Re: [PATCH] netfilter: conntrack: fix -Wformat

From: Nick Desaulniers <ndesaulniers@google.com>
To: Joe Perches <joe@perches.com>
Cc: Pablo Neira Ayuso <pablo@netfilter.org>,
	Jozsef Kadlecsik <kadlec@netfilter.org>,
	Florian Westphal <fw@strlen.de>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	Nathan Chancellor <natechancellor@gmail.com>,
	netfilter-devel@vger.kernel.org, coreteam@netfilter.org,
	Network Development <netdev@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	clang-built-linux <clang-built-linux@googlegroups.com>
Subject: Re: [PATCH] netfilter: conntrack: fix -Wformat
Date: Tue, 10 Nov 2020 14:00:05 -0800	[thread overview]
Message-ID: <CAKwvOdn50VP4h7tidMnnFeMA1M-FevykP+Y0ozieisS7Nn4yoQ@mail.gmail.com> (raw)
In-Reply-To: <4910042649a4f3ab22fac93191b8c1fa0a2e17c3.camel@perches.com>

On Sat, Nov 7, 2020 at 2:33 AM Joe Perches <joe@perches.com> wrote:
>
> On Fri, 2020-11-06 at 23:55 -0800, Nick Desaulniers wrote:
> > Clang is more aggressive about -Wformat warnings when the format flag
> > specifies a type smaller than the parameter. Fixes 8 instances of:
> >
> > warning: format specifies type 'unsigned short' but the argument has
> > type 'int' [-Wformat]
>
> Likely clang's -Wformat message is still bogus.
> Wasn't that going to be fixed?
>
> Integer promotions are already done on these types to int anyway.
> Didn't we have this discussion last year?
>
> https://lore.kernel.org/lkml/CAKwvOd=mqzj2pAZEUsW-M_62xn4pijpCJmP=B1h_-wEb0NeZsA@mail.gmail.com/
> https://lore.kernel.org/lkml/CAHk-=wgoxnmsj8GEVFJSvTwdnWm8wVJthefNk2n6+4TC=20e0Q@mail.gmail.com/
> https://lore.kernel.org/lkml/a68114afb134b8633905f5a25ae7c4e6799ce8f1.camel@perches.com/

Now I'll have to page in some old context...

The case we addressed last year was printing char with a wider format
string like %hd: https://reviews.llvm.org/rL369791,
https://bugs.llvm.org/show_bug.cgi?id=41467 and
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95588 have a little more
info but not much.  Which is the case that Linus commented on.  Let's
say we're printing a "wider format than intended." Those have been
fixed in Clang.  These cases are printing a "narrower format than
intended."  Two distinct cases.

>
> Look at commit cbacb5ab0aa0 ("docs: printk-formats: Stop encouraging use
> of unnecessary %h[xudi] and %hh[xudi]")
>
> The "h" and "hh" things should never be used. The only reason for them
> being used if if you have an "int", but you want to print it out as a
> "char" (and honestly, that is a really bad reason, you'd be better off
> just using a proper cast to make the code more obvious).
>
> So if what you have a "char" (or unsigned char) you should always just
> print it out as an "int", knowing that the compiler already did the
> proper type conversion.

Yeah, we could go through and remove %h and %hh to solve this, too, right?
-- 
Thanks,
~Nick Desaulniers