From: koraynilay <koray.fra@gmail.com>
To: Namjae Jeon <linkinjeon@kernel.org>,
Hyunchul Lee <hyc.lee@gmail.com>,
Alexander Viro <viro@zeniv.linux.org.uk>,
Christian Brauner <brauner@kernel.org>
Cc: "Chuck Lever" <cel@kernel.org>,
"Jeff Layton" <jlayton@kernel.org>, "Jan Kara" <jack@suse.cz>,
"Darrick J . Wong" <djwong@kernel.org>,
"Andrey Albershteyn" <aalbersh@kernel.org>,
"Amir Goldstein" <amir73il@gmail.com>,
"Pali Rohár" <pali@kernel.org>,
"Lauri Gates" <lauri.gates@gmail.com>,
"Christoph Hellwig" <hch@infradead.org>,
"Dave Chinner" <dgc@kernel.org>,
linux-btrfs@vger.kernel.org, linux-xfs@vger.kernel.org,
linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
ntfs@lists.linux.dev, linux-nfs@vger.kernel.org,
linux-api@vger.kernel.org, koraynilay <koray.fra@gmail.com>
Subject: [PATCH 2/3] vfs: add xflags mask to fileattr API
Date: Sun, 22 Nov 2026 01:08:55 +0100 [thread overview]
Message-ID: <20261122000856.536902-3-koray.fra@gmail.com> (raw)
In-Reply-To: <20261122000856.536902-1-koray.fra@gmail.com>
Add fa_xflags_rw_mask and fa_xflags_ro_mask to struct file_attr and
equivalent masks to struct file_kattr.
Also add has_masks selector to struct file_kattr so we can know when a
user of this API (userspace or filesystem) supports them.
This lets filesystems advertise their support for specific xflags, like
statx.stx_attributes_mask does, and userspace can specify only which
xflags it wants to change without needing a RMW cycle:
- on GET: each filesystem that wants to support this new API fills the
masks according to its supported and read-only xflags, then:
* for boolean xflags: it fills fsx_xflags with the current on-disk value
* for value xflags: it sets the xflag bit to 1 in fsx_xflags and the
xflag value in the struct field;
- on SET: userspace fills the rw_mask with the bits of the xflags it
wants to change, then:
* for boolean xflags: it provides the xflag value as a bit in fa_xflags
* for value xflags: it sets the xflag bit to 1 in fa_xflags and then
provides the value in the struct field; if the xflag bit is set to 0
instead, this informs the kernel to unset/clear that value.
Since fa_xflags_rw_mask means "I want to change the xflags specified in
here", if userspace specifies fa_xflags_rw_mask = 0, only check for
permissions and fileattr support without checking any xflags.
Behaviour of older programs that already use file_[gs]etattr() is
preserved by checking against FILE_ATTR_SIZE_VER*.
Also add the missing FS_XFLAG_PROJID xflag, to allow userspace to set
the projid using file_[gs]etattr() by leveraging the new masks.
Finally, in order to prevent the legacy FS_IOC_FSSETXATTR from accepting
FS_XFLAG_PROJID, add a FS_FILE_ATTR_XFLAGS_MASK to support new xflags
and use it to validate/clamp both the _xflags fields and the masks.
Assisted-by: Gemini:3.8-flash-high antigravity-cli-1.5.11
Signed-off-by: koraynilay <koray.fra@gmail.com>
---
fs/file_attr.c | 128 ++++++++++++++++++++++++++++++++++++---
include/linux/fileattr.h | 7 +++
include/uapi/linux/fs.h | 7 ++-
3 files changed, 131 insertions(+), 11 deletions(-)
diff --git a/fs/file_attr.c b/fs/file_attr.c
index 40d6518f1d8e..0e16c23368b2 100644
--- a/fs/file_attr.c
+++ b/fs/file_attr.c
@@ -102,13 +102,20 @@ EXPORT_SYMBOL(vfs_fileattr_get);
static void fileattr_to_file_attr(const struct file_kattr *fa,
struct file_attr *fattr)
{
- __u32 mask = FS_XFLAGS_MASK;
-
memset(fattr, 0, sizeof(struct file_attr));
- fattr->fa_xflags = fa->fsx_xflags & mask;
+ /* Pass the masks to userspace if underlying fs uses them. */
+ if (fa->has_masks) {
+ fattr->fa_xflags_rw_mask = fa->fsx_xflags_rw_mask &
+ FS_FILE_ATTR_XFLAGS_MASK;
+ fattr->fa_xflags_ro_mask = fa->fsx_xflags_ro_mask &
+ FS_FILE_ATTR_XFLAGS_MASK;
+ }
+ fattr->fa_xflags = fa->fsx_xflags & FS_FILE_ATTR_XFLAGS_MASK;
fattr->fa_extsize = fa->fsx_extsize;
fattr->fa_nextents = fa->fsx_nextents;
fattr->fa_projid = fa->fsx_projid;
+ if (fa->fsx_projid)
+ fattr->fa_xflags |= FS_XFLAG_PROJID;
fattr->fa_cowextsize = fa->fsx_cowextsize;
}
@@ -141,12 +148,30 @@ EXPORT_SYMBOL(copy_fsxattr_to_user);
static int file_attr_to_fileattr(const struct file_attr *fattr,
struct file_kattr *fa)
{
- __u64 mask = FS_XFLAGS_MASK;
+ u64 rw_mask = fa->has_masks ? fattr->fa_xflags_rw_mask : FS_XFLAGS_MASK;
- if (fattr->fa_xflags & ~mask)
+ /*
+ * No need to check for has_masks here, as fa_xflags_rw_mask is 0 in
+ * VER0 mode.
