Merge tag 'iomap-5.12-merge-2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull iomap updates from Darrick Wong:
 "The big change in this cycle is some new code to make it possible for
  XFS to try unaligned directio overwrites without taking locks. If the
  block is fully written and within EOF (i.e. doesn't require any
  further fs intervention) then we can let the unlocked write proceed.
  If not, we fall back to synchronizing direct writes.

  Summary:

   - Adjust the final parameter of iomap_dio_rw.

   - Add a new flag to request that iomap directio writes return EAGAIN
     if the write is not a pure overwrite within EOF; this will be used
     to reduce lock contention with unaligned direct writes on XFS.

   - Amend XFS' directio code to eliminate exclusive locking for
     unaligned direct writes if the circumstances permit"

* tag 'iomap-5.12-merge-2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: reduce exclusive locking on unaligned dio
  xfs: split the unaligned DIO write code out
  xfs: improve the reflink_bounce_dio_write tracepoint
  xfs: simplify the read/write tracepoints
  xfs: remove the buffered I/O fallback assert
  xfs: cleanup the read/write helper naming
  xfs: make xfs_file_aio_write_checks IOCB_NOWAIT-aware
  xfs: factor out a xfs_ilock_iocb helper
  iomap: add a IOMAP_DIO_OVERWRITE_ONLY flag
  iomap: pass a flags argument to iomap_dio_rw
  iomap: rename the flags variable in __iomap_dio_rw

9 files changed, 253 insertions, 184 deletions

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 01a72f53fb5d..be9e3900cce8 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1942,8 +1942,8 @@ relock:
 		goto buffered;
 	}
 
-	dio = __iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops,
-			     &btrfs_dio_ops, is_sync_kiocb(iocb));
+	dio = __iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops, &btrfs_dio_ops,
+			     0);
 
 	btrfs_inode_unlock(inode, ilock_flags);
 
@@ -3618,8 +3618,7 @@ static ssize_t btrfs_direct_read(struct kiocb *iocb, struct iov_iter *to)
 		return 0;
 
 	btrfs_inode_lock(inode, BTRFS_ILOCK_SHARED);
-	ret = iomap_dio_rw(iocb, to, &btrfs_dio_iomap_ops, &btrfs_dio_ops,
-			   is_sync_kiocb(iocb));
+	ret = iomap_dio_rw(iocb, to, &btrfs_dio_iomap_ops, &btrfs_dio_ops, 0);
 	btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED);
 	return ret;
 }
diff --git a/fs/ext4/file.c b/fs/ext4/file.c
index 349b27f0dda0..194f5d00fa32 100644
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -74,8 +74,7 @@ static ssize_t ext4_dio_read_iter(struct kiocb *iocb, struct iov_iter *to)
 		return generic_file_read_iter(iocb, to);
 	}
 
-	ret = iomap_dio_rw(iocb, to, &ext4_iomap_ops, NULL,
-			   is_sync_kiocb(iocb));
+	ret = iomap_dio_rw(iocb, to, &ext4_iomap_ops, NULL, 0);
 	inode_unlock_shared(inode);
 
 	file_accessed(iocb->ki_filp);
@@ -550,7 +549,7 @@ static ssize_t ext4_dio_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (ilock_shared)
 		iomap_ops = &ext4_iomap_overwrite_ops;
 	ret = iomap_dio_rw(iocb, from, iomap_ops, &ext4_dio_write_ops,
-			   is_sync_kiocb(iocb) || unaligned_io || extend);
+			   (unaligned_io || extend) ? IOMAP_DIO_FORCE_WAIT : 0);
 	if (ret == -ENOTBLK)
 		ret = 0;
 
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index b39b339feddc..89609c299717 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -797,9 +797,7 @@ static ssize_t gfs2_file_direct_read(struct kiocb *iocb, struct iov_iter *to,
 	if (ret)
 		goto out_uninit;
 
-	ret = iomap_dio_rw(iocb, to, &gfs2_iomap_ops, NULL,
-			   is_sync_kiocb(iocb));
-
+	ret = iomap_dio_rw(iocb, to, &gfs2_iomap_ops, NULL, 0);
 	gfs2_glock_dq(gh);
 out_uninit:
 	gfs2_holder_uninit(gh);
@@ -833,8 +831,7 @@ static ssize_t gfs2_file_direct_write(struct kiocb *iocb, struct iov_iter *from,
 	if (offset + len > i_size_read(&ip->i_inode))
 		goto out;
 
