[cfe-dev] [StaticAnalyzer]: Cant get my checker to output

[Anna Zaks]

Hi Arjun,

The reason why the callback is not called is that IfStmt is not a CFGElement - it's a CFGTerminator (it expresses only control flow). See the CFG block dump below. We only call checkPostStmt/checkPreStmt for statements that correspond to CFG elements. 

If you want to visit the conditions within the if statements (or other branches), use checkBranchCondition() instead. 

Cheers,
Anna.

[ B3 ]
      1: foo
      2: [B3.1]
      3: [B3.2]()
      4: int a = foo();
      5: foo
      6: [B3.5]
      7: [B3.6]()
      8: int b = foo();
      9: a
     10: &[B3.9]
     11: "%d"
     12: [B3.11]
     13: [B3.12]
     14: scanf
     15: [B3.14]
     16: [B3.15]([B3.13], [B3.10])
     17: a
     18: [B3.17]
     19: b
     20: [B3.19]
     21: [B3.18] + [B3.20]
     22: a
     23: [B3.22] = [B3.21]
     24: a
     25: [B3.24]
     26: 2
     27: [B3.25] == [B3.26]
      T: if [B3.27]
    Predecessors (1): B4
    Successors (2): B2 B1

On Sep 28, 2011, at 9:03 PM, Arjun Singri wrote:

> My checker was able to output (the debug messages show up) when I changed the IfStmt to BinaryOperator. So my checker has been registered correctly. It is not able to deal with IfStmt for some reason.
> 
> Here is the code I am trying to analyze:
> 
> #include<stdio.h>
> 
>   int foo() { return 2; }
> 
>   int main(void)
>   {
>       int a = foo();
>       int b = foo();
> 
>       scanf("%d", &a);
> 
>       a = a + b;
>       if (a == 2)
>       {
>           a = 4;
>       }
> 
>       printf("%d", a);
> 
>       return 0;
>   }
> 
> 
> 
> 
> On Wed, Sep 28, 2011 at 7:54 PM, Jim Goodnow II <Jim at thegoodnows.net> wrote:
> Hi Arjun,
> 
> Do you have a short example of the code that you are analyzing. It is
> possible that the static analyzer has determined that the if statement
> is in a path that never gets executed. It should still be in the AST
> though. You can use -ast-dump to look at the AST directly.
> 
>  - jim
> 
> On 9/28/2011 7:40 PM, John McCall wrote:
> > On Sep 28, 2011, at 7:34 PM, funceval wrote:
> >> Not even constant folding and constant propagation?
> > Correct.  A few places in the AST, like array dimensions, do compute and store the value of a constant expression, but only in unevaluated contexts, and the original source information is never thrown away.
> >
> > John.
> > _______________________________________________
> > cfe-dev mailing list
> > cfe-dev at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
> >
> >
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20110928/fc740094/attachment.html>