>> -             if (likely(blk_queue_enter(q, false) == 0)) {
>> +             if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
>>                       struct bio_list hold;
>>                       struct bio_list lower, same;
>>
>> @@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
>>                       bio_list_merge(&bio_list_on_stack, &same);
>>                       bio_list_merge(&bio_list_on_stack, &hold);
>>               } else {
>> -                     bio_io_error(bio);
>> +                     if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
>> +                             bio_wouldblock_error(bio);
>> +                     else
>> +                             bio_io_error(bio);
>
> This doesn't look right. What if the queue is dying, and BIO_NOWAIT just
> happened to be set?
>
> And you're missing wbt_wait() as well as a blocking point. Ditto in
> blk-mq.

There are also tons of block points in dm/md/bcache, which need to be
considered or be documented at least, :-)



Thanks,
Ming Lei

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 03/16/2017 04:31 PM, Dave Chinner wrote:
> On Wed, Mar 15, 2017 at 04:51:04PM -0500, Goldwyn Rodrigues wrote:
>> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
>>
>> A new flag BIO_NOWAIT is introduced to identify bio's
>> orignating from iocb with IOCB_NOWAIT. This flag indicates
>> to return immediately if a request cannot be made instead
>> of retrying.
> 
> So this makes a congested block device run the bio IO completion
> callback with an -EAGAIN error present? Are all the filesystem
> direct IO submission and completion routines OK with that? i.e. does
> such a congestion case cause filesystems to temporarily expose stale
> data to unprivileged users when the IO is requeued in this way?
> 
> e.g. filesystem does allocation without blocking, submits bio,
> device is congested, runs IO completion with error, so nothing
> written to allocated blocks, write gets queued, so other read
> comes in while the write is queued, reads data from uninitialised
> blocks that were allocated during the write....
> 
> Seems kinda problematic to me to have a undocumented design
> constraint (i.e a landmine) where we submit the AIO only to have it
> error out and then expect the filesystem to do something special and
> different /without blocking/ on EAGAIN.


If the filesystems has to perform block allocation, we would return
-EAGAIN early enough. However, I agree there is a problem, since not all
filesystems know this. I worked on only three of them.

> 
> Why isn't the congestion check at a higher layer like we do for page
> cache readahead? i.e. using the bdi*congested() API at the time we
> are doing all the other filesystem blocking checks.
> 

Yes, that may work better. We will have to call bdi_read_congested() on
a write path. (will have to comment that part of the code). Would it
encompass all possible waits in the block layer?


-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 03/16/2017 09:33 AM, Jens Axboe wrote:
> On 03/15/2017 03:51 PM, Goldwyn Rodrigues wrote:
>> diff --git a/block/blk-core.c b/block/blk-core.c
>> index 0eeb99e..2e5cba2 100644
>> --- a/block/blk-core.c
>> +++ b/block/blk-core.c
>> @@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
>>  	do {
>>  		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
>>  
>> -		if (likely(blk_queue_enter(q, false) == 0)) {
>> +		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
>>  			struct bio_list hold;
>>  			struct bio_list lower, same;
>>  
>> @@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
>>  			bio_list_merge(&bio_list_on_stack, &same);
>>  			bio_list_merge(&bio_list_on_stack, &hold);
>>  		} else {
>> -			bio_io_error(bio);
>> +			if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
>> +				bio_wouldblock_error(bio);
>> +			else
>> +				bio_io_error(bio);
> 
> This doesn't look right. What if the queue is dying, and BIO_NOWAIT just
> happened to be set?

Yes, need a check for that.
> 
> And you're missing wbt_wait() as well as a blocking point. Ditto in
> blk-mq.

Agree.

> 
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index 159187a..942ce8c 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -1518,6 +1518,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
>>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>>  	if (unlikely(!rq)) {
>>  		__wbt_done(q->rq_wb, wb_acct);
>> +		if (bio && bio_flagged(bio, BIO_NOWAIT))
>> +			bio_wouldblock_error(bio);
>>  		return BLK_QC_T_NONE;
>>  	}
>>  
> 
> This seems a little fragile now, since not both paths free the bio.
> 

bio_put() is handled in dio_bio_complete(). However, I am not sure why
there is no bio_endio() call for !rq. IIRC, it was not meant to be.

-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Jan Kara]

On Fri 17-03-17 07:23:24, Goldwyn Rodrigues wrote:
> On 03/16/2017 04:31 PM, Dave Chinner wrote:
> > On Wed, Mar 15, 2017 at 04:51:04PM -0500, Goldwyn Rodrigues wrote:
> >> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
> >>
> >> A new flag BIO_NOWAIT is introduced to identify bio's
> >> orignating from iocb with IOCB_NOWAIT. This flag indicates
> >> to return immediately if a request cannot be made instead
> >> of retrying.
> > 
> > So this makes a congested block device run the bio IO completion
> > callback with an -EAGAIN error present? Are all the filesystem
> > direct IO submission and completion routines OK with that? i.e. does
> > such a congestion case cause filesystems to temporarily expose stale
> > data to unprivileged users when the IO is requeued in this way?
> > 
> > e.g. filesystem does allocation without blocking, submits bio,
> > device is congested, runs IO completion with error, so nothing
> > written to allocated blocks, write gets queued, so other read
> > comes in while the write is queued, reads data from uninitialised
> > blocks that were allocated during the write....
> > 
> > Seems kinda problematic to me to have a undocumented design
> > constraint (i.e a landmine) where we submit the AIO only to have it
> > error out and then expect the filesystem to do something special and
> > different /without blocking/ on EAGAIN.
> 
> If the filesystems has to perform block allocation, we would return
> -EAGAIN early enough. However, I agree there is a problem, since not all
> filesystems know this. I worked on only three of them.

