Re: [PATCH 3/3] RFC: p9auth: add p9auth fs

[Oleg Nesterov <oleg@redhat.com>]

On 04/28, Serge E. Hallyn wrote:
>
> Quoting Oleg Nesterov (oleg@redhat.com):
>
> > > +static ssize_t p9auth_use_write(struct file *file, const char __user *buffer,
> > > +			       size_t count, loff_t *ppos)
> > > +{
> > > +	ssize_t retval = -ENOMEM;
> > > +	char *user_buf;
> > > +
> > > +	if (mutex_lock_interruptible(&cap_mutex))
> > > +		return -EINTR;
> >
> > EINTR doesn't look exactly right here, especially if TIF_SIGPENDING is
> > spurious. Probably ERESTARTNOINTR makes more sense. Or mutex_lock_killable().
>
> Ashwin had had this as ERESTARTSYS I believe.  I'd read something about
> userspace should only see -EINTR so I changed it.

Yes, ERESTARTxxx should not be visible to the user-space. But it is OK
to return it from the syscall if signal_pending() is true, in this case
it will be changed to EINTR or the system call will be restarted.

ERESTARTNOINTR always restarts the syscall, perphaps after handling the
signal if it is really pending.

> > > +	if (copy_from_user(user_buf, buffer, count)) {
> > > +		retval = -EFAULT;
> > > +		goto out;
> > > +	}
> > > +
> > > +	retval = use_setuid_capability(user_buf);
> >
> > It seems that use_setuid_capability() pathes assume that user_buf is
> > null terminated? Say, parse_user_capability() does kstrdup(user_buf).
>
> I kzalloc()d to count+1 before, and only copy_from_user() count bytes,
> so the last byte should always be 0.

Ah, indeed, thanks.

Oleg.