On Jun 8, 2007, at 10:47 AM, Chris Lattner wrote:
>> --- llvm/include/llvm/Analysis/Dominators.h:1.95 Thu Jun 7
>> 16:34:22 2007
>> +++ llvm/include/llvm/Analysis/Dominators.h Thu Jun 7 17:17:16 2007
>> @@ -142,6 +142,16 @@
>> return getNode(BB);
>> }
>>
>> + /// getIDomBlock - return basic block BB's immediate domiantor
>> basic block.
>> + ///
>> + BasicBlock *getIDomBlock(BasicBlock *BB) {
>> + DomTreeNode *N = getNode(BB);
>> + assert (N && "Missing dominator tree node");
>> + DomTreeNode *I = N->getIDom();
>> + assert (N && "Missing immediate dominator");
>> + return I->getBlock();
>> + }
>
> This will assert and die if called on the entry node, because it has
> no idom. Would it make sense to have this function return null in
> this case? If so, please document it as returning null in that
> case.
Current clients do not expect null here.
> Also, "domiantor" is misspelled in the comment,
oops
-
Devang
_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