-	ret = iomap_dio_rw(iocb, from, &gfs2_iomap_ops, NULL,
-			   is_sync_kiocb(iocb));
+	ret = iomap_dio_rw(iocb, from, &gfs2_iomap_ops, NULL, 0);
 	if (ret == -ENOTBLK)
 		ret = 0;
 out:
diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
index 2273120d8ed7..e4c258a2cefc 100644
--- a/fs/iomap/direct-io.c
+++ b/fs/iomap/direct-io.c
@@ -451,23 +451,22 @@ iomap_dio_actor(struct inode *inode, loff_t pos, loff_t length,
 struct iomap_dio *
 __iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 		const struct iomap_ops *ops, const struct iomap_dio_ops *dops,
-		bool wait_for_completion)
+		unsigned int dio_flags)
 {
 	struct address_space *mapping = iocb->ki_filp->f_mapping;
 	struct inode *inode = file_inode(iocb->ki_filp);
 	size_t count = iov_iter_count(iter);
 	loff_t pos = iocb->ki_pos;
 	loff_t end = iocb->ki_pos + count - 1, ret = 0;
-	unsigned int flags = IOMAP_DIRECT;
+	bool wait_for_completion =
+		is_sync_kiocb(iocb) || (dio_flags & IOMAP_DIO_FORCE_WAIT);
+	unsigned int iomap_flags = IOMAP_DIRECT;
 	struct blk_plug plug;
 	struct iomap_dio *dio;
 
 	if (!count)
 		return NULL;
 
-	if (WARN_ON(is_sync_kiocb(iocb) && !wait_for_completion))
-		return ERR_PTR(-EIO);
-
 	dio = kmalloc(sizeof(*dio), GFP_KERNEL);
 	if (!dio)
 		return ERR_PTR(-ENOMEM);
@@ -492,7 +491,7 @@ __iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 		if (iter_is_iovec(iter))
 			dio->flags |= IOMAP_DIO_DIRTY;
 	} else {
-		flags |= IOMAP_WRITE;
+		iomap_flags |= IOMAP_WRITE;
 		dio->flags |= IOMAP_DIO_WRITE;
 
 		/* for data sync or sync, we need sync completion processing */
@@ -514,7 +513,14 @@ __iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 			ret = -EAGAIN;
 			goto out_free_dio;
 		}
-		flags |= IOMAP_NOWAIT;
+		iomap_flags |= IOMAP_NOWAIT;
+	}
+
+	if (dio_flags & IOMAP_DIO_OVERWRITE_ONLY) {
+		ret = -EAGAIN;
+		if (pos >= dio->i_size || pos + count > dio->i_size)
+			goto out_free_dio;
+		iomap_flags |= IOMAP_OVERWRITE_ONLY;
 	}
 
 	ret = filemap_write_and_wait_range(mapping, pos, end);
@@ -545,7 +551,7 @@ __iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 
 	blk_start_plug(&plug);
 	do {
-		ret = iomap_apply(inode, pos, count, flags, ops, dio,
+		ret = iomap_apply(inode, pos, count, iomap_flags, ops, dio,
 				iomap_dio_actor);
 		if (ret <= 0) {
 			/* magic error code to fall back to buffered I/O */
@@ -629,11 +635,11 @@ EXPORT_SYMBOL_GPL(__iomap_dio_rw);
 ssize_t
 iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 		const struct iomap_ops *ops, const struct iomap_dio_ops *dops,
-		bool wait_for_completion)
+		unsigned int dio_flags)
 {
 	struct iomap_dio *dio;
 
-	dio = __iomap_dio_rw(iocb, iter, ops, dops, wait_for_completion);
+	dio = __iomap_dio_rw(iocb, iter, ops, dops, dio_flags);
 	if (IS_ERR_OR_NULL(dio))
 		return PTR_ERR_OR_ZERO(dio);
 	return iomap_dio_complete(dio);
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 5b0f93f73837..39695b59dfcc 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -197,30 +197,42 @@ xfs_file_fsync(
 	return error;
 }
 
+static int
+xfs_ilock_iocb(
+	struct kiocb		*iocb,
+	unsigned int		lock_mode)
+{
+	struct xfs_inode	*ip = XFS_I(file_inode(iocb->ki_filp));
+
+	if (iocb->ki_flags & IOCB_NOWAIT) {
+		if (!xfs_ilock_nowait(ip, lock_mode))
+			return -EAGAIN;
+	} else {
+		xfs_ilock(ip, lock_mode);
+	}
+
+	return 0;
+}
+
 STATIC ssize_t
-xfs_file_dio_aio_read(
+xfs_file_dio_read(
 	struct kiocb		*iocb,
 	struct iov_iter		*to)
 {
 	struct xfs_inode	*ip = XFS_I(file_inode(iocb->ki_filp));
-	size_t			count = iov_iter_count(to);
 	ssize_t			ret;
 
