[PATCH] D30322: [SCCP] Resolve indirect branch target when possible.

[Davide Italiano via Phabricator via llvm-commits]

davide added inline comments.


================
Comment at: lib/Transforms/Scalar/SCCP.cpp:1531-1548
+   if (auto *IBR = dyn_cast<IndirectBrInst>(TI)) {
+      if (!getValueState(IBR->getAddress()).isUnknown())
+        continue;
+
+      // If the input to SCCP is actually branch on undef, fix the undef to
+      // the first successor of the indirect branch.
+      if (isa<UndefValue>(IBR->getAddress())) {
----------------
I'm very wary of adding more code here, as it caused a fair amount of pain in the past. This, tho, seems to be in line with what was already there.


================
Comment at: lib/Transforms/Scalar/SCCP.cpp:1895-1896
             assert(isa<UndefValue>(SI->getCondition()) && "Switch should fold");
+          } else if (auto *IBR = dyn_cast<IndirectBrInst>(I)) {
+            assert(isa<UndefValue>(IBR->getAddress()) && "IndirectBr should fold");
           } else {
----------------
Can this assertion ever fire?


https://reviews.llvm.org/D30322