From: Christian Schoenebeck <linux_oss@crudebyte.com>
To: Dominique Martinet <asmadeus@codewreck.org>, Tingmao Wang <m@maowtm.org>
Cc: Tingmao Wang <m@maowtm.org>,
Eric Van Hensbergen <ericvh@kernel.org>,
Latchesar Ionkov <lucho@ionkov.net>,
v9fs@lists.linux.dev
Subject: Re: [PATCH v2] fs/9p: Don't open remote file with APPEND mode when writeback cache is used
Date: Mon, 10 Nov 2025 15:22:48 +0100 [thread overview]
Message-ID: <5562649.tBn8OvxF64@silver> (raw)
In-Reply-To: <20251102235631.8724-1-m@maowtm.org>
On Monday, November 3, 2025 12:56:30 AM CET Tingmao Wang wrote:
> When page cache is used, writebacks are done on a page granularity, and it
> is expected that the underlying filesystem (such as v9fs) should respect
> the write position. However, currently v9fs will passthrough O_APPEND to
> the server even on cached mode. This causes data corruption if a sync or
> fstat gets between two writes to the same file.
>
> This patch removes the APPEND flag from the open request we send to the
> server when writeback caching is involved. I believe keeping server-side
> APPEND is probably fine for uncached mode (even if two fds are opened, one
> without O_APPEND and one with it, this should still be fine since they
> would use separate fid for the writes).
>
> Signed-off-by: Tingmao Wang <m@maowtm.org>
> ---
>
> I haven't done a bisect to figure out if this regression was introduced by
> a change or was this behaviour always present - 6.14 has the same problem
> and 6.13 did not compile for me due to some problem with bool / true /
> false being a keyword in c23, and it could not compile with c17 either.
>
> fs/9p/vfs_file.c | 11 ++++++++---
> fs/9p/vfs_inode.c | 3 +--
> fs/9p/vfs_inode_dotl.c | 2 +-
> 3 files changed, 10 insertions(+), 6 deletions(-)
>
> diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
> index 612a230bc012..6f3880208587 100644
> --- a/fs/9p/vfs_file.c
> +++ b/fs/9p/vfs_file.c
> @@ -43,14 +43,18 @@ int v9fs_file_open(struct inode *inode, struct file
> *file) struct v9fs_session_info *v9ses;
> struct p9_fid *fid;
> int omode;
> + int o_append;
>
> p9_debug(P9_DEBUG_VFS, "inode: %p file: %p\n", inode, file);
> v9ses = v9fs_inode2v9ses(inode);
> - if (v9fs_proto_dotl(v9ses))
> + if (v9fs_proto_dotl(v9ses)) {
> omode = v9fs_open_to_dotl_flags(file->f_flags);
> - else
> + o_append = P9_DOTL_APPEND;
> + } else {
> omode = v9fs_uflags2omode(file->f_flags,
> v9fs_proto_dotu(v9ses));
> + o_append = P9_OAPPEND;
> + }
> fid = file->private_data;
> if (!fid) {
> fid = v9fs_fid_clone(file_dentry(file));
> @@ -58,9 +62,10 @@ int v9fs_file_open(struct inode *inode, struct file
> *file) return PTR_ERR(fid);
>
> if ((v9ses->cache & CACHE_WRITEBACK) && (omode & P9_OWRITE)) {
> - int writeback_omode = (omode & ~P9_OWRITE) | P9_ORDWR;
> + int writeback_omode = (omode & ~(P9_OWRITE | o_append)) | P9_ORDWR;
I guess changes in this function could be simplified by either pulling
O_APPEND at the very beginning right after v9ses = v9fs_inode2v9ses(inode); or
by simply pulling both P9_OAPPEND and P9_DOTL_APPEND here, but it's doing the
job so:
Reviewed-by: Christian Schoenebeck <linux_oss@crudebyte.com>
/Christian
>
> p9_debug(P9_DEBUG_CACHE, "write-only file with writeback enabled, try
> opening O_RDWR\n"); +
> err = p9_client_open(fid, writeback_omode);
> if (err < 0) {
> p9_debug(P9_DEBUG_CACHE, "could not open O_RDWR, disabling caches\n");
> diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
> index 8666c9c62258..97abe65bf7c1 100644
> --- a/fs/9p/vfs_inode.c
> +++ b/fs/9p/vfs_inode.c
> @@ -786,7 +786,7 @@ v9fs_vfs_atomic_open(struct inode *dir, struct dentry
> *dentry, p9_omode = v9fs_uflags2omode(flags, v9fs_proto_dotu(v9ses));
>
> if ((v9ses->cache & CACHE_WRITEBACK) && (p9_omode & P9_OWRITE)) {
> - p9_omode = (p9_omode & ~P9_OWRITE) | P9_ORDWR;
> + p9_omode = (p9_omode & ~(P9_OWRITE | P9_OAPPEND)) | P9_ORDWR;
> p9_debug(P9_DEBUG_CACHE,
> "write-only file with writeback enabled, creating w/ O_RDWR\n");
> }
> @@ -1393,4 +1393,3 @@ static const struct inode_operations
> v9fs_symlink_inode_operations = { .getattr = v9fs_vfs_getattr,
> .setattr = v9fs_vfs_setattr,
> };
> -
> diff --git a/fs/9p/vfs_inode_dotl.c b/fs/9p/vfs_inode_dotl.c
> index 1661a25f2772..643e759eacb2 100644
> --- a/fs/9p/vfs_inode_dotl.c
> +++ b/fs/9p/vfs_inode_dotl.c
> @@ -282,7 +282,7 @@ v9fs_vfs_atomic_open_dotl(struct inode *dir, struct
> dentry *dentry, }
>
> if ((v9ses->cache & CACHE_WRITEBACK) && (p9_omode & P9_OWRITE)) {
> - p9_omode = (p9_omode & ~P9_OWRITE) | P9_ORDWR;
> + p9_omode = (p9_omode & ~(P9_OWRITE | P9_DOTL_APPEND)) | P9_ORDWR;
> p9_debug(P9_DEBUG_CACHE,
> "write-only file with writeback enabled, creating w/ O_RDWR\n");
> }
next prev parent reply other threads:[~2025-11-10 14:22 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-02 20:24 [PATCH 0/1] fs/9p: Do not open remote file with APPEND mode when writeback cache is used Tingmao Wang
2025-11-02 20:24 ` [PATCH 1/1] " Tingmao Wang
2025-11-02 23:07 ` Dominique Martinet
2025-11-02 23:56 ` [PATCH v2] fs/9p: Don't " Tingmao Wang
2025-11-03 7:34 ` Dominique Martinet
2025-11-10 13:25 ` Christian Schoenebeck
2025-11-10 14:22 ` Christian Schoenebeck [this message]
2025-11-02 23:58 ` [PATCH 1/1] fs/9p: Do not " Tingmao Wang
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=5562649.tBn8OvxF64@silver \
--to=linux_oss@crudebyte.com \
--cc=asmadeus@codewreck.org \
--cc=ericvh@kernel.org \
--cc=lucho@ionkov.net \
--cc=m@maowtm.org \
--cc=v9fs@lists.linux.dev \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.