netdev - Re: [PATCH -next v2] wifi: mac80211: use max to simplify the code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <4967a09d4c00532148aacd7dc4b5e21902ef7721.camel@sipsolutions.net>
Date: Tue, 27 Aug 2024 09:39:00 +0200
From: Johannes Berg <johannes@...solutions.net>
To: Hongbo Li <lihongbo22@...wei.com>, davem@...emloft.net,
 edumazet@...gle.com,  kuba@...nel.org, pabeni@...hat.com, kvalo@...nel.org
Cc: linux-wireless@...r.kernel.org, netdev@...r.kernel.org
Subject: Re: [PATCH -next v2] wifi: mac80211: use max to simplify the code

On Tue, 2024-08-27 at 15:29 +0800, Hongbo Li wrote:
> 
> On 2024/8/27 15:25, Johannes Berg wrote:
> > On Tue, 2024-08-27 at 11:03 +0800, Hongbo Li wrote:
> > > The following Coccinelle/coccicheck warning reported by
> > > minmax.cocci:
> > >      WARNING opportunity for max()
> > 
> > Yeah well, maybe sometimes we shouldn't blindly follow tools ...
> > 
> > > Let's use max() to simplify the code and fix the warning.
> > 
> > You should explain why.
> > 
> > I think only one out of four changes in this patch is correct,
> > semantically.
> > 
> You mean sometimes we should keep the variable type in comparison?

No, I just don't think these are semantically calculations of a maximum,
even if they look that way.

That's why I asked: Why are you making this change? It looks like you're
making this change just because you want coccicheck to be silent here.
But that's *really* not a good reason! Don't do that, ever, *think*
about the changes you're making too.

We should consider the primary consumer of the code to be *people*, not
the compiler or tools like coccicheck. And for *people*, applying max()
to a link ID makes no sense. It's a link ID, not any kind of value that
applying max() to makes any sense.
In contrast, for the timeout value there that you changed, that _does_
make sense: it clearly wants to take the longer of the two durations.

So then why do we have patterns that look like max(0, link_id)? That's
because we treat -1 as a special value indicating "no link, but for the
whole sta/vif/...", "don't care about the link" or "MLD not used"
(depending on the context). Internally in the code, however, we use 0
for non-MLD to simplify older drivers and internal logic.

That's why we end up with "link_id >= 0 ? link_id : 0" in some places.
But it's fundamentally not max() even though it looks like it. Replacing
it with max() does this a disservice.

Now arguably open-coding it often (though three perhaps isn't often, but
I'm surprised it's only three times) maybe isn't a great idea either,
but then that should be solved differently.

So yeah, please think about changes, don't make them blindly.

johannes