From: Christoph Hellwig <hch@lst.de>
To: Christian Brauner <brauner@kernel.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>,
"Darrick J. Wong" <djwong@kernel.org>, Jan Kara <jack@suse.cz>,
Jens Axboe <axboe@kernel.dk>, Avi Kivity <avi@scylladb.com>,
Damien Le Moal <dlemoal@kernel.org>,
Naohiro Aota <naohiro.aota@wdc.com>,
Johannes Thumshirn <jth@kernel.org>,
linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
io-uring@vger.kernel.org
Subject: [PATCH 3/5] iomap: rework REQ_FUA selection
Date: Wed, 12 Nov 2025 08:21:27 +0100 [thread overview]
Message-ID: <20251112072214.844816-4-hch@lst.de> (raw)
In-Reply-To: <20251112072214.844816-1-hch@lst.de>
The way how iomap_dio_can_use_fua and the caller is structured is
a bit confusing, as the main guarding condition is hidden in the
helper, and the secondary conditions are split between caller and
callee.
Refactor the code, so that there is a main IOMAP_DIO_WRITE_THROUGH
guard in iomap_dio_bio_iter, which is directly tied to clearing it
when not supported, and a helper that just checks if the I/O is a
pure overwrite.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
fs/iomap/direct-io.c | 36 ++++++++++++++++++++----------------
1 file changed, 20 insertions(+), 16 deletions(-)
diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
index 765ab6dd6637..c4a883fa8ea5 100644
--- a/fs/iomap/direct-io.c
+++ b/fs/iomap/direct-io.c
@@ -288,20 +288,14 @@ static int iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio,
}
/*
- * Use a FUA write if we need datasync semantics and this is a pure data I/O
- * that doesn't require any metadata updates (including after I/O completion
- * such as unwritten extent conversion) and the underlying device either
- * doesn't have a volatile write cache or supports FUA.
- * This allows us to avoid cache flushes on I/O completion.
+ * Check if this mapping is a pure overwrite that does not need metadata updates
+ * at I/O completion time.
*/
-static inline bool iomap_dio_can_use_fua(const struct iomap *iomap,
- struct iomap_dio *dio)
+static inline bool iomap_dio_is_overwrite(const struct iomap *iomap)
{
- if (iomap->flags & (IOMAP_F_SHARED | IOMAP_F_DIRTY))
+ if (iomap->type != IOMAP_MAPPED)
return false;
- if (!(dio->flags & IOMAP_DIO_WRITE_THROUGH))
- return false;
- return !bdev_write_cache(iomap->bdev) || bdev_fua(iomap->bdev);
+ return !(iomap->flags & (IOMAP_F_NEW | IOMAP_F_SHARED));
}
static int iomap_dio_bio_iter(struct iomap_iter *iter, struct iomap_dio *dio)
@@ -355,12 +349,22 @@ static int iomap_dio_bio_iter(struct iomap_iter *iter, struct iomap_dio *dio)
if (iomap->flags & IOMAP_F_NEW)
need_zeroout = true;
- else if (iomap->type == IOMAP_MAPPED &&
- iomap_dio_can_use_fua(iomap, dio))
- bio_opf |= REQ_FUA;
- if (!(bio_opf & REQ_FUA))
- dio->flags &= ~IOMAP_DIO_WRITE_THROUGH;
+ /*
+ * Use a FUA write if we need datasync semantics and this is a
+ * pure overwrite that doesn't require any metadata updates.
+ *
+ * This allows us to avoid cache flushes on I/O completion.
+ */
+ if (dio->flags & IOMAP_DIO_WRITE_THROUGH) {
+ if (iomap_dio_is_overwrite(iomap) &&
+ !(iomap->flags & IOMAP_F_DIRTY) &&
+ (!bdev_write_cache(iomap->bdev) ||
+ bdev_fua(iomap->bdev)))
+ bio_opf |= REQ_FUA;
+ else
+ dio->flags &= ~IOMAP_DIO_WRITE_THROUGH;
+ }
} else {
bio_opf |= REQ_OP_READ;
}
--
2.47.3
next prev parent reply other threads:[~2025-11-12 7:22 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-12 7:21 enable iomap dio write completions from interrupt context Christoph Hellwig
2025-11-12 7:21 ` [PATCH 1/5] fs, iomap: remove IOCB_DIO_CALLER_COMP Christoph Hellwig
2025-11-12 19:59 ` Jan Kara
2025-11-13 0:00 ` Chaitanya Kulkarni
2025-11-12 7:21 ` [PATCH 2/5] iomap: always run error completions in user context Christoph Hellwig
2025-11-12 20:01 ` Jan Kara
2025-11-13 0:02 ` Chaitanya Kulkarni
2025-11-12 7:21 ` Christoph Hellwig [this message]
2025-11-12 20:07 ` [PATCH 3/5] iomap: rework REQ_FUA selection Jan Kara
2025-11-12 7:21 ` [PATCH 4/5] iomap: support write completions from interrupt context Christoph Hellwig
2025-11-12 20:25 ` Jan Kara
2025-11-13 6:50 ` Christoph Hellwig
2025-11-13 9:54 ` Jan Kara
2025-11-13 10:06 ` Christoph Hellwig
2025-11-13 12:06 ` Jan Kara
2025-11-13 12:35 ` Christoph Hellwig
2025-11-12 7:21 ` [PATCH 5/5] iomap: invert the polarity of IOMAP_DIO_INLINE_COMP Christoph Hellwig
2025-11-13 12:09 ` Jan Kara
2025-11-12 8:43 ` enable iomap dio write completions from interrupt context Damien Le Moal
2025-11-12 8:44 ` Christoph Hellwig
2025-11-12 8:46 ` Damien Le Moal
2025-11-13 9:58 ` Jan Kara
2025-11-13 10:05 ` Christoph Hellwig
2025-11-13 10:07 ` Damien Le Moal
2025-11-13 11:52 ` Jan Kara
-- strict thread matches above, loose matches on Subject: below --
2025-11-13 17:06 enable iomap dio write completions from interrupt context v2 Christoph Hellwig
2025-11-13 17:06 ` [PATCH 3/5] iomap: rework REQ_FUA selection Christoph Hellwig
2025-11-24 11:05 ` Jan Kara
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20251112072214.844816-4-hch@lst.de \
--to=hch@lst.de \
--cc=avi@scylladb.com \
--cc=axboe@kernel.dk \
--cc=brauner@kernel.org \
--cc=djwong@kernel.org \
--cc=dlemoal@kernel.org \
--cc=io-uring@vger.kernel.org \
--cc=jack@suse.cz \
--cc=jth@kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=naohiro.aota@wdc.com \
--cc=viro@zeniv.linux.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox