Re: violating function pointer signature

From: Segher Boessenkool <segher@kernel.crashing.org>
To: Florian Weimer <fw@deneb.enyo.de>
Cc: Steven Rostedt <rostedt@goodmis.org>,
	Nick Desaulniers <ndesaulniers@google.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Sami Tolvanen <samitolvanen@google.com>,
	Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Matt Mullins <mmullins@mmlx.us>, Ingo Molnar <mingo@redhat.com>,
	Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Dmitry Vyukov <dvyukov@google.com>,
	Martin KaFai Lau <kafai@fb.com>, Song Liu <songliubraving@fb.com>,
	Yonghong Song <yhs@fb.com>, Andrii Nakryiko <andriin@fb.com>,
	John Fastabend <john.fastabend@gmail.com>,
	KP Singh <kpsingh@chromium.org>, netdev <netdev@vger.kernel.org>,
	bpf <bpf@vger.kernel.org>, Kees Cook <keescook@chromium.org>,
	Josh Poimboeuf <jpoimboe@redhat.com>,
	linux-toolchains@vger.kernel.org
Subject: Re: violating function pointer signature
Date: Wed, 18 Nov 2020 12:55:40 -0600	[thread overview]
Message-ID: <20201118185540.GL2672@gate.crashing.org> (raw)
In-Reply-To: <87o8jutt2h.fsf@mid.deneb.enyo.de>

On Wed, Nov 18, 2020 at 07:31:50PM +0100, Florian Weimer wrote:
> * Segher Boessenkool:
> 
> > On Wed, Nov 18, 2020 at 12:17:30PM -0500, Steven Rostedt wrote:
> >> I could change the stub from (void) to () if that would be better.
> >
> > Don't?  In a function definition they mean exactly the same thing (and
> > the kernel uses (void) everywhere else, which many people find clearer).
> 
> And I think () functions expected a caller-provided parameter save
> area on powerpc64le, while (void) functions do not.

Like I said (but you cut off, didn't realise it matters I guess):

> > In a function declaration that is not part of a definition it means no
> > information about the arguments is specified, a quite different thing.

Since the caller does not know if the callee will need a save area, it
has to assume it does.  Similar is true for many ABIs.

> It does not
> matter for an empty function, but GCC prefers to use the parameter
> save area instead of setting up a stack frame if it is present.  So
> you get stack corruption if you call a () function as a (void)
> function.  (The other way round is fine.)

If you have no prototype for a function, you have to assume worst case,
yes.  Calling things "a () function" can mean two things (a declaration
that is or isn't a definition, two very different things), so it helps
to be explicit about it.

Just use (void) and do not worry :-)

Segher