Initialize securityfs as a pseudo filesystem. This raises SB_I_PSEUDO and SB_I_NOEXEC. securityfs never holds executables.
But securityfs needs device nodes. AppArmor puts a .null device node into securityfs. On exec it replaces file descriptors the new profile isn't allowed to use with that node and opens it via dentry_open() which doesn't care about SB_I_NODEV. Such a file descriptor can still be reopened through /proc/<pid>/fd/<nr> though and that's what opening /dev/stderr does. The node itself can also be opened by path as /sys/kernel/security/apparmor/.null. Both go through may_open_dev() which refuses device nodes with SB_I_NODEV raised. So clear SB_I_NODEV again in securityfs_fill_super(). Reviewed-by: Alexander Mikhalitsyn <[email protected]> Reviewed-by: Chris Down <[email protected]> Signed-off-by: Christian Brauner (Amutable) <[email protected]> --- security/inode.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/security/inode.c b/security/inode.c index 0804023676741..edaebf0f1972d 100644 --- a/security/inode.c +++ b/security/inode.c @@ -14,6 +14,7 @@ #include <linux/kobject.h> #include <linux/fs.h> #include <linux/fs_context.h> +#include <linux/pseudo_fs.h> #include <linux/mount.h> #include <linux/pagemap.h> #include <linux/init.h> @@ -49,6 +50,8 @@ static int securityfs_fill_super(struct super_block *sb, struct fs_context *fc) return error; sb->s_op = &securityfs_super_operations; + /* Userspace may open AppArmor's .null device node by path. */ + sb->s_iflags &= ~SB_I_NODEV; return 0; } @@ -64,6 +67,7 @@ static const struct fs_context_operations securityfs_context_ops = { static int securityfs_init_fs_context(struct fs_context *fc) { + fc_init_pseudo(fc); fc->ops = &securityfs_context_ops; return 0; } -- 2.53.0
