Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 30 Jun 2023 08:51:17 +0200
From:   Christian Brauner <brauner@kernel.org>
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Kees Cook <keescook@chromium.org>, linux-kernel@vger.kernel.org
Subject: Re: [GIT PULL] pid: use flex array
Message-ID: <20230630-testphasen-orangen-0e54486a267d@brauner>
References: <20230628-pokal-puzzeln-5199c679b051@brauner>
 <CAHk-=wiBXJOzkez2Rd=cQ5ckttJq6OdYtArFmCtVQHyeuQBGrw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <CAHk-=wiBXJOzkez2Rd=cQ5ckttJq6OdYtArFmCtVQHyeuQBGrw@mail.gmail.com>
Precedence: bulk

On Thu, Jun 29, 2023 at 04:52:43PM -0700, Linus Torvalds wrote:
> On Wed, 28 Jun 2023 at 03:37, Christian Brauner <brauner@kernel.org> wrote:
> >
> > This contains Kees' work to make struct upid in struct pid a proper
> > flexible array and thus gets rid of a bunch of syzbot UBSAN warnings.
> 
> Hmm. Of this, about half were replacing "array + index" with "&array[index]".
> 
> Honestly, it makes no difference, but the reverse is also true: the
> "array + index" is *very* traditional, and if people have problems
> with that simple syntax I really don't know what to say. It's kind of
> core C. It's *literally* how arrays work, and what the '[]' operator
> means.

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.

> 
> And of the remaining half, half again is using a truly disgusting
> 
>     struct_size((struct pid *)0, numbers, X)

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
which is why I didn't ask for it to be changed. See commits
48658213202c ("scsi: megaraid_sas: Use struct_size() in code related to struct MR_PD_CFG_SEQ_NUM_SYNC") 
5b12a568cc6f ("scsi: hptiop: Use struct_size() helper in code related to struct hpt_iop_request_scsi_command"
as examples.

> 
> thing. That is *GARBAGE*. It's garbage for too many reasons for me to
> actually pull this sh*t, but let me just name them:
> 
>  - 0 isn't a pointer. Stop doing that.
> 
>  - dammit, we have 'struct_size_t' that does the above disgusting cast
> without getting that simple thing wrong.
> 
> In other words, this pull request contained half pointless and
> unrelated churn, and 25% actual garbage.
> 
> In other words, I'm not pulling this to just get the remaining 25%.

Sure. @Kees, I'd appreciate it if you could change the patch according
to the comments here.