So Dave has a good point about the missing documentation explaining the
expectation from the filesystem. Other than that the filesystem is
responsible for determining in its direct IO submission routine whether it
can deal with NOWAIT direct IO without blocking (including backing out of
block layer refuses to do the IO) and if not, just return with EAGAIN right
away.

Probably we should also have a feature flag in filesystem_type telling
whether a particular filesystem knows about NOWAIT direct IO at all so that
we don't have to add stub to each and every filesystem supporting direct IO
that returns EAGAIN when NOWAIT is set (OTOH adding that stub won't be that
hard either since there are not *that* many filesystems supporting direct
IO). But one or the other has to be done.

> > Why isn't the congestion check at a higher layer like we do for page
> > cache readahead? i.e. using the bdi*congested() API at the time we
> > are doing all the other filesystem blocking checks.
> > 
> 
> Yes, that may work better. We will have to call bdi_read_congested() on
> a write path. (will have to comment that part of the code). Would it
> encompass all possible waits in the block layer?

Congestion mechanism is not really reliable. The bdi can show the device is
not congested but block layer will still block you waiting for some
resource. And for some of the stuff like tag allocation you cannot really
do a reliable check in advance so I don't think congestion checks are
really a viable alternative.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 03/16/2017 09:33 AM, Jens Axboe wrote:
> On 03/15/2017 03:51 PM, Goldwyn Rodrigues wrote:
>> diff --git a/block/blk-core.c b/block/blk-core.c
>> index 0eeb99e..2e5cba2 100644
>> --- a/block/blk-core.c
>> +++ b/block/blk-core.c
>> @@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
>>  	do {
>>  		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
>>  
>> -		if (likely(blk_queue_enter(q, false) == 0)) {
>> +		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
>>  			struct bio_list hold;
>>  			struct bio_list lower, same;
>>  
>> @@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
>>  			bio_list_merge(&bio_list_on_stack, &same);
>>  			bio_list_merge(&bio_list_on_stack, &hold);
>>  		} else {
>> -			bio_io_error(bio);
>> +			if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
>> +				bio_wouldblock_error(bio);
>> +			else
>> +				bio_io_error(bio);
> 
> This doesn't look right. What if the queue is dying, and BIO_NOWAIT just
> happened to be set?
> 

Yes, I need to add a condition here to check for blk_queue_dying(). Thanks.

> And you're missing wbt_wait() as well as a blocking point. Ditto in
> blk-mq.

wbt_wait() does not apply to WRITE_ODIRECT


> 
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index 159187a..942ce8c 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -1518,6 +1518,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
>>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>>  	if (unlikely(!rq)) {
>>  		__wbt_done(q->rq_wb, wb_acct);
>> +		if (bio && bio_flagged(bio, BIO_NOWAIT))
>> +			bio_wouldblock_error(bio);
>>  		return BLK_QC_T_NONE;
>>  	}
>>  
> 
> This seems a little fragile now, since not both paths free the bio.
> 

Direct I/O should free the bios in bio_dio_complete(). I am not sure why
it would not free bio here originally, but IIRC, this path is for
bio==NULL only. So, with this patch we would get a rq==NULL here and
hence the bio_wouldblock_error() call.

-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Jens Axboe]

On 03/24/2017 05:32 AM, Goldwyn Rodrigues wrote:
> 
> 
> On 03/16/2017 09:33 AM, Jens Axboe wrote:
>> On 03/15/2017 03:51 PM, Goldwyn Rodrigues wrote:
>>> diff --git a/block/blk-core.c b/block/blk-core.c
>>> index 0eeb99e..2e5cba2 100644
>>> --- a/block/blk-core.c
>>> +++ b/block/blk-core.c
>>> @@ -2014,7 +2019,7 @@ blk_qc_t generic_make_request(struct bio *bio)
>>>  	do {
>>>  		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
>>>  
>>> -		if (likely(blk_queue_enter(q, false) == 0)) {
>>> +		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
>>>  			struct bio_list hold;
>>>  			struct bio_list lower, same;
>>>  
>>> @@ -2040,7 +2045,10 @@ blk_qc_t generic_make_request(struct bio *bio)
>>>  			bio_list_merge(&bio_list_on_stack, &same);
>>>  			bio_list_merge(&bio_list_on_stack, &hold);
>>>  		} else {
>>> -			bio_io_error(bio);
>>> +			if (unlikely(bio_flagged(bio, BIO_NOWAIT)))
>>> +				bio_wouldblock_error(bio);
>>> +			else
>>> +				bio_io_error(bio);
>>
>> This doesn't look right. What if the queue is dying, and BIO_NOWAIT just
>> happened to be set?
>>
> 
> Yes, I need to add a condition here to check for blk_queue_dying(). Thanks.
> 
>> And you're missing wbt_wait() as well as a blocking point. Ditto in
>> blk-mq.
> 
> wbt_wait() does not apply to WRITE_ODIRECT

It doesn't _currently_ block for it, but that could change. It should
have an explicit check.