-	trace_xfs_file_direct_read(ip, count, iocb->ki_pos);
+	trace_xfs_file_direct_read(iocb, to);
 
-	if (!count)
+	if (!iov_iter_count(to))
 		return 0; /* skip atime */
 
 	file_accessed(iocb->ki_filp);
 
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		if (!xfs_ilock_nowait(ip, XFS_IOLOCK_SHARED))
-			return -EAGAIN;
-	} else {
-		xfs_ilock(ip, XFS_IOLOCK_SHARED);
-	}
-	ret = iomap_dio_rw(iocb, to, &xfs_read_iomap_ops, NULL,
-			is_sync_kiocb(iocb));
+	ret = xfs_ilock_iocb(iocb, XFS_IOLOCK_SHARED);
+	if (ret)
+		return ret;
+	ret = iomap_dio_rw(iocb, to, &xfs_read_iomap_ops, NULL, 0);
 	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
 	return ret;
@@ -232,21 +244,16 @@ xfs_file_dax_read(
 	struct iov_iter		*to)
 {
 	struct xfs_inode	*ip = XFS_I(iocb->ki_filp->f_mapping->host);
-	size_t			count = iov_iter_count(to);
 	ssize_t			ret = 0;
 
-	trace_xfs_file_dax_read(ip, count, iocb->ki_pos);
+	trace_xfs_file_dax_read(iocb, to);
 
-	if (!count)
+	if (!iov_iter_count(to))
 		return 0; /* skip atime */
 
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		if (!xfs_ilock_nowait(ip, XFS_IOLOCK_SHARED))
-			return -EAGAIN;
-	} else {
-		xfs_ilock(ip, XFS_IOLOCK_SHARED);
-	}
-
+	ret = xfs_ilock_iocb(iocb, XFS_IOLOCK_SHARED);
+	if (ret)
+		return ret;
 	ret = dax_iomap_rw(iocb, to, &xfs_read_iomap_ops);
 	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
@@ -255,21 +262,18 @@ xfs_file_dax_read(
 }
 
 STATIC ssize_t
-xfs_file_buffered_aio_read(
+xfs_file_buffered_read(
 	struct kiocb		*iocb,
 	struct iov_iter		*to)
 {
 	struct xfs_inode	*ip = XFS_I(file_inode(iocb->ki_filp));
 	ssize_t			ret;
 
-	trace_xfs_file_buffered_read(ip, iov_iter_count(to), iocb->ki_pos);
+	trace_xfs_file_buffered_read(iocb, to);
 
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		if (!xfs_ilock_nowait(ip, XFS_IOLOCK_SHARED))
-			return -EAGAIN;
-	} else {
-		xfs_ilock(ip, XFS_IOLOCK_SHARED);
-	}
+	ret = xfs_ilock_iocb(iocb, XFS_IOLOCK_SHARED);
+	if (ret)
+		return ret;
 	ret = generic_file_read_iter(iocb, to);
 	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
@@ -293,9 +297,9 @@ xfs_file_read_iter(
 	if (IS_DAX(inode))
 		ret = xfs_file_dax_read(iocb, to);
 	else if (iocb->ki_flags & IOCB_DIRECT)
-		ret = xfs_file_dio_aio_read(iocb, to);
+		ret = xfs_file_dio_read(iocb, to);
 	else
-		ret = xfs_file_buffered_aio_read(iocb, to);
+		ret = xfs_file_buffered_read(iocb, to);
 
 	if (ret > 0)
 		XFS_STATS_ADD(mp, xs_read_bytes, ret);
@@ -310,7 +314,7 @@ xfs_file_read_iter(
  * if called for a direct write beyond i_size.
  */
 STATIC ssize_t
-xfs_file_aio_write_checks(
+xfs_file_write_checks(
 	struct kiocb		*iocb,
 	struct iov_iter		*from,
 	int			*iolock)
@@ -328,7 +332,14 @@ restart:
 	if (error <= 0)
 		return error;
 
-	error = xfs_break_layouts(inode, iolock, BREAK_WRITE);
+	if (iocb->ki_flags & IOCB_NOWAIT) {
+		error = break_layout(inode, false);
+		if (error == -EWOULDBLOCK)
+			error = -EAGAIN;
+	} else {
+		error = xfs_break_layouts(inode, iolock, BREAK_WRITE);
+	}
+
 	if (error)
 		return error;
 
