Re: [GIT PULL] pid: use flex array

[Kees Cook <keescook@chromium.org>]

On Fri, Jun 30, 2023 at 10:04:14AM +0200, Christian Brauner wrote:
> On Fri, Jun 30, 2023 at 12:12:22AM -0700, Linus Torvalds wrote:
> > On Thu, 29 Jun 2023 at 23:51, Christian Brauner <brauner@kernel.org> wrote:
> > >
> > > I have no preference for either syntax. Both work. But this is probably
> > > more an objection to this being mixed in with the flex array change in
> > > the first place.
> > 
> > Yes. I looked at it, and tried to figure out if it was related
> > somehow, and decided that no, it can't possibly be, and must be just
> > an unrelated change.

Yes, those changes were style changes because I was annoyed that a grep
for 'numbers[' didn't turn anything up. :P Since it's an array I think
it's just good form to use [] when accessing an element. But yes, it's
conceptually the same.

> > > I did react to that in the original review here:
> > > https://lore.kernel.org/all/20230518-zuneigen-brombeeren-0a57cd32b1a7@brauner
> > > but then I grepped for it and saw it done in a few other places already
> > 
> > Yeah, we do end up growing new uses of 'use 0 as a pointer' almost as
> > quickly as we get rid of them.

Apologies on this -- this patch was just before the addition of
struct_size_t(), so I missed it in the cleanup I did for that:
https://git.kernel.org/linus/d67790ddf0219aa0ad3e13b53ae0a7619b3425a2

> I've grepped around a bit and I saw that the
> struct_size((struct bla *)NULL, ...)
> pattern seems to be used in most places that have similar needs. Not
> sure if there's something nicer.

The above patch fixes them all (excepting struct pid). In retrospect, I
should have asked to carry the struct pid fix in the hardening tree due
to that.

> I gave this thing a stab myself since I have a few minutes and so Kees
> doesn't have to do it. Authorship retained and dropped the ack. Is the
> following more acceptable? 

Thanks for reworking it!

> [...]
> [brauner: dropped unrelated changes and remove 0 with NULL cast]

However, this should use struct_size_t(); I'll send a new patch and
double check that UBSAN stays happy, etc.

Sorry for the mess!

-Kees

-- 
Kees Cook