Additionally, the more weird assumptions you make, the more half assed
this will be. The other assumption made here is that only direct IO
would set nonblock. We have to assume that any IO can have NONBLOCK set,
and not make any assumptions about the caller.

-- 
Jens Axboe

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 05/11/2017 02:44 AM, Christoph Hellwig wrote:
> Looks fine,
> 
> Reviewed-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
> 
> Although lifting the make_request limit is something a lot of users
> would appreciate in the near future..
> 

Yes, I understand. That will be on my todo list next on priority.

-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Christoph Hellwig]

Looks fine,

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

Although lifting the make_request limit is something a lot of users
would appreciate in the near future..

[PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

A new bio operation flag REQ_NOWAIT is introduced to identify bio's
orignating from iocb with IOCB_NOWAIT. This flag indicates
to return immediately if a request cannot be made instead
of retrying.

Stacked devices such as md (the ones with make_request_fn hooks)
currently are not supported because it may block for housekeeping.
For example, an md can have a part of the device suspended.
For this reason, only request based devices are supported.
In the future, this feature will be expanded to stacked devices
by teaching them how to handle the REQ_NOWAIT flags.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 block/blk-core.c          | 24 ++++++++++++++++++++++--
 block/blk-mq-sched.c      |  3 +++
 block/blk-mq.c            |  4 ++++
 fs/direct-io.c            | 10 ++++++++--
 include/linux/bio.h       |  6 ++++++
 include/linux/blk_types.h |  2 ++
 6 files changed, 45 insertions(+), 4 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index d772c221cc17..effe934b806b 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1232,6 +1232,11 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
 	if (!IS_ERR(rq))
 		return rq;
 
+	if (op & REQ_NOWAIT) {
+		blk_put_rl(rl);
+		return ERR_PTR(-EAGAIN);
+	}
+
 	if (!gfpflags_allow_blocking(gfp_mask) || unlikely(blk_queue_dying(q))) {
 		blk_put_rl(rl);
 		return rq;
@@ -1870,6 +1875,17 @@ generic_make_request_checks(struct bio *bio)
 		goto end_io;
 	}
 
+	/*
+	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
+	 * if queue does not have QUEUE_FLAG_NOWAIT_SUPPORT set
+	 * and if it is not a request based queue.
+	 */
+
+	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_rq_based(q)) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
+
 	part = bio->bi_bdev->bd_part;
 	if (should_fail_request(part, bio->bi_iter.bi_size) ||
 	    should_fail_request(&part_to_disk(part)->part0,
@@ -2021,7 +2037,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 	do {
 		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 
-		if (likely(blk_queue_enter(q, false) == 0)) {
+		if (likely(blk_queue_enter(q, bio->bi_opf & REQ_NOWAIT) == 0)) {
 			struct bio_list lower, same;
 
 			/* Create a fresh bio_list for all subordinate requests */
@@ -2046,7 +2062,11 @@ blk_qc_t generic_make_request(struct bio *bio)
 			bio_list_merge(&bio_list_on_stack[0], &same);
 			bio_list_merge(&bio_list_on_stack[0], &bio_list_on_stack[1]);
 		} else {
-			bio_io_error(bio);
+			if (unlikely(!blk_queue_dying(q) &&
+					(bio->bi_opf & REQ_NOWAIT)))
+				bio_wouldblock_error(bio);
+			else
+				bio_io_error(bio);
 		}
 		bio = bio_list_pop(&bio_list_on_stack[0]);
 	} while (bio);
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index c974a1bbf4cb..019d881d62b7 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
 	if (likely(!data->hctx))
 		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
 
+	if (op & REQ_NOWAIT)
+		data->flags |= BLK_MQ_REQ_NOWAIT;
+
 	if (e) {
 		data->flags |= BLK_MQ_REQ_INTERNAL;
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index c7836a1ded97..d7613ae6a269 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1538,6 +1538,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio->bi_opf & REQ_NOWAIT)
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
@@ -1662,6 +1664,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio->bi_opf & REQ_NOWAIT)
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
diff --git a/fs/direct-io.c b/fs/direct-io.c
index a04ebea77de8..139ebd5ae1c7 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -480,8 +480,12 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
 	unsigned i;
 	int err;
 
-	if (bio->bi_error)
-		dio->io_error = -EIO;
+	if (bio->bi_error) {
+		if (bio->bi_error == -EAGAIN && (bio->bi_opf & REQ_NOWAIT))
+			dio->io_error = -EAGAIN;
+		else
+			dio->io_error = -EIO;
+	}
 
 	if (dio->is_async && dio->op == REQ_OP_READ && dio->should_dirty) {
 		err = bio->bi_error;
@@ -1197,6 +1201,8 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
 	if (iov_iter_rw(iter) == WRITE) {
 		dio->op = REQ_OP_WRITE;
 		dio->op_flags = REQ_SYNC | REQ_IDLE;
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			dio->op_flags |= REQ_NOWAIT;
 	} else {
 		dio->op = REQ_OP_READ;
 	}
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 8e521194f6fc..1a9270744b1e 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -425,6 +425,12 @@ static inline void bio_io_error(struct bio *bio)
 	bio_endio(bio);
 }
 