@@ -339,7 +350,11 @@ restart:
 	if (*iolock == XFS_IOLOCK_SHARED && !IS_NOSEC(inode)) {
 		xfs_iunlock(ip, *iolock);
 		*iolock = XFS_IOLOCK_EXCL;
-		xfs_ilock(ip, *iolock);
+		error = xfs_ilock_iocb(iocb, *iolock);
+		if (error) {
+			*iolock = 0;
+			return error;
+		}
 		goto restart;
 	}
 	/*
@@ -361,6 +376,10 @@ restart:
 	isize = i_size_read(inode);
 	if (iocb->ki_pos > isize) {
 		spin_unlock(&ip->i_flags_lock);
+
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			return -EAGAIN;
+
 		if (!drained_dio) {
 			if (*iolock == XFS_IOLOCK_SHARED) {
 				xfs_iunlock(ip, *iolock);
@@ -480,122 +499,149 @@ static const struct iomap_dio_ops xfs_dio_write_ops = {
 };
 
 /*
- * xfs_file_dio_aio_write - handle direct IO writes
- *
- * Lock the inode appropriately to prepare for and issue a direct IO write.
- * By separating it from the buffered write path we remove all the tricky to
- * follow locking changes and looping.
- *
- * If there are cached pages or we're extending the file, we need IOLOCK_EXCL
- * until we're sure the bytes at the new EOF have been zeroed and/or the cached
- * pages are flushed out.
- *
- * In most cases the direct IO writes will be done holding IOLOCK_SHARED
- * allowing them to be done in parallel with reads and other direct IO writes.
- * However, if the IO is not aligned to filesystem blocks, the direct IO layer
- * needs to do sub-block zeroing and that requires serialisation against other
- * direct IOs to the same block. In this case we need to serialise the
- * submission of the unaligned IOs so that we don't get racing block zeroing in
- * the dio layer.  To avoid the problem with aio, we also need to wait for
- * outstanding IOs to complete so that unwritten extent conversion is completed
- * before we try to map the overlapping block. This is currently implemented by
- * hitting it with a big hammer (i.e. inode_dio_wait()).
- *
- * Returns with locks held indicated by @iolock and errors indicated by
- * negative return values.
+ * Handle block aligned direct I/O writes
  */
-STATIC ssize_t
-xfs_file_dio_aio_write(
+static noinline ssize_t
+xfs_file_dio_write_aligned(
+	struct xfs_inode	*ip,
 	struct kiocb		*iocb,
 	struct iov_iter		*from)
 {
-	struct file		*file = iocb->ki_filp;
-	struct address_space	*mapping = file->f_mapping;
-	struct inode		*inode = mapping->host;
-	struct xfs_inode	*ip = XFS_I(inode);
-	struct xfs_mount	*mp = ip->i_mount;
-	ssize_t			ret = 0;
-	int			unaligned_io = 0;
-	int			iolock;
-	size_t			count = iov_iter_count(from);
-	struct xfs_buftarg      *target = xfs_inode_buftarg(ip);
+	int			iolock = XFS_IOLOCK_SHARED;
+	ssize_t			ret;
 
-	/* DIO must be aligned to device logical sector size */
-	if ((iocb->ki_pos | count) & target->bt_logical_sectormask)
-		return -EINVAL;
+	ret = xfs_ilock_iocb(iocb, iolock);
+	if (ret)
+		return ret;
+	ret = xfs_file_write_checks(iocb, from, &iolock);
+	if (ret)
+		goto out_unlock;
 
 	/*
-	 * Don't take the exclusive iolock here unless the I/O is unaligned to
-	 * the file system block size.  We don't need to consider the EOF
-	 * extension case here because xfs_file_aio_write_checks() will relock
-	 * the inode as necessary for EOF zeroing cases and fill out the new
-	 * inode size as appropriate.
+	 * We don't need to hold the IOLOCK exclusively across the IO, so demote
+	 * the iolock back to shared if we had to take the exclusive lock in
+	 * xfs_file_write_checks() for other reasons.
 	 */
-	if ((iocb->ki_pos & mp->m_blockmask) ||
-	    ((iocb->ki_pos + count) & mp->m_blockmask)) {
-		unaligned_io = 1;
-
-		/*
-		 * We can't properly handle unaligned direct I/O to reflink
-		 * files yet, as we can't unshare a partial block.
-		 */
-		if (xfs_is_cow_inode(ip)) {
-			trace_xfs_reflink_bounce_dio_write(ip, iocb->ki_pos, count);
-			return -ENOTBLK;
-		}
-		iolock = XFS_IOLOCK_EXCL;
-	} else {
+	if (iolock == XFS_IOLOCK_EXCL) {
+		xfs_ilock_demote(ip, XFS_IOLOCK_EXCL);
 		iolock = XFS_IOLOCK_SHARED;
 	}
+	trace_xfs_file_direct_write(iocb, from);
+	ret = iomap_dio_rw(iocb, from, &xfs_direct_write_iomap_ops,
+			   &xfs_dio_write_ops, 0);
+out_unlock:
+	if (iolock)
+		xfs_iunlock(ip, iolock);
+	return ret;
+}
 
