Thread (12 messages) 12 messages, 2 authors, 2025-01-08
STALE560d
Revisions (10)
  1. v1 [diff vs current]
  2. v2 [diff vs current]
  3. v3 current
  4. v6 [diff vs current]
  5. v7 [diff vs current]
  6. v8 [diff vs current]
  7. v9 [diff vs current]
  8. v10 [diff vs current]
  9. v11 [diff vs current]
  10. v12 [diff vs current]

[PATCH v3 1/7] iomap: Increase iomap_dio_zero() size limit

From: John Garry <john.g.garry@oracle.com>
Date: 2025-01-02 14:04:52
Also in: linux-fsdevel, lkml
Subsystem: filesystems (vfs and infrastructure), filesystems [iomap], the rest · Maintainers: Alexander Viro, Christian Brauner, Linus Torvalds

Currently iomap_dio_zero() is limited to using a single bio to write up to
64K.

To support atomic writes larger than the FS block size, it may be required
to pre-zero some extents larger than 64K.

To increase the limit, fill each bio up in a loop.

Signed-off-by: John Garry <john.g.garry@oracle.com>
---
 fs/iomap/direct-io.c | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
index b521eb15759e..23fdad16e6a8 100644
--- a/fs/iomap/direct-io.c
+++ b/fs/iomap/direct-io.c
@@ -240,27 +240,35 @@ void iomap_dio_bio_end_io(struct bio *bio)
 EXPORT_SYMBOL_GPL(iomap_dio_bio_end_io);
 
 static int iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio,
-		loff_t pos, unsigned len)
+		const loff_t pos, const unsigned len)
 {
 	struct inode *inode = file_inode(dio->iocb->ki_filp);
+	unsigned int remaining = len;
+	unsigned int nr_vecs;
 	struct bio *bio;
+	int i;
 
 	if (!len)
 		return 0;
-	/*
-	 * Max block size supported is 64k
-	 */
-	if (WARN_ON_ONCE(len > IOMAP_ZERO_PAGE_SIZE))
+
+	nr_vecs = DIV_ROUND_UP(len, IOMAP_ZERO_PAGE_SIZE);
+	if (WARN_ON_ONCE(nr_vecs > BIO_MAX_VECS))
 		return -EINVAL;
 
-	bio = iomap_dio_alloc_bio(iter, dio, 1, REQ_OP_WRITE | REQ_SYNC | REQ_IDLE);
+	bio = iomap_dio_alloc_bio(iter, dio, nr_vecs,
+			REQ_OP_WRITE | REQ_SYNC | REQ_IDLE);
 	fscrypt_set_bio_crypt_ctx(bio, inode, pos >> inode->i_blkbits,
 				  GFP_KERNEL);
 	bio->bi_iter.bi_sector = iomap_sector(&iter->iomap, pos);
 	bio->bi_private = dio;
 	bio->bi_end_io = iomap_dio_bio_end_io;
 
-	__bio_add_page(bio, zero_page, len, 0);
+	for (i = 0; i < nr_vecs; i++) {
+		__bio_add_page(bio, zero_page,
+			min(remaining, IOMAP_ZERO_PAGE_SIZE), 0);
+		remaining -= IOMAP_ZERO_PAGE_SIZE;
+	}
+
 	iomap_dio_submit_bio(iter, dio, bio, pos);
 	return 0;
 }
-- 
2.31.1
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help