Re: [PATCH v2 07/10] PM: hibernate: Use TPM-backed keys to encrypt image

[Kees Cook <keescook@chromium.org>]

On Fri, Sep 23, 2022 at 03:23:43PM -0700, Evan Green wrote:
> On Tue, Sep 20, 2022 at 4:16 PM Kees Cook <keescook@chromium.org> wrote:
> > On Tue, Aug 23, 2022 at 03:25:23PM -0700, Evan Green wrote:
> > > [...]
> > > +     ret = key_instantiate_and_link(key, keyinfo, strlen(keyinfo) + 1, NULL,
> > > +                                    NULL);
> >
> > You want to keep the trailing NUL byte here so it's easier to read back
> > later? Reading it back will need to verify the trailing NUL regardless.
> > (Does this get read back?)
> 
> Are you referring to the trailing nul on keyinfo? The keyinfo string
> is only used within this function, as key_instantiate_and_link()
> receives it as a parameter string. I can see that
> trusted_instantiate() also null-terminates the buffer defensively for
> itself, but it still seemed prudent to hand in a terminated string. I
> can remove the + 1 if you think it's better.

No, I like having the trailing NUL byte -- it seems more robust that
way. I just wanted to understand who the consumer was going to be. I'm
clear now; thank you! :)

-- 
Kees Cook