[Soot-list] calls to objects instantiated through reflection not detected

[Denis Bogdanas]

Strange,
I see there is Kind.REFL_CLASS_NEWINSTANCE

used in a few places to add edges to CG. What is it used for?

On 4 May 2016 at 09:02, Steven Arzt <Steven.Arzt at cased.de> wrote:

> Hi Denis,
>
>
>
> That is by design. Spark needs to propagate type information to the base
> object of a method call. If these base object comes from a reflective call
> or a factory method inside a phantom class, the type information is not
> available through the PAG propagation and there will not be any  outgoing
> call edges. In that case, your best bet is to take the declared callee.
>
>
>
> Best regards,
>
>   Steven
>
>
>
> *Von:* soot-list-bounces at CS.McGill.CA [mailto:
> soot-list-bounces at CS.McGill.CA] *Im Auftrag von *Denis Bogdanas
> *Gesendet:* Mittwoch, 4. Mai 2016 00:42
> *An:* soot-list at CS.McGill.CA
> *Betreff:* [Soot-list] calls to objects instantiated through reflection
> not detected
>
>
>
> Hi,
>
> Looks like Spark doesn't process the objects instantiated through
> reflection.
>
>
>
> For the following code:
>
> try {
>     locationManager = LocationManager.class.newInstance();
> } catch (I...) { }
> locationManager.addGpsStatusListener(null);
>
> edge to addGpsStatusListener() is not in the call graph. I've looked into
> the support for reflection in soot source code, looks like at least this
> scenario should be supported.
>
> Am I wrong somewhere?
>
> thanks,
>
> --
>
> Denis
>



-- 
Denis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.CS.McGill.CA/pipermail/soot-list/attachments/20160504/4015a90a/attachment.html