+static inline void bio_wouldblock_error(struct bio *bio)
+{
+	bio->bi_error = -EAGAIN;
+	bio_endio(bio);
+}
+
 struct request_queue;
 extern int bio_phys_segments(struct request_queue *, struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index d703acb55d0f..5ce4da30ba43 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -187,6 +187,7 @@ enum req_flag_bits {
 	__REQ_PREFLUSH,		/* request for cache flush */
 	__REQ_RAHEAD,		/* read ahead, can fail anytime */
 	__REQ_BACKGROUND,	/* background IO */
+	__REQ_NOWAIT,		/* Don't wait if request will block */
 	__REQ_NR_BITS,		/* stops here */
 };
 
@@ -203,6 +204,7 @@ enum req_flag_bits {
 #define REQ_PREFLUSH		(1ULL << __REQ_PREFLUSH)
 #define REQ_RAHEAD		(1ULL << __REQ_RAHEAD)
 #define REQ_BACKGROUND		(1ULL << __REQ_BACKGROUND)
+#define REQ_NOWAIT		(1ULL << __REQ_NOWAIT)
 
 #define REQ_FAILFAST_MASK \
 	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
-- 
2.12.0

Re: [PATCH 5/8] nowait aio: return on congested block device

[Jens Axboe]

On 04/24/2017 03:10 PM, Goldwyn Rodrigues wrote:
> 
> 
> On 04/19/2017 01:45 AM, Christoph Hellwig wrote:
>> On Fri, Apr 14, 2017 at 07:02:54AM -0500, Goldwyn Rodrigues wrote:
>>> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
>>>
>>
>>> +	/* Request queue supports BIO_NOWAIT */
>>> +	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);
>>
>> BIO_NOWAIT is gone.  And the comment would not be needed if the
>> flag had a more descriptive name, e.g. QUEUE_FLAG_NOWAIT_SUPPORT.
>>
>> And I think all request based drivers should set the flag implicitly
>> as ->queuecommand can't sleep, and ->queue_rq only when it's always
>> offloaded to a workqueue when the BLK_MQ_F_BLOCKING flag is set.
>>
> 
> We introduced QUEUE_FLAG_NOWAIT for devices which would not wait for
> request completions. The ones which wait are MD devices because of sync
> or suspend operations.
> 
> The only user of BLK_MQ_F_NONBLOCKING seems to be nbd. As you mentioned,
> it uses the flag to offload it to a workqueue.
> 
> The other way to do it implicitly is to change the flag to
> BLK_MAY_BLOCK_REQS and use it for devices which do wait such as md/dm.
> Is that what you are hinting at? Or do you have something else in mind?

You are misunderstanding. What Christoph (correctly) says is that request
based drivers do not block, so all of those are fine. What you need to
worry about is drivers that are NOT request based. In other words, drivers
that hook into make_request_fn and process bio's.

-- 
Jens Axboe

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 04/19/2017 01:45 AM, Christoph Hellwig wrote:
> On Fri, Apr 14, 2017 at 07:02:54AM -0500, Goldwyn Rodrigues wrote:
>> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
>>
> 
>> +	/* Request queue supports BIO_NOWAIT */
>> +	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);
> 
> BIO_NOWAIT is gone.  And the comment would not be needed if the
> flag had a more descriptive name, e.g. QUEUE_FLAG_NOWAIT_SUPPORT.
> 
> And I think all request based drivers should set the flag implicitly
> as ->queuecommand can't sleep, and ->queue_rq only when it's always
> offloaded to a workqueue when the BLK_MQ_F_BLOCKING flag is set.
> 

We introduced QUEUE_FLAG_NOWAIT for devices which would not wait for
request completions. The ones which wait are MD devices because of sync
or suspend operations.

The only user of BLK_MQ_F_NONBLOCKING seems to be nbd. As you mentioned,
it uses the flag to offload it to a workqueue.

The other way to do it implicitly is to change the flag to
BLK_MAY_BLOCK_REQS and use it for devices which do wait such as md/dm.
Is that what you are hinting at? Or do you have something else in mind?


-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Jan Kara]

On Wed 19-04-17 10:21:39, Goldwyn Rodrigues wrote:
> 
> 
> On 04/19/2017 01:45 AM, Christoph Hellwig wrote:
> > 
> >> +	if (bio->bi_opf & REQ_NOWAIT) {
> >> +		if (!blk_queue_nowait(q)) {
> >> +			err = -EOPNOTSUPP;
> >> +			goto end_io;
> >> +		}
> >> +		if (!(bio->bi_opf & REQ_SYNC)) {
> > 
> > I don't understand this check at all..
> 
> It is to ensure at the block layer that NOWAIT comes only for DIRECT
> calls only. I should probably change it to REQ_SYNC | REQ_IDLE.

Ouch. Checking 'REQ_SYNC' for this is

a) unreliable hack
b) layering violation

You just don't care why someone marked bio with REQ_NOWAIT at this place.
Just obey the request if you can, return error if you cannot, but advisory
REQ_SYNC or REQ_IDLE flags have nothing to do with the ability of the block
layer to submit the bio without blocking...

