llvm dev - Sep 2006 - [LLVMdev] FunctionPass requiring SCCs

I have a FunctionPass F that needs a list of all the SCCs for use in its 
doFinalization() method.  Let's say I write a CallGraphSCCPass C that 
creates an array of all SCCs.  Let C be required by F, and let F call 
getAnalysis<C>() from its doFinalization() method.  Am I guaranteed that 
C's runOnSCC() method will have executed on all SCCs before F's 
doFinalization() method?  In other words, when F calls getAnalysis<C>() 
from its doFinalization() method, am I guaranteed that C will have been 
able to build the array of SCCs?  If not, how else might I structure things?

Regards,
Ryan

Hi Ryan,

On 9/29/06, Ryan M. Lefever <lefever at crhc.uiuc.edu>
wrote:
> I have a FunctionPass F that needs a list of all the SCCs for use in its
> doFinalization() method.  Let's say I write a CallGraphSCCPass C that
> creates an array of all SCCs.  Let C be required by F, and let F call
> getAnalysis<C>() from its doFinalization() method.  Am I guaranteed
that
> C's runOnSCC() method will have executed on all SCCs before F's
> doFinalization() method?  In other words, when F calls
getAnalysis<C>()
> from its doFinalization() method, am I guaranteed that C will have been
> able to build the array of SCCs?  If not, how else might I structure
things?
That should work. I've written such code, something like:
bool doFinalization(Module &M) {
  CallGraph &CG = getAnalysis<CallGraph>();
  for (scc_iterator<CallGraph*> SCCI = scc_begin(&CG),
    SCCE = scc_end(&CG); SCCI != SCCE; ++SCCI) {
    unsigned size = (*SCCI).size();
    for (unsigned i = 0; i < size; ++i) {
      Function *F = (*SCCI)[i]->getFunction();
      ......
    }
  }
  ....
}

Check out scc_* iterators. Also note that the call graph
is not aware of the indirect calls, so you will need to write your
own CG implementation if you need to handle function pointers
soundly.

Domagoj Babic

On Sep 29, 2006, at 2:05 PM, Domagoj Babic wrote:
>
> Check out scc_* iterators. Also note that the call graph
> is not aware of the indirect calls, so you will need to write your
> own CG implementation if you need to handle function pointers
> soundly.
>

Chris, is this true?  If so, it seems like a bad property for the  
CallGraphSCCPass framework.

--Vikram
http://www.cs.uiuc.edu/~vadve
http://llvm.cs.uiuc.edu/