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		/* unaligned dio always waits, bail */
-		if (unaligned_io)
-			return -EAGAIN;
-		if (!xfs_ilock_nowait(ip, iolock))
+/*
+ * Handle block unaligned direct I/O writes
+ *
+ * In most cases direct I/O writes will be done holding IOLOCK_SHARED, allowing
+ * them to be done in parallel with reads and other direct I/O writes.  However,
+ * if the I/O is not aligned to filesystem blocks, the direct I/O layer may need
+ * to do sub-block zeroing and that requires serialisation against other direct
+ * I/O to the same block.  In this case we need to serialise the submission of
+ * the unaligned I/O so that we don't get racing block zeroing in the dio layer.
+ * In the case where sub-block zeroing is not required, we can do concurrent
+ * sub-block dios to the same block successfully.
+ *
+ * Optimistically submit the I/O using the shared lock first, but use the
+ * IOMAP_DIO_OVERWRITE_ONLY flag to tell the lower layers to return -EAGAIN
+ * if block allocation or partial block zeroing would be required.  In that case
+ * we try again with the exclusive lock.
+ */
+static noinline ssize_t
+xfs_file_dio_write_unaligned(
+	struct xfs_inode	*ip,
+	struct kiocb		*iocb,
+	struct iov_iter		*from)
+{
+	size_t			isize = i_size_read(VFS_I(ip));
+	size_t			count = iov_iter_count(from);
+	int			iolock = XFS_IOLOCK_SHARED;
+	unsigned int		flags = IOMAP_DIO_OVERWRITE_ONLY;
+	ssize_t			ret;
+
+	/*
+	 * Extending writes need exclusivity because of the sub-block zeroing
+	 * that the DIO code always does for partial tail blocks beyond EOF, so
+	 * don't even bother trying the fast path in this case.
+	 */
+	if (iocb->ki_pos > isize || iocb->ki_pos + count >= isize) {
+retry_exclusive:
+		if (iocb->ki_flags & IOCB_NOWAIT)
 			return -EAGAIN;
-	} else {
-		xfs_ilock(ip, iolock);
+		iolock = XFS_IOLOCK_EXCL;
+		flags = IOMAP_DIO_FORCE_WAIT;
 	}
 
-	ret = xfs_file_aio_write_checks(iocb, from, &iolock);
+	ret = xfs_ilock_iocb(iocb, iolock);
 	if (ret)
-		goto out;
-	count = iov_iter_count(from);
+		return ret;
 
 	/*
-	 * If we are doing unaligned IO, we can't allow any other overlapping IO
-	 * in-flight at the same time or we risk data corruption. Wait for all
-	 * other IO to drain before we submit. If the IO is aligned, demote the
-	 * iolock if we had to take the exclusive lock in
-	 * xfs_file_aio_write_checks() for other reasons.
+	 * We can't properly handle unaligned direct I/O to reflink files yet,
+	 * as we can't unshare a partial block.
 	 */
-	if (unaligned_io) {
-		inode_dio_wait(inode);
-	} else if (iolock == XFS_IOLOCK_EXCL) {
-		xfs_ilock_demote(ip, XFS_IOLOCK_EXCL);
-		iolock = XFS_IOLOCK_SHARED;
+	if (xfs_is_cow_inode(ip)) {
+		trace_xfs_reflink_bounce_dio_write(iocb, from);
+		ret = -ENOTBLK;
+		goto out_unlock;
 	}
 