+ */
+ if (fattr->fa_xflags_rw_mask & ~FS_FILE_ATTR_XFLAGS_MASK)
return -EINVAL;
- fileattr_fill_xflags(fa, fattr->fa_xflags & ~FS_XFLAG_RDONLY_MASK);
+ if (fattr->fa_xflags & ~rw_mask)
+ return -EINVAL;
+
+ /*
+ * Filter with ~FS_XFLAG_RDONLY_MASK only when in VER0 mode, so
+ * filtering with the actual filesystem read-only xflags can be done in
+ * vfs_fileattr_set() once we have the fsx_xflags_ro_mask from the fs.
+ */
+ if (fa->has_masks) {
+ fileattr_fill_xflags(fa, fattr->fa_xflags);
+ fa->fsx_xflags_rw_mask = fattr->fa_xflags_rw_mask;
+ } else {
+ fileattr_fill_xflags(fa,
+ fattr->fa_xflags & ~FS_XFLAG_RDONLY_MASK);
+ }
fa->fsx_extsize = fattr->fa_extsize;
fa->fsx_projid = fattr->fa_projid;
fa->fsx_cowextsize = fattr->fa_cowextsize;
@@ -187,6 +212,14 @@ static int fileattr_set_prepare(struct inode *inode,
{
int err;
+ /*
+ * If the fs supports the fsx_xflags_{rw,ro}_mask API, fail when the
+ * requested rw flags are not in the supported ones for that fs
+ */
+ if (old_ma->has_masks &&
+ (fa->fsx_xflags_rw_mask & ~old_ma->fsx_xflags_rw_mask))
+ return -EOPNOTSUPP;
+
/*
* The IMMUTABLE and APPEND_ONLY flags can only be changed by
* the relevant capability.
@@ -211,6 +244,11 @@ static int fileattr_set_prepare(struct inode *inode,
FS_XFLAG_PROJINHERIT)
return -EINVAL;
} else {
+ /* If user wants to unset projid, set it to 0 */
+ if ((fa->fsx_xflags_rw_mask & FS_XFLAG_PROJID) &&
+ ((fa->fsx_xflags & FS_XFLAG_PROJID) == 0))
+ fa->fsx_projid = 0;
+
/*
* Caller is allowed to change the project ID. If it is being
* changed, make sure that the new value is valid.
