Re: violating function pointer signature

[Steven Rostedt]

On Wed, 18 Nov 2020 13:48:37 -0600
Segher Boessenkool <segher@xxxxxxxxxxxxxxxxxxx> wrote:

> > With it_func being the func from the struct tracepoint_func, which is a
> > void pointer, it is typecast to the function that is defined by the
> > tracepoint. args is defined as the arguments that match the proto.  
> 
> If you have at most four or so args, what you wnat to do will work on
> all systems the kernel currently supports, as far as I can tell.  It
> is not valid C, and none of the compilers have an extension for this
> either.  But it will likely work.

Well, unfortunately, there's tracepoints with many more than 4 arguments. I
think there's one with up to 13!

> 
> > The problem we are solving is on the removal case, if the memory is tight,
> > it is possible that the new array can not be allocated. But we must still
> > remove the called function. The idea in this case is to replace the
> > function saved with a stub. The above loop will call the stub and not the
> > removed function until another update happens.
> > 
> > This thread is about how safe is it to call:
> > 
> > void tp_stub_func(void) { return ; }
> > 
> > instead of the function that was removed?  
> 
> Exactly as safe as calling a stub defined in asm.  The undefined
> behaviour happens if your program has such a call, it doesn't matter
> how the called function is defined, it doesn't have to be C.
> 
> > Thus, we are indeed calling that stub function from a call site that is not
> > using the same parameters.
> > 
> > The question is, will this break?  
> 
> It is unlikely to break if you use just a few arguments, all of simple
> scalar types.  Just hope you will never encounter a crazy ABI :-)

But in most cases, all the arguments are of scaler types, as anything else
is not recommended, because copying is always slower than just passing a
pointer, especially since it would need to be copied for every instance of
that loop. I could do an audit to see if there's any that exist, and perhaps
even add some static checker to make sure they don't.

-- Steve