Re: violating function pointer signature

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Nov 18, 2020 at 02:59:29PM +0100, Florian Weimer wrote:
> * Peter Zijlstra:
> 
> > I think that as long as the function is completely empty (it never
> > touches any of the arguments) this should work in practise.
> >
> > That is:
> >
> >   void tp_nop_func(void) { }
> >
> > can be used as an argument to any function pointer that has a void
> > return. In fact, I already do that, grep for __static_call_nop().
> 
> You can pass it as a function parameter, but in general, you cannot
> call the function with a different prototype.  Even trivial
> differences such as variadic vs non-variadic prototypes matter.

I don't think any tracepoint uses variadic argument.

> The default Linux calling conventions are all of the cdecl family,
> where the caller pops the argument off the stack.  You didn't quote
> enough to context to tell whether other calling conventions matter in
> your case.

This is strictly in-kernel, and I think we're all cdecl, of which the
important part is caller-cleanup. The function compiles to:

	RET

so whatever the arguments are is irrelevant.

> > I'm not sure what the LLVM-CFI crud makes of it, but that's their
> > problem.
> 
> LTO can cause problems as well, particularly with whole-program
> optimization.

I don't think LTO can de-virtualize a dynamic array of function
pointers, so there's very little risk. That said, the __static_call_nop
case, where everything is inlined, is compiled sub-optimally for both
LLVM and GCC.