@@ -250,6 +288,8 @@ static int fileattr_set_prepare(struct inode *inode,
fa->fsx_xflags &= ~(FS_XFLAG_EXTSIZE | FS_XFLAG_EXTSZINHERIT);
if (fa->fsx_cowextsize == 0)
fa->fsx_xflags &= ~FS_XFLAG_COWEXTSIZE;
+ if (fa->fsx_projid == 0)
+ fa->fsx_xflags &= ~FS_XFLAG_PROJID;
return 0;
}
@@ -275,6 +315,8 @@ int vfs_fileattr_set(const struct mnt_idmap *idmap, struct dentry *dentry,
{
struct inode *inode = d_inode(dentry);
struct file_kattr old_ma = {};
+ u32 rw_user_mask = fa->fsx_xflags_rw_mask;
+ u32 rw_fs_mask = 0, ro_fs_mask = 0;
int err;
if (!inode->i_op->fileattr_set)
@@ -288,15 +330,53 @@ int vfs_fileattr_set(const struct mnt_idmap *idmap, struct dentry *dentry,
if (err)
goto out;
+ if (old_ma.has_masks) {
+ old_ma.fsx_xflags_rw_mask &= FS_FILE_ATTR_XFLAGS_MASK;
+ old_ma.fsx_xflags_ro_mask &= FS_FILE_ATTR_XFLAGS_MASK;
+ rw_fs_mask = old_ma.fsx_xflags_rw_mask;
+ ro_fs_mask = old_ma.fsx_xflags_ro_mask;
+ }
+ old_ma.fsx_xflags &= FS_FILE_ATTR_XFLAGS_MASK;
+
/* initialize missing bits from old_ma */
if (fa->flags_valid) {
+ /* FS_IOC_SETFLAGS */
+ fa->fsx_xflags_rw_mask = 0;
fa->fsx_xflags |= old_ma.fsx_xflags & ~FS_XFLAG_COMMON;
fa->fsx_extsize = old_ma.fsx_extsize;
fa->fsx_nextents = old_ma.fsx_nextents;
fa->fsx_projid = old_ma.fsx_projid;
fa->fsx_cowextsize = old_ma.fsx_cowextsize;
} else {
+ /*
+ * Preserve flags not being set when using FS_IOC_FSSETXATTR or
+ * file_setattr VER0+.
+ */
fa->flags |= old_ma.flags & ~FS_COMMON_FL;
+
+ /* file_setattr VER1+ */
+ if (fa->has_masks) {
+ /* Preserve old xflags that aren't being set. */
+ fa->fsx_xflags |= old_ma.fsx_xflags & ~rw_user_mask;
+
+ /*
+ * Filter fsx_xflags by removing xflags that are:
+ * - filesystem-reported read only
+ * - globally read only
+ * but keeping xflags that are:
+ * - filesystem-reported rw, globally ro
+ */
+ ro_fs_mask |= FS_XFLAG_RDONLY_MASK & ~rw_fs_mask;
+ fileattr_fill_xflags(fa, fa->fsx_xflags & ~ro_fs_mask);
+
+ if (!(rw_user_mask &
+ (FS_XFLAG_EXTSIZE | FS_XFLAG_EXTSZINHERIT)))
+ fa->fsx_extsize = old_ma.fsx_extsize;
+ if (!(rw_user_mask & FS_XFLAG_PROJID))
+ fa->fsx_projid = old_ma.fsx_projid;
+ if (!(rw_user_mask & FS_XFLAG_COWEXTSIZE))
+ fa->fsx_cowextsize = old_ma.fsx_cowextsize;
+ }
}
err = fileattr_set_prepare(inode, &old_ma, fa);
@@ -424,6 +504,14 @@ SYSCALL_DEFINE5(file_getattr, int, dfd, const char __user *, filename,
return error;
fileattr_to_file_attr(&fa, &fattr);
+ /*
+ * Remove xflags added in or after VER1 so userspace programs compiled
+ * against VER0 don't try to write xflags they don't know about (since
+ * they do RMW).
+ */
+ if (usize < FILE_ATTR_SIZE_VER1)
+ fattr.fa_xflags &= FS_XFLAGS_MASK;
+
error = copy_struct_to_user(ufattr, usize, &fattr,
sizeof(struct file_attr), NULL);
@@ -460,10 +548,6 @@ SYSCALL_DEFINE5(file_setattr, int, dfd, const char __user *, filename,
if (error)
return error;
- error = file_attr_to_fileattr(&fattr, &fa);
- if (error)
- return error;
-
CLASS(filename_maybe_null, name)(filename, at_flags);
if (!name && dfd >= 0) {
CLASS(fd, f)(dfd);
@@ -479,6 +563,30 @@ SYSCALL_DEFINE5(file_setattr, int, dfd, const char __user *, filename,
return error;
}
+ /*
+ * Userspace said "I don't want to change any flag", so be it I guess.
+ * Still check for fileattr support and permissions, also in case this
+ * is just a probe.