-	trace_xfs_file_direct_write(ip, count, iocb->ki_pos);
+	ret = xfs_file_write_checks(iocb, from, &iolock);
+	if (ret)
+		goto out_unlock;
+
 	/*
-	 * If unaligned, this is the only IO in-flight. Wait on it before we
-	 * release the iolock to prevent subsequent overlapping IO.
+	 * If we are doing exclusive unaligned I/O, this must be the only I/O
+	 * in-flight.  Otherwise we risk data corruption due to unwritten extent
+	 * conversions from the AIO end_io handler.  Wait for all other I/O to
+	 * drain first.
 	 */
+	if (flags & IOMAP_DIO_FORCE_WAIT)
+		inode_dio_wait(VFS_I(ip));
+
+	trace_xfs_file_direct_write(iocb, from);
 	ret = iomap_dio_rw(iocb, from, &xfs_direct_write_iomap_ops,
-			   &xfs_dio_write_ops,
-			   is_sync_kiocb(iocb) || unaligned_io);
-out:
-	xfs_iunlock(ip, iolock);
+			   &xfs_dio_write_ops, flags);
 
 	/*
-	 * No fallback to buffered IO after short writes for XFS, direct I/O
-	 * will either complete fully or return an error.
+	 * Retry unaligned I/O with exclusive blocking semantics if the DIO
+	 * layer rejected it for mapping or locking reasons. If we are doing
+	 * nonblocking user I/O, propagate the error.
 	 */
-	ASSERT(ret < 0 || ret == count);
+	if (ret == -EAGAIN && !(iocb->ki_flags & IOCB_NOWAIT)) {
+		ASSERT(flags & IOMAP_DIO_OVERWRITE_ONLY);
+		xfs_iunlock(ip, iolock);
+		goto retry_exclusive;
+	}
+
+out_unlock:
+	if (iolock)
+		xfs_iunlock(ip, iolock);
 	return ret;
 }
 
+static ssize_t
+xfs_file_dio_write(
+	struct kiocb		*iocb,
+	struct iov_iter		*from)
+{
+	struct xfs_inode	*ip = XFS_I(file_inode(iocb->ki_filp));
+	struct xfs_buftarg      *target = xfs_inode_buftarg(ip);
+	size_t			count = iov_iter_count(from);
+
+	/* direct I/O must be aligned to device logical sector size */
+	if ((iocb->ki_pos | count) & target->bt_logical_sectormask)
+		return -EINVAL;
+	if ((iocb->ki_pos | count) & ip->i_mount->m_blockmask)
+		return xfs_file_dio_write_unaligned(ip, iocb, from);
+	return xfs_file_dio_write_aligned(ip, iocb, from);
+}
+
 static noinline ssize_t
 xfs_file_dax_write(
 	struct kiocb		*iocb,
@@ -605,31 +651,26 @@ xfs_file_dax_write(
 	struct xfs_inode	*ip = XFS_I(inode);
 	int			iolock = XFS_IOLOCK_EXCL;
 	ssize_t			ret, error = 0;
-	size_t			count;
 	loff_t			pos;
 
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		if (!xfs_ilock_nowait(ip, iolock))
-			return -EAGAIN;
-	} else {
-		xfs_ilock(ip, iolock);
-	}
-
-	ret = xfs_file_aio_write_checks(iocb, from, &iolock);
+	ret = xfs_ilock_iocb(iocb, iolock);
+	if (ret)
+		return ret;
+	ret = xfs_file_write_checks(iocb, from, &iolock);
 	if (ret)
 		goto out;
 
 	pos = iocb->ki_pos;
-	count = iov_iter_count(from);
 
-	trace_xfs_file_dax_write(ip, count, pos);
+	trace_xfs_file_dax_write(iocb, from);
 	ret = dax_iomap_rw(iocb, from, &xfs_direct_write_iomap_ops);
 	if (ret > 0 && iocb->ki_pos > i_size_read(inode)) {
 		i_size_write(inode, iocb->ki_pos);
 		error = xfs_setfilesize(ip, pos, ret);
 	}
 out:
-	xfs_iunlock(ip, iolock);
+	if (iolock)
+		xfs_iunlock(ip, iolock);
 	if (error)
 		return error;
 
@@ -643,7 +684,7 @@ out:
 }
 
 STATIC ssize_t
-xfs_file_buffered_aio_write(
+xfs_file_buffered_write(
 	struct kiocb		*iocb,
 	struct iov_iter		*from)
 {
@@ -662,14 +703,14 @@ write_retry:
 	iolock = XFS_IOLOCK_EXCL;
 	xfs_ilock(ip, iolock);
 
-	ret = xfs_file_aio_write_checks(iocb, from, &iolock);
+	ret = xfs_file_write_checks(iocb, from, &iolock);
 	if (ret)
 		goto out;
 
 	/* We can write back this queue in page reclaim */
 	current->backing_dev_info = inode_to_bdi(inode);
 