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

On 04/19/2017 01:45 AM, Christoph Hellwig wrote:
> 
>> +	if (bio->bi_opf & REQ_NOWAIT) {
>> +		if (!blk_queue_nowait(q)) {
>> +			err = -EOPNOTSUPP;
>> +			goto end_io;
>> +		}
>> +		if (!(bio->bi_opf & REQ_SYNC)) {
> 
> I don't understand this check at all..

It is to ensure at the block layer that NOWAIT comes only for DIRECT
calls only. I should probably change it to REQ_SYNC | REQ_IDLE.

> 
>> +			if (unlikely(!blk_queue_dying(q) && (bio->bi_opf & REQ_NOWAIT)))
> 
> Please break lines after 80 characters.
> 
>> @@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
>>  	if (likely(!data->hctx))
>>  		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
>>  
>> +	if (bio && (bio->bi_opf & REQ_NOWAIT))
>> +		data->flags |= BLK_MQ_REQ_NOWAIT;
> 
> Check the op flag again here.
> 
>> +++ b/block/blk-mq.c
>> @@ -1538,6 +1538,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
>>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>>  	if (unlikely(!rq)) {
>>  		__wbt_done(q->rq_wb, wb_acct);
>> +		if (bio && (bio->bi_opf & REQ_NOWAIT))
>> +			bio_wouldblock_error(bio);
> 
> bio iѕ dereferences unconditionally above, so you can do the same.
> 
>> @@ -1662,6 +1664,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
>>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>>  	if (unlikely(!rq)) {
>>  		__wbt_done(q->rq_wb, wb_acct);
>> +		if (bio && (bio->bi_opf & REQ_NOWAIT))
>> +			bio_wouldblock_error(bio);
> 
> Same here.  Although blk_sq_make_request is gone anyway in the current
> block tree..
> 
>> +	/* Request queue supports BIO_NOWAIT */
>> +	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);
> 
> BIO_NOWAIT is gone.  And the comment would not be needed if the
> flag had a more descriptive name, e.g. QUEUE_FLAG_NOWAIT_SUPPORT.
> 
> And I think all request based drivers should set the flag implicitly
> as ->queuecommand can't sleep, and ->queue_rq only when it's always
> offloaded to a workqueue when the BLK_MQ_F_BLOCKING flag is set.
> 

Yes, Do we have a central point (like a probe() function call?) where
this can be done?


-- 
Goldwyn

Re: [PATCH 5/8] nowait aio: return on congested block device

[Christoph Hellwig]

On Fri, Apr 14, 2017 at 07:02:54AM -0500, Goldwyn Rodrigues wrote:
> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
> 
> A new bio operation flag REQ_NOWAIT is introduced to identify bio's

s/bio/block/

> @@ -1232,6 +1232,11 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
>  	if (!IS_ERR(rq))
>  		return rq;
>  
> +	if (bio && (bio->bi_opf & REQ_NOWAIT)) {
> +		blk_put_rl(rl);
> +		return ERR_PTR(-EAGAIN);
> +	}

Please check the op argument instead of touching bio.

> +	if (bio->bi_opf & REQ_NOWAIT) {
> +		if (!blk_queue_nowait(q)) {
> +			err = -EOPNOTSUPP;
> +			goto end_io;
> +		}
> +		if (!(bio->bi_opf & REQ_SYNC)) {

I don't understand this check at all..

> +			if (unlikely(!blk_queue_dying(q) && (bio->bi_opf & REQ_NOWAIT)))

Please break lines after 80 characters.

> @@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
>  	if (likely(!data->hctx))
>  		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
>  
> +	if (bio && (bio->bi_opf & REQ_NOWAIT))
> +		data->flags |= BLK_MQ_REQ_NOWAIT;

Check the op flag again here.

> +++ b/block/blk-mq.c
> @@ -1538,6 +1538,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>  	if (unlikely(!rq)) {
>  		__wbt_done(q->rq_wb, wb_acct);
> +		if (bio && (bio->bi_opf & REQ_NOWAIT))
> +			bio_wouldblock_error(bio);

bio iѕ dereferences unconditionally above, so you can do the same.

> @@ -1662,6 +1664,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
>  	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
>  	if (unlikely(!rq)) {
>  		__wbt_done(q->rq_wb, wb_acct);
> +		if (bio && (bio->bi_opf & REQ_NOWAIT))
> +			bio_wouldblock_error(bio);

Same here.  Although blk_sq_make_request is gone anyway in the current
block tree..

> +	/* Request queue supports BIO_NOWAIT */
> +	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);

BIO_NOWAIT is gone.  And the comment would not be needed if the
flag had a more descriptive name, e.g. QUEUE_FLAG_NOWAIT_SUPPORT.

And I think all request based drivers should set the flag implicitly
as ->queuecommand can't sleep, and ->queue_rq only when it's always
offloaded to a workqueue when the BLK_MQ_F_BLOCKING flag is set.

> --- a/fs/direct-io.c
> +++ b/fs/direct-io.c
> @@ -480,8 +480,12 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
>  	unsigned i;
>  	int err;
>  
> -	if (bio->bi_error)
> -		dio->io_error = -EIO;
> +	if (bio->bi_error) {
> +		if (bio->bi_opf & REQ_NOWAIT)
> +			dio->io_error = -EAGAIN;
> +		else
> +			dio->io_error = -EIO;
> +	}

Huh?  Once REQ_NOWAIT is set all errors are -EAGAIN?

[PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

A new bio operation flag REQ_NOWAIT is introduced to identify bio's
orignating from iocb with IOCB_NOWAIT. This flag indicates
to return immediately if a request cannot be made instead
of retrying.

To facilitate this, QUEUE_FLAG_NOWAIT is set to devices
which support this. While currently this is set to
virtio and sd only. Support to more devices will be added soon
once I am sure they don't block. Currently blocks such as dm/md
block while performing sync.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 block/blk-core.c           | 24 ++++++++++++++++++++++--
 block/blk-mq-sched.c       |  3 +++
 block/blk-mq.c             |  4 ++++
 drivers/block/virtio_blk.c |  3 +++
 drivers/scsi/sd.c          |  3 +++
 fs/direct-io.c             | 10 ++++++++--
 include/linux/bio.h        |  6 ++++++
 include/linux/blk_types.h  |  2 ++
 include/linux/blkdev.h     |  3 +++
 9 files changed, 54 insertions(+), 4 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index d772c221cc17..54698521756b 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1232,6 +1232,11 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
 	if (!IS_ERR(rq))
 		return rq;
 
+	if (bio && (bio->bi_opf & REQ_NOWAIT)) {
+		blk_put_rl(rl);
+		return ERR_PTR(-EAGAIN);
+	}
+
 	if (!gfpflags_allow_blocking(gfp_mask) || unlikely(blk_queue_dying(q))) {
 		blk_put_rl(rl);
 		return rq;
@@ -1870,6 +1875,18 @@ generic_make_request_checks(struct bio *bio)
 		goto end_io;
 	}
 
+	if (bio->bi_opf & REQ_NOWAIT) {
+		if (!blk_queue_nowait(q)) {
+			err = -EOPNOTSUPP;
+			goto end_io;
+		}
+		if (!(bio->bi_opf & REQ_SYNC)) {
+			err = -EINVAL;
+			goto end_io;
+		}
+	}
+
+
 	part = bio->bi_bdev->bd_part;
 	if (should_fail_request(part, bio->bi_iter.bi_size) ||
 	    should_fail_request(&part_to_disk(part)->part0,
@@ -2021,7 +2038,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 	do {
 		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 
-		if (likely(blk_queue_enter(q, false) == 0)) {
+		if (likely(blk_queue_enter(q, bio->bi_opf & REQ_NOWAIT) == 0)) {
 			struct bio_list lower, same;
 
 			/* Create a fresh bio_list for all subordinate requests */
@@ -2046,7 +2063,10 @@ blk_qc_t generic_make_request(struct bio *bio)
 			bio_list_merge(&bio_list_on_stack[0], &same);
 			bio_list_merge(&bio_list_on_stack[0], &bio_list_on_stack[1]);
 		} else {
-			bio_io_error(bio);
+			if (unlikely(!blk_queue_dying(q) && (bio->bi_opf & REQ_NOWAIT)))
+				bio_wouldblock_error(bio);
+			else
+				bio_io_error(bio);
 		}
 		bio = bio_list_pop(&bio_list_on_stack[0]);
 	} while (bio);
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index c974a1bbf4cb..9f88190ff395 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
 	if (likely(!data->hctx))
 		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
 
+	if (bio && (bio->bi_opf & REQ_NOWAIT))
+		data->flags |= BLK_MQ_REQ_NOWAIT;
+
 	if (e) {
 		data->flags |= BLK_MQ_REQ_INTERNAL;
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 572966f49596..8b9b1a411ce2 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1538,6 +1538,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && (bio->bi_opf & REQ_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
@@ -1662,6 +1664,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && (bio->bi_opf & REQ_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 1d4c9f8bc1e1..7481124c5025 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -731,6 +731,9 @@ static int virtblk_probe(struct virtio_device *vdev)
 	/* No real sector limit. */
 	blk_queue_max_hw_sectors(q, -1U);
 
+	/* Request queue supports BIO_NOWAIT */
+	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);
+
 	/* Host can optionally specify maximum segment size and number of
 	 * segments. */
 	err = virtio_cread_feature(vdev, VIRTIO_BLK_F_SIZE_MAX,
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index fcfeddc79331..9df85ee165be 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -3177,6 +3177,9 @@ static int sd_probe(struct device *dev)
 					     SD_MOD_TIMEOUT);
 	}
 
+	/* Support BIO_NOWAIT */
+	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, sdp->request_queue);
+
 	device_initialize(&sdkp->dev);
 	sdkp->dev.parent = dev;
 	sdkp->dev.class = &sd_disk_class;
diff --git a/fs/direct-io.c b/fs/direct-io.c
index a04ebea77de8..a802168284e1 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -480,8 +480,12 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
 	unsigned i;
 	int err;
 
-	if (bio->bi_error)
-		dio->io_error = -EIO;
+	if (bio->bi_error) {
+		if (bio->bi_opf & REQ_NOWAIT)
+			dio->io_error = -EAGAIN;
+		else
+			dio->io_error = -EIO;
+	}
 
 	if (dio->is_async && dio->op == REQ_OP_READ && dio->should_dirty) {
 		err = bio->bi_error;
@@ -1197,6 +1201,8 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
 	if (iov_iter_rw(iter) == WRITE) {
 		dio->op = REQ_OP_WRITE;
 		dio->op_flags = REQ_SYNC | REQ_IDLE;
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			dio->op_flags |= REQ_NOWAIT;
 	} else {
 		dio->op = REQ_OP_READ;
 	}
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 8e521194f6fc..1a9270744b1e 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -425,6 +425,12 @@ static inline void bio_io_error(struct bio *bio)
 	bio_endio(bio);
 }
 
