[Soot-list] Re: Global variables in soot

[Eric Bodden]

Ok, for this you have to write your own analysis. The good thing is
that the only case that a field can be referenced is by assigning it
into a local variable. So you need a flow analysis (see
ForwardFlowAnalysis) that keeps track of each local variable and
stores the information whether this local was assigned a field
reference and if so, which one. Then, when this local is used in a
binary (or unary?) expression, you access that analysis information.

Eric

On 01/05/07, John Chaitanya Kati <johnkati at cse.iitb.ac.in> wrote:
> by expressions i mean those involving two operands like a*b, a+b etc...
>
>
> On Fri, 4 May 2007, Eric Bodden wrote:
>
> > So in your example, would it ot suffice to say that...
> >
> > $i2 = <Test: int a>;
> >
> > ... here the expression "<Test: int a>" is available in $i2?
> >
> > So whenever a field reference is assigned to a local, it's expression
> > is available in that local.
> >
> > Eric
> >
> > On 04/05/07, John Chaitanya Kati <johnkati at cse.iitb.ac.in> wrote:
> >> Hi,
> >>     I am trying to implement inter-procedural available expression
> >> analysis using call string approach. As far as i know, i have to take care
> >> of expressions which have only one or both operands as global and create a
> >> bitvector of only those expressions. If both variables in the expression
> >> are local then i need not care for such type of expressions. But its
> >> getting difficult for me now...
> >>
> >>
> >>
> >>
> >>
> >> On Fri, 4 May 2007, Eric Bodden wrote:
> >>
> >> > Hi again.
> >> >
> >> > Sorry but I overlooked something. As Patrick just reminded me,
> >> > points-to analysis only works for reference types, while the field in
> >> > your example is an int. So it seems like you would have to write your
> >> > own analysis for this.
> >> >
> >> > What are you actually trying to achieve?
> >> >
> >> > Eric
> >> >
> >> > On 04/05/07, Eric Bodden <eric.bodden at mail.mcgill.ca> wrote:
> >> >> Ah, I see.
> >> >>
> >> >> In this case, I think you need points-to analysis. You want to find
> >> >> out which values $i2 could possibly point to. We have some tutorials
> >> >> and papers about how to use points-to analysis in Soot online.
> >> >>
> >> >> Eric
> >> >>
> >> >> On 04/05/07, John Chaitanya Kati <johnkati at cse.iitb.ac.in> wrote:
> >> >> > for my below program, jimple is giving the following output
> >> >> >         int i0, i1, $i2, $i3;
> >> >> >          $i2 = <Test: int a>;
> >> >> >          i1 = $i2 * i0;
> >> >> >
> >> >> > so its treating $i2 as local since its getting declared. But actually
> >> it
> >> >> > represents variable 'a' which is global. I need this information for
> >> >> > interprocedural available expression analysis. So how do i resolve it
> >> ?
> >> >> >
> >> >> >
> >> >> >
> >> >> >
> >> >> > On Fri, 4 May 2007, Eric Bodden wrote:
> >> >> >
> >> >> > > Ok, so what you mean by global variables are actually fields. it's
> >> as
> >> >> > > I said: local variables are represented by soot.Local's and
> >> >> > > fields/global variables by soot.jimple.FieldRef's.
> >> >> > >
> >> >> > > Eric
> >> >> > >
> >> >> > > On 04/05/07, John Chaitanya Kati <johnkati at cse.iitb.ac.in> wrote:
> >> >> > >> By global variables i meant, which are available to all the methods
> >> of
> >> >> the
> >> >> > >> class.
> >> >> > >> class Test
> >> >> > >> {       static public int a=5;
> >> >> > >>         static public void main(String args[])
> >> >> > >>         {
> >> >> > >>                 int b=10,c;
> >> >> > >>                 c = a*b;
> >> >> > >>                 add(a,b);
> >> >> > >>         }
> >> >> > >>         static void add(int i, int j)
> >> >> > >>         {
> >> >> > >>                 i=i+j;
> >> >> > >>                 a=15;
> >> >> > >>         }
> >> >> > >> }
> >> >> > >> a is global, b,c are local to main. I want to find out such
> >> variables.
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >>
> >> >> > >> On Fri, 4 May 2007, Eric Bodden wrote:
> >> >> > >>
> >> >> > >> > What's a global variable?
> >> >> > >> >
> >> >> > >> > Soot knows variables of type soot.Local. Those are obviously
> >> local.
> >> >> > >> > Then there also are field references, which are obviously field
> >> >> > >> > references. Are those what you mean by global? (Of course, a
> >> local
> >> >> can
> >> >> > >> > well be assigned a field reference, the same way as it can happen
> >> in
> >> >> > >> > Java code.)
> >> >> > >> >
> >> >> > >> > Eric
> >> >> > >> >
> >> >> > >> > On 04/05/07, John Chaitanya Kati <johnkati at cse.iitb.ac.in> wrote:
> >> >> > >> >> Hi,
> >> >> > >> >>    Is there a way to find out if a variable is global or not in
> >> >> soot ?
> >> >> > >> >>
> >> >> > >> >> --
> >> >> > >> >> Thanks for any help in advance
> >> >> > >> >> John Chaitanya
> >> >> > >> >>
> >> >> > >> >
> >> >> > >> >
> >> >> > >> >
> >> >> > >>
> >> >> > >> --
> >> >> > >>
> >> >> > >> Thanks
> >> >> > >> John Chaitanya
> >> >> > >> Like a rose trampled on the ground, you took the fall and thought
> >> of
> >> >> me...
> >> >> > >>
> >> >> > >
> >> >> > >
> >> >> > >
> >> >> >
> >> >> > --
> >> >> >
> >> >> > Thanks
> >> >> > John Chaitanya
> >> >> > Like a rose trampled on the ground, you took the fall and thought of
> >> >> me...
> >> >> >
> >> >>
> >> >>
> >> >> --
> >> >> Eric Bodden
> >> >> Sable Research Group
> >> >> McGill University, Montréal, Canada
> >> >>
> >> >
> >> >
> >> >
> >>
> >> --
> >>
> >> Thanks
> >> John Chaitanya
> >> Like a rose trampled on the ground, you took the fall and thought of me...
> >
> >
> >
>
> --
>
> Thanks
> John Chaitanya
> Like a rose trampled on the ground, you took the fall and thought of me...


-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada