[PATCHv8,4/5] iomap: Allocate iof in ->write_begin() early

Commit Message

Ritesh Harjani (IBM) June 6, 2023, 11:43 a.m. UTC

We dont need to allocate an iof in ->write_begin() for writes where the
position and length completely overlap with the given folio.
Therefore, such cases are skipped.

Currently when the folio is uptodate, we only allocate iof at writeback
time (in iomap_writepage_map()). This is ok until now, but when we are
going to add support for per-block dirty state bitmap in iof, this
could cause some performance degradation. The reason is that if we don't
allocate iof during ->write_begin(), then we will never mark the
necessary dirty bits in ->write_end() call. And we will have to mark all
the bits as dirty at the writeback time, that could cause the same write
amplification and performance problems as it is now.

Reviewed-by: Darrick J. Wong <[e-mail]>
Signed-off-by: Ritesh Harjani (IBM) <[e-mail]>
---
 fs/iomap/buffered-io.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

Comments

Christoph Hellwig June 7, 2023, 6:53 a.m. UTC | #1

On Tue, Jun 06, 2023 at 05:13:51PM +0530, Ritesh Harjani (IBM) wrote:
> We dont need to allocate an iof in ->write_begin() for writes where the
> position and length completely overlap with the given folio.
> Therefore, such cases are skipped.
> 
> Currently when the folio is uptodate, we only allocate iof at writeback
> time (in iomap_writepage_map()). This is ok until now, but when we are
> going to add support for per-block dirty state bitmap in iof, this
> could cause some performance degradation. The reason is that if we don't
> allocate iof during ->write_begin(), then we will never mark the
> necessary dirty bits in ->write_end() call. And we will have to mark all
> the bits as dirty at the writeback time, that could cause the same write
> amplification and performance problems as it is now.
> 
> Reviewed-by: Darrick J. Wong <[e-mail]>
> Signed-off-by: Ritesh Harjani (IBM) <[e-mail]>

Looks good:

Reviewed-by: Christoph Hellwig <[e-mail]>

(and didn't I ack an earlier of this already?)

Patch

diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
index 89489aed49c0..2b72ca3ba37a 100644
--- a/fs/iomap/buffered-io.c
+++ b/fs/iomap/buffered-io.c
@@ -567,14 +567,23 @@  static int __iomap_write_begin(const struct iomap_iter *iter, loff_t pos,
 	size_t from = offset_in_folio(folio, pos), to = from + len;
 	size_t poff, plen;
 
-	if (folio_test_uptodate(folio))
+	/*
+	 * If the write completely overlaps the current folio, then
+	 * entire folio will be dirtied so there is no need for
+	 * per-block state tracking structures to be attached to this folio.
+	 */
+	if (pos <= folio_pos(folio) &&
+	    pos + len >= folio_pos(folio) + folio_size(folio))
 		return 0;
-	folio_clear_error(folio);
 
 	iof = iomap_iof_alloc(iter->inode, folio, iter->flags);
 	if ((iter->flags & IOMAP_NOWAIT) && !iof && nr_blocks > 1)
 		return -EAGAIN;
 
+	if (folio_test_uptodate(folio))
+		return 0;
+	folio_clear_error(folio);
+
 	do {
 		iomap_adjust_read_range(iter->inode, folio, &block_start,
 				block_end - block_start, &poff, &plen);