[SimplifyCFG] Remove limitation on sinking of load/store of alloca (#104788)

This is a followup to https://github.com/llvm/llvm-project/pull/104579
to remove the limitation on sinking loads/stores of allocas entirely,
even if this would introduce a phi node.

Nowadays, SROA supports speculating load/store over select/phi.
Additionally, SimplifyCFG with sinking only runs at the end of the
function simplification pipeline, after SROA. I checked that the two
tests modified here still successfully SROA after the SimplifyCFG
transform.

We should, however, keep the limitation on lifetime intrinsics. SROA
does not have speculation support for these, and I've also found that
the way these are handled in the backend is very problematic
(https://github.com/llvm/llvm-project/issues/104776), so I think we
should leave them alone.

1 files changed, 4 insertions, 15 deletions

diff --git a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
index da4d57f..92e2d18 100644
--- a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -2051,21 +2051,10 @@ static bool canSinkInstructions(
       return I->getOperand(OI) == I0->getOperand(OI);
     };
     if (!all_of(Insts, SameAsI0)) {
-      // Because SROA historically couldn't handle speculating stores of
-      // selects, we try not to sink loads, stores or lifetime markers of
-      // allocas when we'd have to create a PHI for the address operand.
-      // TODO: SROA supports speculation for loads and stores now -- remove
-      // this hack?
-      if (isa<StoreInst>(I0) && OI == 1 &&
-          any_of(Insts, [](const Instruction *I) {
-            return isa<AllocaInst>(I->getOperand(1)->stripPointerCasts());
-          }))
-        return false;
-      if (isa<LoadInst>(I0) && OI == 0 &&
-          any_of(Insts, [](const Instruction *I) {
-            return isa<AllocaInst>(I->getOperand(0)->stripPointerCasts());
-          }))
-        return false;
+      // SROA can't speculate lifetime markers of selects/phis, and the
+      // backend may handle such lifetimes incorrectly as well (#104776).
+      // Don't sink lifetimes if it would introduce a phi on the pointer
+      // argument.
       if (isLifeTimeMarker(I0) && OI == 1 &&
           any_of(Insts, [](const Instruction *I) {
             return isa<AllocaInst>(I->getOperand(1)->stripPointerCasts());