> One thing, whatever you end up passing to vfs_create() please make sure
> to retrieve mnt_userns once so permission checking and object creation
> line-up:
>
> int vfs_create(struct vfsmount *mnt, struct inode *dir,
> struct dentry *dentry, umode_t mode, bool want_excl)
> {
> struct user_namespace *mnt_userns;
>
> mnt_userns = mnt_user_ns(mnt);
>
> int error = may_create(mnt_userns, dir, dentry);
> if (error)
> return error;
>
> if (!dir->i_op->create)
> return -EACCES; /* shouldn't it be ENOSYS? */
> mode &= S_IALLUGO;
> mode |= S_IFREG;
> error = security_inode_create(dir, dentry, mode);
> if (error)
> return error;
> error = dir->i_op->create(mnt_userns, dir, dentry, mode, want_excl);
> if (!error)
> fsnotify_create(mnt, dir, dentry);
> return error;
> }
>
Christian,
What is the concern here?
Can mnt_user_ns() change under us?
I am asking because Al doesn't like both mnt_userns AND path to
be passed to do_tuncate() => notify_change()
So I will need to retrieve mnt_userns again inside notify_change()
after it had been used for security checks in do_open().
Would that be acceptable to you?
Thanks,
Amir.
