maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Fix bug: test/Regression/Other/2002-08-02-DomSetProblem.ll 

llvm-svn: 3213
This commit is contained in:
Chris Lattner 2002-08-02 16:51:27 +00:00
parent 35d69af46f
commit 6d149e0215
1 changed files with 12 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
llvm/lib/VMCore/Dominators.cpp
View File

@ -60,7 +60,7 @@ bool DominatorSet::runOnFunction(Function &F) {
// in at least once. We are guaranteed to have this because we are
// traversing the graph in DFO and have handled start nodes specially.
//
while (Doms[*PI].size() == 0) ++PI;
while (Doms[*PI].empty()) ++PI;
WorkingSet = Doms[*PI];
for (++PI; PI != PEnd; ++PI) { // Intersect all of the predecessor sets
@ -79,6 +79,17 @@ bool DominatorSet::runOnFunction(Function &F) {
WorkingSet.clear(); // Clear out the set for next iteration
}
} while (Changed);
// Every basic block in the function should at least dominate themselves, and
// thus every basic block should have an entry in Doms. The one case where we
// miss this is when a basic block is unreachable. To get these we now do an
// extra pass adding self dominance info to the DomSet if the block doesn't
// already have an entry.
//
for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I)
if (!Doms.count(I))
Doms[I].insert(I);
return false;
}