+ */
+ if (usize >= FILE_ATTR_SIZE_VER1 && fattr.fa_xflags_rw_mask == 0) {
+ struct inode *inode = d_inode(filepath.dentry);
+
+ if (!inode->i_op->fileattr_set)
+ return -EOPNOTSUPP;
+
+ if (!inode_owner_or_capable(mnt_idmap(filepath.mnt), inode))
+ return -EPERM;
+
+ return error;
+ }
+
+ if (usize >= FILE_ATTR_SIZE_VER1)
+ fa.has_masks = true;
+
+ error = file_attr_to_fileattr(&fattr, &fa);
+ if (error)
+ return error;
+
error = mnt_want_write(filepath.mnt);
if (!error) {
error = vfs_fileattr_set(mnt_idmap(filepath.mnt),
diff --git a/include/linux/fileattr.h b/include/linux/fileattr.h
index c8f69dde5dd3..4faa40febf11 100644
--- a/include/linux/fileattr.h
+++ b/include/linux/fileattr.h
@@ -35,6 +35,10 @@
(FS_XFLAG_COMMON | FS_XFLAG_RDONLY_MASK | FS_XFLAG_VALUES_MASK | \
FS_XFLAG_DIRONLY_MASK | FS_XFLAG_MISC_MASK)
+/* Flags only for fa_ fields */
+#define FS_FILE_ATTR_XFLAGS_MASK \
+ (FS_XFLAGS_MASK | FS_XFLAG_PROJID)
+
/*
* Merged interface for miscellaneous file attributes. 'flags' originates from
* ext* and 'fsx_flags' from xfs. There's some overlap between the two, which
@@ -45,6 +49,8 @@ struct file_kattr {
u32 flags; /* flags (FS_IOC_GETFLAGS/FS_IOC_SETFLAGS) */
/* struct fsxattr: */
u32 fsx_xflags; /* xflags field value (get/set) */
+ u32 fsx_xflags_rw_mask; /* supported writeable xflags (get) / flags to set (set) */
+ u32 fsx_xflags_ro_mask; /* supported read-only xflags (get) */
u32 fsx_extsize; /* extsize field value (get/set)*/
u32 fsx_nextents; /* nextents field value (get) */
u32 fsx_projid; /* project identifier (get/set) */
@@ -52,6 +58,7 @@ struct file_kattr {
/* selectors: */
bool flags_valid:1;
bool fsx_valid:1;
+ bool has_masks:1;
};
int copy_fsxattr_to_user(const struct file_kattr *fa, struct fsxattr __user *ufa);
diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
index a13c7b32a1bd..e40806b3a69c 100644
--- a/include/uapi/linux/fs.h
+++ b/include/uapi/linux/fs.h
@@ -238,10 +238,13 @@ struct file_attr {
__u32 fa_nextents; /* nextents field value (get) */
__u32 fa_projid; /* project identifier (get/set) */
__u32 fa_cowextsize; /* CoW extsize field value (get/set) */
+ __u32 fa_xflags_rw_mask;/* supported writeable xflags (get) / flags to set (set) */
+ __u32 fa_xflags_ro_mask;/* supported read-only xflags (get) */
};
#define FILE_ATTR_SIZE_VER0 24
-#define FILE_ATTR_SIZE_LATEST FILE_ATTR_SIZE_VER0
+#define FILE_ATTR_SIZE_VER1 32
+#define FILE_ATTR_SIZE_LATEST FILE_ATTR_SIZE_VER1
/*
* Flags for the fsx_xflags field
@@ -270,6 +273,8 @@ struct file_attr {
*/
#define FS_XFLAG_CASEFOLD 0x00040000 /* case-insensitive lookups */
#define FS_XFLAG_CASENONPRESERVING 0x00080000 /* case not preserved */
+
+#define FS_XFLAG_PROJID 0x00100000 /* project ID */
#define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
/* the read-only stuff doesn't really belong here, but any other place is
--
2.56.0
next prev parent reply other threads:[~2026-11-22 0:09 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-22 0:08 [PATCH 0/3] Add f*_xflags_{rw,ro}_mask flags to the " koraynilay
2026-11-22 0:08 ` [PATCH 1/3] vfs: use goto out on error in vfs_fileattr_set() koraynilay
2026-11-22 0:08 ` koraynilay [this message]
2026-11-22 0:08 ` [PATCH 3/3] ntfs: switch to the new xflags fileattr API koraynilay
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=20261122000856.536902-3-koray.fra@gmail.com \
--to=koray.fra@gmail.com \
--cc=aalbersh@kernel.org \
--cc=amir73il@gmail.com \
--cc=brauner@kernel.org \
--cc=cel@kernel.org \
--cc=dgc@kernel.org \
--cc=djwong@kernel.org \
--cc=hch@infradead.org \
--cc=hyc.lee@gmail.com \
--cc=jack@suse.cz \
--cc=jlayton@kernel.org \
--cc=lauri.gates@gmail.com \
--cc=linkinjeon@kernel.org \
--cc=linux-api@vger.kernel.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=ntfs@lists.linux.dev \
--cc=pali@kernel.org \
--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
all inboxes | Powered by JetHome®