From: Chao Yu <chao@kernel.org>
To: Daniel Lee <chullee@google.com>, Jaegeuk Kim <jaegeuk@kernel.org>
Cc: chao@kernel.org, linux-f2fs-devel@lists.sourceforge.net,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 1/2] f2fs: Apply bio flags to direct I/O
Date: Mon, 16 Jun 2025 20:41:20 +0800 [thread overview]
Message-ID: <0f09a845-fbaf-4ddf-b684-a1182f85a9ff@kernel.org> (raw)
In-Reply-To: <20250615144235.1836469-2-chullee@google.com>
On 6/15/25 22:42, Daniel Lee wrote:
> Bio flags like REQ_PRIO, REQ_META, and REQ_FUA, determined by
> f2fs_io_flags(), were not being applied to direct I/O (DIO) writes.
> This meant that DIO writes would not respect filesystem-level hints
> (for REQ_META/FUA) or inode-level hints (like F2FS_IOPRIO_WRITE).
>
> This patch refactors f2fs_io_flags() to use a direct inode pointer
> instead of deriving it from a page. The function is then called from
> the DIO write path, ensuring that bio flags are handled consistently
> for both buffered and DIO writes.
>
> Signed-off-by: Daniel Lee <chullee@google.com>
> ---
> fs/f2fs/data.c | 10 +++++-----
> fs/f2fs/f2fs.h | 1 +
> fs/f2fs/file.c | 11 +++++++++++
> 3 files changed, 17 insertions(+), 5 deletions(-)
>
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index 31e892842625..71dde494b892 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -416,10 +416,9 @@ int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr)
> return 0;
> }
>
> -static blk_opf_t f2fs_io_flags(struct f2fs_io_info *fio)
> +blk_opf_t f2fs_io_flags(struct f2fs_io_info *fio, struct inode *inode)
> {
> unsigned int temp_mask = GENMASK(NR_TEMP_TYPE - 1, 0);
> - struct folio *fio_folio = page_folio(fio->page);
> unsigned int fua_flag, meta_flag, io_flag;
> blk_opf_t op_flags = 0;
>
> @@ -446,8 +445,8 @@ static blk_opf_t f2fs_io_flags(struct f2fs_io_info *fio)
> if (BIT(fio->temp) & fua_flag)
> op_flags |= REQ_FUA;
>
> - if (fio->type == DATA &&
> - F2FS_I(fio_folio->mapping->host)->ioprio_hint == F2FS_IOPRIO_WRITE)
> + if (inode && fio->type == DATA &&
> + F2FS_I(inode)->ioprio_hint == F2FS_IOPRIO_WRITE)
> op_flags |= REQ_PRIO;
>
> return op_flags;
> @@ -459,10 +458,11 @@ static struct bio *__bio_alloc(struct f2fs_io_info *fio, int npages)
> struct block_device *bdev;
> sector_t sector;
> struct bio *bio;
> + struct inode *inode = fio->page ? fio->page->mapping->host : NULL;
fio->page will always be true now? We can pass fio->page->mapping->host to f2fs_io_flags()
directly?
Thanks,
>
> bdev = f2fs_target_device(sbi, fio->new_blkaddr, §or);
> bio = bio_alloc_bioset(bdev, npages,
> - fio->op | fio->op_flags | f2fs_io_flags(fio),
> + fio->op | fio->op_flags | f2fs_io_flags(fio, inode),
> GFP_NOIO, &f2fs_bioset);
> bio->bi_iter.bi_sector = sector;
> if (is_read_io(fio->op)) {
> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> index 9333a22b9a01..3e02687c1b58 100644
> --- a/fs/f2fs/f2fs.h
> +++ b/fs/f2fs/f2fs.h
> @@ -3972,6 +3972,7 @@ void f2fs_submit_page_write(struct f2fs_io_info *fio);
> struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
> block_t blk_addr, sector_t *sector);
> int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
> +blk_opf_t f2fs_io_flags(struct f2fs_io_info *fio, struct inode *inode);
> void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
> void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
> int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
> diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> index 696131e655ed..3eb40d7bf602 100644
> --- a/fs/f2fs/file.c
> +++ b/fs/f2fs/file.c
> @@ -5015,6 +5015,17 @@ static void f2fs_dio_write_submit_io(const struct iomap_iter *iter,
> enum log_type type = f2fs_rw_hint_to_seg_type(sbi, inode->i_write_hint);
> enum temp_type temp = f2fs_get_segment_temp(sbi, type);
>
> + /* if fadvise set to hot, override the temperature */
> + struct f2fs_io_info fio = {
> + .sbi = sbi,
> + .type = DATA,
> + .op = REQ_OP_WRITE,
> + .temp = file_is_hot(inode) ? HOT : temp,
> + .op_flags = bio->bi_opf,
> + .page = NULL,
> + };
> + bio->bi_opf |= f2fs_io_flags(&fio, inode);
> +
> bio->bi_write_hint = f2fs_io_type_to_rw_hint(sbi, DATA, temp);
> submit_bio(bio);
> }
next prev parent reply other threads:[~2025-06-16 12:41 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-06-15 14:42 [PATCH v2 0/2] f2fs: Fix DIO flags and add ioprio hint Daniel Lee
2025-06-15 14:42 ` [PATCH v2 1/2] f2fs: Apply bio flags to direct I/O Daniel Lee
2025-06-16 12:41 ` Chao Yu [this message]
2025-06-30 23:02 ` Daniel Lee
2025-06-15 14:42 ` [PATCH v2 2/2] f2fs: use ioprio hint for hot and pinned files Daniel Lee
2025-06-16 12:50 ` Chao Yu
2025-06-30 22:56 ` Daniel Lee
2025-07-01 3:39 ` Chao Yu
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=0f09a845-fbaf-4ddf-b684-a1182f85a9ff@kernel.org \
--to=chao@kernel.org \
--cc=chullee@google.com \
--cc=jaegeuk@kernel.org \
--cc=linux-f2fs-devel@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.org \
/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
all inboxes | Powered by JetHome®