security: Place security_path_post_mknod() where the original IMA call was
authorRoberto Sassu <roberto.sassu@huawei.com>
Wed, 3 Apr 2024 07:57:29 +0000 (09:57 +0200)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 3 Apr 2024 17:21:32 +0000 (10:21 -0700)
Commit 08abce60d63f ("security: Introduce path_post_mknod hook")
introduced security_path_post_mknod(), to replace the IMA-specific call
to ima_post_path_mknod().

For symmetry with security_path_mknod(), security_path_post_mknod() was
called after a successful mknod operation, for any file type, rather
than only for regular files at the time there was the IMA call.

However, as reported by VFS maintainers, successful mknod operation does
not mean that the dentry always has an inode attached to it (for
example, not for FIFOs on a SAMBA mount).

If that condition happens, the kernel crashes when
security_path_post_mknod() attempts to verify if the inode associated to
the dentry is private.

Move security_path_post_mknod() where the ima_post_path_mknod() call was,
which is obviously correct from IMA/EVM perspective. IMA/EVM are the only
in-kernel users, and only need to inspect regular files.

Reported-by: Steve French <smfrench@gmail.com>
Closes: https://lore.kernel.org/linux-kernel/CAH2r5msAVzxCUHHG8VKrMPUKQHmBpE6K9_vjhgDa1uAvwx4ppw@mail.gmail.com/
Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Fixes: 08abce60d63f ("security: Introduce path_post_mknod hook")
Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Reviewed-by: Mimi Zohar <zohar@linux.ibm.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/namei.c
security/security.c

index ceb9ddf8dfdd4e3d8b711ab655bc928eeb3b3e50..c5b2a25be7d048b613a11ef77b6f2f37ec1d0142 100644 (file)
@@ -4050,6 +4050,8 @@ retry:
                case 0: case S_IFREG:
                        error = vfs_create(idmap, path.dentry->d_inode,
                                           dentry, mode, true);
+                       if (!error)
+                               security_path_post_mknod(idmap, dentry);
                        break;
                case S_IFCHR: case S_IFBLK:
                        error = vfs_mknod(idmap, path.dentry->d_inode,
@@ -4060,11 +4062,6 @@ retry:
                                          dentry, mode, 0);
                        break;
        }
-
-       if (error)
-               goto out2;
-
-       security_path_post_mknod(idmap, dentry);
 out2:
        done_path_create(&path, dentry);
        if (retry_estale(error, lookup_flags)) {
index 7e118858b545c15bc4d0d03787e01d715423bafe..0a9a0ac3f2662447699ec355253737fe6a8d67f0 100644 (file)
@@ -1793,11 +1793,11 @@ int security_path_mknod(const struct path *dir, struct dentry *dentry,
 EXPORT_SYMBOL(security_path_mknod);
 
 /**
- * security_path_post_mknod() - Update inode security field after file creation
+ * security_path_post_mknod() - Update inode security after reg file creation
  * @idmap: idmap of the mount
  * @dentry: new file
  *
- * Update inode security field after a file has been created.
+ * Update inode security field after a regular file has been created.
  */
 void security_path_post_mknod(struct mnt_idmap *idmap, struct dentry *dentry)
 {