Re: [PATCH v10 7/9] proc: move hidepid values to uapi as they are user interface to mount

[Kees Cook]

On Sun, Mar 29, 2020 at 12:00:46AM +0100, Alexey Gladkov wrote:
> On Sat, Mar 28, 2020 at 02:53:49PM -0700, Kees Cook wrote:
> > > > > +/* definitions for hide_pid field */
> > > > > +enum {
> > > > > +	HIDEPID_OFF            = 0,
> > > > > +	HIDEPID_NO_ACCESS      = 1,
> > > > > +	HIDEPID_INVISIBLE      = 2,
> > > > > +	HIDEPID_NOT_PTRACEABLE = 4,
> > > > > +};
> > > > Should the numeric values still be UAPI if there is string parsing now?
> > > 
> > > I think yes, because these are still valid hidepid= values.
> > 
> > But if we don't expose the values, we can do whatever we like with
> > future numbers (e.g. the "is this a value or a bit field?" question).
> 
> Alexey Dobriyan suggested to put these parameters into the UAPI and it
> makes sense because these are user parameters.

Okidokey. :) Anyway, ignore my HIDEPID_MAX idea then, since this could
become a bitfield. Just checking for individual bits is the way to go
for now. Sorry for the noise.

-- 
Kees Cook