From: Suyuan Wang via B4 Relay <devnull+shyckless.gmail.com@kernel.org>
To: ebiggers@kernel.org
Cc: tytso@mit.edu, jaegeuk@kernel.org, linux-kernel@vger.kernel.org,
linux-fscrypt@vger.kernel.org
Subject: [PATCH v2] fscrypt: Don't set a policy on immutable or append-only directories
Date: Fri, 20 Nov 2026 04:26:54 +0000 [thread overview]
Message-ID: <20261120-fscrypt-v2-1-fb752135ea32@gmail.com> (raw)
From: Suyuan Wang <stevewong@hust.edu.cn>
FS_IOC_SET_ENCRYPTION_POLICY stores an encryption context in the
directory (an xattr on ext4 and f2fs) and marks the inode as encrypted.
fscrypt_ioctl_set_policy() checks inode_owner_or_capable(), but not the
immutable or append-only inode flags, so the ioctl also succeeds on a
directory with one of those flags set. Setting any other xattr on the
same directory fails with -EPERM in may_write_xattr(), and
FS_IOC_ENABLE_VERITY refuses such files as well.
For example, on ext4 with the encrypt feature:
mkdir d && chattr +i d
setfattr -n user.x -v 1 d -> Operation not permitted
FS_IOC_SET_ENCRYPTION_POLICY on d -> succeeds, d now has a policy
Return -EPERM for an immutable or append-only directory before setting
the policy, as may_write_xattr() does. Calls that find the same policy
already set still return 0, since they don't change the inode. Update
the documented EPERM cases to match.
Found with a static analysis tool and checked by hand.
Fixes: 9bd8212f981e ("ext4 crypto: add encryption policy and password salt support")
Fixes: f424f664f0e8 ("f2fs crypto: add encryption policy and password salt support")
Cc: stable@vger.kernel.org
Co-developed-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Suyuan Wang <stevewong@hust.edu.cn>
---
Changes in v2:
- Point the Fixes tags at the ext4 and f2fs commits that added the
ioctls, and ask for a stable backport (Eric).
- Document the new EPERM case in fscrypt.rst (Eric).
Tested in QEMU (x86_64, ext4 with -O encrypt, v1 policy). Before the
patch the ioctl succeeds on both immutable and append-only empty
directories; with it both get -EPERM, and an ordinary empty directory
still gets its policy.
v1: https://lore.kernel.org/all/20261119-fscrypt-v1-1-c07e1dd93644@gmail.com/
---
Documentation/filesystems/fscrypt.rst | 2 +-
fs/crypto/policy.c | 2 ++
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/Documentation/filesystems/fscrypt.rst b/Documentation/filesystems/fscrypt.rst
index 6e2e511a70bb..188295f09b21 100644
--- a/Documentation/filesystems/fscrypt.rst
+++ b/Documentation/filesystems/fscrypt.rst
@@ -736,7 +736,7 @@ FS_IOC_SET_ENCRYPTION_POLICY can fail with the following errors:
feature flag enabled using ``tune2fs -O encrypt`` or ``mkfs.ext4 -O
encrypt``.)
- ``EPERM``: this directory may not be encrypted, e.g. because it is
- the root directory of an ext4 filesystem
+ the root directory of an ext4 filesystem or is immutable or append-only
- ``EROFS``: the filesystem is readonly
Getting an encryption policy
diff --git a/fs/crypto/policy.c b/fs/crypto/policy.c
index 6dd510f93e6d..86a8254bc3a8 100644
--- a/fs/crypto/policy.c
+++ b/fs/crypto/policy.c
@@ -553,6 +553,8 @@ int fscrypt_ioctl_set_policy(struct file *filp, const void __user *arg)
ret = -ENOTDIR;
else if (IS_DEADDIR(inode))
ret = -ENOENT;
+ else if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+ ret = -EPERM;
else if (!inode->i_sb->s_cop->empty_dir(inode))
ret = -ENOTEMPTY;
else
WARNING: multiple messages have this Message-ID (diff)
From: Suyuan Wang <shyckless@gmail.com>
To: ebiggers@kernel.org
Cc: tytso@mit.edu, jaegeuk@kernel.org, linux-kernel@vger.kernel.org,
linux-fscrypt@vger.kernel.org
Subject: [PATCH v2] fscrypt: Don't set a policy on immutable or append-only directories
Date: Fri, 20 Nov 2026 04:26:54 +0000 [thread overview]
Message-ID: <20261120-fscrypt-v2-1-fb752135ea32@gmail.com> (raw)
From: Suyuan Wang <stevewong@hust.edu.cn>
FS_IOC_SET_ENCRYPTION_POLICY stores an encryption context in the
directory (an xattr on ext4 and f2fs) and marks the inode as encrypted.
fscrypt_ioctl_set_policy() checks inode_owner_or_capable(), but not the
immutable or append-only inode flags, so the ioctl also succeeds on a
directory with one of those flags set. Setting any other xattr on the
same directory fails with -EPERM in may_write_xattr(), and
FS_IOC_ENABLE_VERITY refuses such files as well.
For example, on ext4 with the encrypt feature:
mkdir d && chattr +i d
setfattr -n user.x -v 1 d -> Operation not permitted
FS_IOC_SET_ENCRYPTION_POLICY on d -> succeeds, d now has a policy
Return -EPERM for an immutable or append-only directory before setting
the policy, as may_write_xattr() does. Calls that find the same policy
already set still return 0, since they don't change the inode. Update
the documented EPERM cases to match.
Found with a static analysis tool and checked by hand.
Fixes: 9bd8212f981e ("ext4 crypto: add encryption policy and password salt support")
Fixes: f424f664f0e8 ("f2fs crypto: add encryption policy and password salt support")
Cc: stable@vger.kernel.org
Co-developed-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Suyuan Wang <stevewong@hust.edu.cn>
---
Changes in v2:
- Point the Fixes tags at the ext4 and f2fs commits that added the
ioctls, and ask for a stable backport (Eric).
- Document the new EPERM case in fscrypt.rst (Eric).
Tested in QEMU (x86_64, ext4 with -O encrypt, v1 policy). Before the
patch the ioctl succeeds on both immutable and append-only empty
directories; with it both get -EPERM, and an ordinary empty directory
still gets its policy.
v1: https://lore.kernel.org/all/20261119-fscrypt-v1-1-c07e1dd93644@gmail.com/
---
Documentation/filesystems/fscrypt.rst | 2 +-
fs/crypto/policy.c | 2 ++
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/Documentation/filesystems/fscrypt.rst b/Documentation/filesystems/fscrypt.rst
index 6e2e511a70bb..188295f09b21 100644
--- a/Documentation/filesystems/fscrypt.rst
+++ b/Documentation/filesystems/fscrypt.rst
@@ -736,7 +736,7 @@ FS_IOC_SET_ENCRYPTION_POLICY can fail with the following errors:
feature flag enabled using ``tune2fs -O encrypt`` or ``mkfs.ext4 -O
encrypt``.)
- ``EPERM``: this directory may not be encrypted, e.g. because it is
- the root directory of an ext4 filesystem
+ the root directory of an ext4 filesystem or is immutable or append-only
- ``EROFS``: the filesystem is readonly
Getting an encryption policy
diff --git a/fs/crypto/policy.c b/fs/crypto/policy.c
index 6dd510f93e6d..86a8254bc3a8 100644
--- a/fs/crypto/policy.c
+++ b/fs/crypto/policy.c
@@ -553,6 +553,8 @@ int fscrypt_ioctl_set_policy(struct file *filp, const void __user *arg)
ret = -ENOTDIR;
else if (IS_DEADDIR(inode))
ret = -ENOENT;
+ else if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+ ret = -EPERM;
else if (!inode->i_sb->s_cop->empty_dir(inode))
ret = -ENOTEMPTY;
else
next reply other threads:[~2026-11-20 4:26 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-20 4:26 Suyuan Wang via B4 Relay [this message]
2026-11-20 4:26 ` [PATCH v2] fscrypt: Don't set a policy on immutable or append-only directories Suyuan 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=20261120-fscrypt-v2-1-fb752135ea32@gmail.com \
--to=devnull+shyckless.gmail.com@kernel.org \
--cc=ebiggers@kernel.org \
--cc=jaegeuk@kernel.org \
--cc=linux-fscrypt@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=shyckless@gmail.com \
--cc=tytso@mit.edu \
/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.