Re: Sparse-LLVM issue compiling NULL pointers

[Luc Van Oostenryck <luc.vanoostenryck@gmail.com>]

On Wed, Mar 01, 2017 at 02:45:03PM +0000, Dibyendu Majumdar wrote:
> On 1 March 2017 at 10:58, Dibyendu Majumdar <mobile@majumdar.org.uk> wrote:
> > On 28 February 2017 at 17:03, Luc Van Oostenryck
> > <luc.vanoostenryck@gmail.com> wrote:
> >> On Tue, Feb 28, 2017 at 4:09 PM, Luc Van Oostenryck
> >> <luc.vanoostenryck@gmail.com> wrote:
> >>> There is indeed some problems regarding this, we looked a bit at this
> >>> some weeks ago. However I firmly believe that the information about
> >>> the type belong to the operations and not the values.
> >>
> >
> > I am trying to work out how a value pseudo correct type can be
> > determined when the pseudo is a function call argument. Would
> > appreciate any pointers on this. The current implementation of
> > pseudo_to_value() uses the function call instruction which is
> > incorrect.
> >
> 
> I have implemented a solution that get the type information from the
> function prototype for pseudo values when processing function
> arguments, but not sure this is correct.

You need something like:
+struct symbol *argument_type(pseudo_t src)
+{
+	struct entrypoint *ep = src->def->bb->ep;
+	struct symbol_list *args = ep->name->ctype.base_type->arguments;
+	struct symbol *arg;
+	int i = 0;
+	FOR_EACH_PTR(args, arg) {
+		if (++i == src->nr)
+			return arg;
+	} END_FOR_EACH_PTR(arg);
+
+	assert(0);
+}
 
> Anyway have hit a bunch of other issues with sparse-llvm ... :-(

Each day its problem (and happily its solution too!).


Luc