-	trace_xfs_file_buffered_write(ip, iov_iter_count(from), iocb->ki_pos);
+	trace_xfs_file_buffered_write(iocb, from);
 	ret = iomap_file_buffered_write(iocb, from,
 			&xfs_buffered_write_iomap_ops);
 	if (likely(ret >= 0))
@@ -749,12 +790,12 @@ xfs_file_write_iter(
 		 * CoW.  In all other directio scenarios we do not
 		 * allow an operation to fall back to buffered mode.
 		 */
-		ret = xfs_file_dio_aio_write(iocb, from);
+		ret = xfs_file_dio_write(iocb, from);
 		if (ret != -ENOTBLK)
 			return ret;
 	}
 
-	return xfs_file_buffered_aio_write(iocb, from);
+	return xfs_file_buffered_write(iocb, from);
 }
 
 static void
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 7b9ff824e82d..ef76f775fabf 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -784,15 +784,28 @@ xfs_direct_write_iomap_begin(
 		goto allocate_blocks;
 
 	/*
-	 * NOWAIT IO needs to span the entire requested IO with a single map so
-	 * that we avoid partial IO failures due to the rest of the IO range not
-	 * covered by this map triggering an EAGAIN condition when it is
-	 * subsequently mapped and aborting the IO.
+	 * NOWAIT and OVERWRITE I/O needs to span the entire requested I/O with
+	 * a single map so that we avoid partial IO failures due to the rest of
+	 * the I/O range not covered by this map triggering an EAGAIN condition
+	 * when it is subsequently mapped and aborting the I/O.
 	 */
-	if ((flags & IOMAP_NOWAIT) &&
-	    !imap_spans_range(&imap, offset_fsb, end_fsb)) {
+	if (flags & (IOMAP_NOWAIT | IOMAP_OVERWRITE_ONLY)) {
 		error = -EAGAIN;
-		goto out_unlock;
+		if (!imap_spans_range(&imap, offset_fsb, end_fsb))
+			goto out_unlock;
+	}
+
+	/*
+	 * For overwrite only I/O, we cannot convert unwritten extents without
+	 * requiring sub-block zeroing.  This can only be done under an
+	 * exclusive IOLOCK, hence return -EAGAIN if this is not a written
+	 * extent to tell the caller to try again.
+	 */
+	if (flags & IOMAP_OVERWRITE_ONLY) {
+		error = -EAGAIN;
+		if (imap.br_state != XFS_EXT_NORM &&
+	            ((offset | length) & mp->m_blockmask))
+			goto out_unlock;
 	}
 
 	xfs_iunlock(ip, lockmode);
@@ -801,7 +814,7 @@ xfs_direct_write_iomap_begin(
 
 allocate_blocks:
 	error = -EAGAIN;
-	if (flags & IOMAP_NOWAIT)
+	if (flags & (IOMAP_NOWAIT | IOMAP_OVERWRITE_ONLY))
 		goto out_unlock;
 
 	/*
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 5a263ae3d4f0..0cfd65cd67c1 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -1287,8 +1287,8 @@ TRACE_EVENT(xfs_log_assign_tail_lsn,
 )
 
 DECLARE_EVENT_CLASS(xfs_file_class,
-	TP_PROTO(struct xfs_inode *ip, size_t count, loff_t offset),
-	TP_ARGS(ip, count, offset),
+	TP_PROTO(struct kiocb *iocb, struct iov_iter *iter),
+	TP_ARGS(iocb, iter),
 	TP_STRUCT__entry(
 		__field(dev_t, dev)
 		__field(xfs_ino_t, ino)
@@ -1297,11 +1297,11 @@ DECLARE_EVENT_CLASS(xfs_file_class,
 		__field(size_t, count)
 	),
 	TP_fast_assign(
-		__entry->dev = VFS_I(ip)->i_sb->s_dev;
-		__entry->ino = ip->i_ino;
-		__entry->size = ip->i_d.di_size;
-		__entry->offset = offset;
-		__entry->count = count;
+		__entry->dev = file_inode(iocb->ki_filp)->i_sb->s_dev;
+		__entry->ino = XFS_I(file_inode(iocb->ki_filp))->i_ino;
+		__entry->size = XFS_I(file_inode(iocb->ki_filp))->i_d.di_size;
+		__entry->offset = iocb->ki_pos;
+		__entry->count = iov_iter_count(iter);
 	),
 	TP_printk("dev %d:%d ino 0x%llx size 0x%llx offset 0x%llx count 0x%zx",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
@@ -1313,14 +1313,16 @@ DECLARE_EVENT_CLASS(xfs_file_class,
 