+static inline void bio_wouldblock_error(struct bio *bio)
+{
+	bio->bi_error = -EAGAIN;
+	bio_endio(bio);
+}
+
 struct request_queue;
 extern int bio_phys_segments(struct request_queue *, struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index d703acb55d0f..5ce4da30ba43 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -187,6 +187,7 @@ enum req_flag_bits {
 	__REQ_PREFLUSH,		/* request for cache flush */
 	__REQ_RAHEAD,		/* read ahead, can fail anytime */
 	__REQ_BACKGROUND,	/* background IO */
+	__REQ_NOWAIT,		/* Don't wait if request will block */
 	__REQ_NR_BITS,		/* stops here */
 };
 
@@ -203,6 +204,7 @@ enum req_flag_bits {
 #define REQ_PREFLUSH		(1ULL << __REQ_PREFLUSH)
 #define REQ_RAHEAD		(1ULL << __REQ_RAHEAD)
 #define REQ_BACKGROUND		(1ULL << __REQ_BACKGROUND)
+#define REQ_NOWAIT		(1ULL << __REQ_NOWAIT)
 
 #define REQ_FAILFAST_MASK \
 	(REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 7548f332121a..df0b1245d955 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -610,6 +610,8 @@ struct request_queue {
 #define QUEUE_FLAG_FLUSH_NQ    25	/* flush not queueuable */
 #define QUEUE_FLAG_DAX         26	/* device supports DAX */
 #define QUEUE_FLAG_STATS       27	/* track rq completion times */
+/* can return immediately on congestion  (for REQ_NOWAIT) */
+#define QUEUE_FLAG_NOWAIT      28
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_STACKABLE)	|	\
@@ -700,6 +702,7 @@ static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
 #define blk_queue_secure_erase(q) \
 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
+#define blk_queue_nowait(q)	test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
 
 #define blk_noretry_request(rq) \
 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
-- 
2.12.0

Re: [PATCH 5/8] nowait aio: return on congested block device

[Christoph Hellwig]

Please make this a REQ_* flag so that it can be passed in the bio,
the request and as an argument to the get_request functions instead
of testing for a bio.

[PATCH 5/8] nowait aio: return on congested block device

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

A new flag BIO_NOWAIT is introduced to identify bio's
orignating from iocb with IOCB_NOWAIT. This flag indicates
to return immediately if a request cannot be made instead
of retrying.

To facilitate this, QUEUE_FLAG_NOWAIT is set to devices
which support this. While currently this is set to
virtio and sd only. Support to more devices will be added soon.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 block/blk-core.c           | 24 ++++++++++++++++++++++--
 block/blk-mq-sched.c       |  3 +++
 block/blk-mq.c             |  4 ++++
 drivers/block/virtio_blk.c |  3 +++
 drivers/scsi/sd.c          |  3 +++
 fs/direct-io.c             | 11 +++++++++--
 include/linux/bio.h        |  6 ++++++
 include/linux/blk_types.h  |  1 +
 include/linux/blkdev.h     |  2 ++
 9 files changed, 53 insertions(+), 4 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index d772c22..95a9b18 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1232,6 +1232,11 @@ static struct request *get_request(struct request_queue *q, unsigned int op,
 	if (!IS_ERR(rq))
 		return rq;
 
+	if (bio && bio_flagged(bio, BIO_NOWAIT)) {
+		blk_put_rl(rl);
+		return ERR_PTR(-EAGAIN);
+	}
+
 	if (!gfpflags_allow_blocking(gfp_mask) || unlikely(blk_queue_dying(q))) {
 		blk_put_rl(rl);
 		return rq;
@@ -1870,6 +1875,18 @@ generic_make_request_checks(struct bio *bio)
 		goto end_io;
 	}
 
+	if (bio_flagged(bio, BIO_NOWAIT)) {
+		if (!blk_queue_nowait(q)) {
+			err = -EOPNOTSUPP;
+			goto end_io;
+		}
+		if (!(bio->bi_opf & (REQ_SYNC | REQ_IDLE))) {
+			err = -EINVAL;
+			goto end_io;
+		}
+	}
+
+
 	part = bio->bi_bdev->bd_part;
 	if (should_fail_request(part, bio->bi_iter.bi_size) ||
 	    should_fail_request(&part_to_disk(part)->part0,
@@ -2021,7 +2038,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 	do {
 		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
 
-		if (likely(blk_queue_enter(q, false) == 0)) {
+		if (likely(blk_queue_enter(q, bio_flagged(bio, BIO_NOWAIT)) == 0)) {
 			struct bio_list lower, same;
 