 #define DEFINE_RW_EVENT(name)		\
 DEFINE_EVENT(xfs_file_class, name,	\
-	TP_PROTO(struct xfs_inode *ip, size_t count, loff_t offset),	\
-	TP_ARGS(ip, count, offset))
+	TP_PROTO(struct kiocb *iocb, struct iov_iter *iter),		\
+	TP_ARGS(iocb, iter))
 DEFINE_RW_EVENT(xfs_file_buffered_read);
 DEFINE_RW_EVENT(xfs_file_direct_read);
 DEFINE_RW_EVENT(xfs_file_dax_read);
 DEFINE_RW_EVENT(xfs_file_buffered_write);
 DEFINE_RW_EVENT(xfs_file_direct_write);
 DEFINE_RW_EVENT(xfs_file_dax_write);
+DEFINE_RW_EVENT(xfs_reflink_bounce_dio_write);
+
 
 DECLARE_EVENT_CLASS(xfs_imap_class,
 	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
@@ -3294,8 +3296,6 @@ DEFINE_INODE_IREC_EVENT(xfs_reflink_cow_found);
 DEFINE_INODE_IREC_EVENT(xfs_reflink_cow_enospc);
 DEFINE_INODE_IREC_EVENT(xfs_reflink_convert_cow);
 
-DEFINE_SIMPLE_IO_EVENT(xfs_reflink_bounce_dio_write);
-
 DEFINE_SIMPLE_IO_EVENT(xfs_reflink_cancel_cow_range);
 DEFINE_SIMPLE_IO_EVENT(xfs_reflink_end_cow);
 DEFINE_INODE_IREC_EVENT(xfs_reflink_cow_remap);
diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c
index bec47f2d074b..0e7ab0bc00ae 100644
--- a/fs/zonefs/super.c
+++ b/fs/zonefs/super.c
@@ -780,7 +780,7 @@ static ssize_t zonefs_file_dio_write(struct kiocb *iocb, struct iov_iter *from)
 		ret = zonefs_file_dio_append(iocb, from);
 	else
 		ret = iomap_dio_rw(iocb, from, &zonefs_iomap_ops,
-				   &zonefs_write_dio_ops, sync);
+				   &zonefs_write_dio_ops, 0);
 	if (zi->i_ztype == ZONEFS_ZTYPE_SEQ &&
 	    (ret > 0 || ret == -EIOCBQUEUED)) {
 		if (ret > 0)
@@ -917,7 +917,7 @@ static ssize_t zonefs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 		}
 		file_accessed(iocb->ki_filp);
 		ret = iomap_dio_rw(iocb, to, &zonefs_iomap_ops,
-				   &zonefs_read_dio_ops, is_sync_kiocb(iocb));
+				   &zonefs_read_dio_ops, 0);
 	} else {
 		ret = generic_file_read_iter(iocb, to);
 		if (ret == -EIO)
diff --git a/include/linux/iomap.h b/include/linux/iomap.h
index 8ebb1fa6f3b7..d202fd2d0f91 100644
--- a/include/linux/iomap.h
+++ b/include/linux/iomap.h
@@ -123,6 +123,7 @@ struct iomap_page_ops {
 #define IOMAP_FAULT		(1 << 3) /* mapping for page fault */
 #define IOMAP_DIRECT		(1 << 4) /* direct I/O */
 #define IOMAP_NOWAIT		(1 << 5) /* do not block */
+#define IOMAP_OVERWRITE_ONLY	(1 << 6) /* only pure overwrites allowed */
 
 struct iomap_ops {
 	/*
@@ -257,12 +258,25 @@ struct iomap_dio_ops {
 			struct bio *bio, loff_t file_offset);
 };
 
+/*
+ * Wait for the I/O to complete in iomap_dio_rw even if the kiocb is not
+ * synchronous.
+ */
+#define IOMAP_DIO_FORCE_WAIT	(1 << 0)
+
+/*
+ * Do not allocate blocks or zero partial blocks, but instead fall back to
+ * the caller by returning -EAGAIN.  Used to optimize direct I/O writes that
+ * are not aligned to the file system block size.
+  */
+#define IOMAP_DIO_OVERWRITE_ONLY	(1 << 1)
+
 ssize_t iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 		const struct iomap_ops *ops, const struct iomap_dio_ops *dops,
-		bool wait_for_completion);
+		unsigned int dio_flags);
 struct iomap_dio *__iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
 		const struct iomap_ops *ops, const struct iomap_dio_ops *dops,
-		bool wait_for_completion);
+		unsigned int dio_flags);
 ssize_t iomap_dio_complete(struct iomap_dio *dio);
 int iomap_dio_iopoll(struct kiocb *kiocb, bool spin);