 			/* Create a fresh bio_list for all subordinate requests */
@@ -2046,7 +2063,10 @@ blk_qc_t generic_make_request(struct bio *bio)
 			bio_list_merge(&bio_list_on_stack[0], &same);
 			bio_list_merge(&bio_list_on_stack[0], &bio_list_on_stack[1]);
 		} else {
-			bio_io_error(bio);
+			if (unlikely(!blk_queue_dying(q) && bio_flagged(bio, BIO_NOWAIT)))
+				bio_wouldblock_error(bio);
+			else
+				bio_io_error(bio);
 		}
 		bio = bio_list_pop(&bio_list_on_stack[0]);
 	} while (bio);
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 09af8ff..40e78b5 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -119,6 +119,9 @@ struct request *blk_mq_sched_get_request(struct request_queue *q,
 	if (likely(!data->hctx))
 		data->hctx = blk_mq_map_queue(q, data->ctx->cpu);
 
+	if (likely(bio) && bio_flagged(bio, BIO_NOWAIT))
+		data->flags |= BLK_MQ_REQ_NOWAIT;
+
 	if (e) {
 		data->flags |= BLK_MQ_REQ_INTERNAL;
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 6b6e7bc..2d90b12 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1511,6 +1511,8 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && bio_flagged(bio, BIO_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
@@ -1635,6 +1637,8 @@ static blk_qc_t blk_sq_make_request(struct request_queue *q, struct bio *bio)
 	rq = blk_mq_sched_get_request(q, bio, bio->bi_opf, &data);
 	if (unlikely(!rq)) {
 		__wbt_done(q->rq_wb, wb_acct);
+		if (bio && bio_flagged(bio, BIO_NOWAIT))
+			bio_wouldblock_error(bio);
 		return BLK_QC_T_NONE;
 	}
 
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 1d4c9f8..7481124 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -731,6 +731,9 @@ static int virtblk_probe(struct virtio_device *vdev)
 	/* No real sector limit. */
 	blk_queue_max_hw_sectors(q, -1U);
 
+	/* Request queue supports BIO_NOWAIT */
+	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, q);
+
 	/* Host can optionally specify maximum segment size and number of
 	 * segments. */
 	err = virtio_cread_feature(vdev, VIRTIO_BLK_F_SIZE_MAX,
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index fcfeddc..9df85ee 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -3177,6 +3177,9 @@ static int sd_probe(struct device *dev)
 					     SD_MOD_TIMEOUT);
 	}
 
+	/* Support BIO_NOWAIT */
+	queue_flag_set_unlocked(QUEUE_FLAG_NOWAIT, sdp->request_queue);
+
 	device_initialize(&sdkp->dev);
 	sdkp->dev.parent = dev;
 	sdkp->dev.class = &sd_disk_class;
diff --git a/fs/direct-io.c b/fs/direct-io.c
index a04ebea..f6835d3 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -386,6 +386,9 @@ dio_bio_alloc(struct dio *dio, struct dio_submit *sdio,
 	else
 		bio->bi_end_io = dio_bio_end_io;
 
+	if (dio->iocb->ki_flags & IOCB_NOWAIT)
+		bio_set_flag(bio, BIO_NOWAIT);
+
 	sdio->bio = bio;
 	sdio->logical_offset_in_bio = sdio->cur_page_fs_offset;
 }
@@ -480,8 +483,12 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
 	unsigned i;
 	int err;
 
-	if (bio->bi_error)
-		dio->io_error = -EIO;
+	if (bio->bi_error) {
+		if (bio_flagged(bio, BIO_NOWAIT))
+			dio->io_error = -EAGAIN;
+		else
+			dio->io_error = -EIO;
+	}
 
 	if (dio->is_async && dio->op == REQ_OP_READ && dio->should_dirty) {
 		err = bio->bi_error;
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 8e52119..1a92707 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -425,6 +425,12 @@ static inline void bio_io_error(struct bio *bio)
 	bio_endio(bio);
 }
 
+static inline void bio_wouldblock_error(struct bio *bio)
+{
+	bio->bi_error = -EAGAIN;
+	bio_endio(bio);
+}
+
 struct request_queue;
 extern int bio_phys_segments(struct request_queue *, struct bio *);
 
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index d703acb..514c08e 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -102,6 +102,7 @@ struct bio {
 #define BIO_REFFED	8	/* bio has elevated ->bi_cnt */
 #define BIO_THROTTLED	9	/* This bio has already been subjected to
 				 * throttling rules. Don't do it again. */
+#define BIO_NOWAIT	10	/* don't block over blk device congestion */
 
 /*
  * Flags starting here get preserved by bio_reset() - this includes
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 5a7da60..ae38ab6 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -611,6 +611,7 @@ struct request_queue {
 #define QUEUE_FLAG_DAX         26	/* device supports DAX */
 #define QUEUE_FLAG_STATS       27	/* track rq completion times */
 #define QUEUE_FLAG_RESTART     28	/* queue needs restart at completion */
+#define QUEUE_FLAG_NOWAIT      29	/* queue supports BIO_NOWAIT */
 
 #define QUEUE_FLAG_DEFAULT	((1 << QUEUE_FLAG_IO_STAT) |		\
 				 (1 << QUEUE_FLAG_STACKABLE)	|	\
@@ -701,6 +702,7 @@ static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
 #define blk_queue_secure_erase(q) \
 	(test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
 #define blk_queue_dax(q)	test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
+#define blk_queue_nowait(q)	test_bit(QUEUE_FLAG_NOWAIT, &(q)->queue_flags)
 
 #define blk_noretry_request(rq) \
 	((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
-- 
2.10.2