[llvm] [CGP]: Optimize mul.overflow. (PR #148343)

[Hassnaa Hamdi via llvm-commits]

================
@@ -6389,6 +6395,184 @@ bool CodeGenPrepare::optimizeGatherScatterInst(Instruction *MemoryInst,
   return true;
 }
 
+// This is a helper for CodeGenPrepare::optimizeMulWithOverflow.
+// Check the pattern we are interested in where there are maximum 2 uses
+// of the intrinsic which are the extract instructions.
+static bool matchOverflowPattern(Instruction *&I, ExtractValueInst *&MulExtract,
+                                 ExtractValueInst *&OverflowExtract) {
+  if (I->getNumUses() > 2)
+    return false;
+
+  for (User *U : I->users()) {
+    auto *Extract = dyn_cast<ExtractValueInst>(U);
+    if (!Extract || Extract->getNumIndices() != 1)
+      return false;
+
+    unsigned Index = Extract->getIndices()[0];
+    if (Index == 0)
+      MulExtract = Extract;
+    else if (Index == 1)
+      OverflowExtract = Extract;
+    else
+      return false;
+  }
+  return true;
+}
+
+// Rewrite the mul_with_overflow intrinsic by checking if both of the
+// operands' value ranges are within the legal type. If so, we can optimize the
+// multiplication algorithm. This code is supposed to be written during the step
+// of type legalization, but given that we need to reconstruct the IR which is
+// not doable there, we do it here.
+// The IR after the optimization will look like:
+// entry:
+//   if signed:
+//     ( (lhs_lo>>BW-1) ^ lhs_hi) || ( (rhs_lo>>BW-1) ^ rhs_hi) ? overflow,
+//     overflow_no
+//   else:
+//     (lhs_hi != 0) || (rhs_hi != 0) ? overflow, overflow_no
+// overflow_no:
+// overflow:
+// overflow.res:
+// \returns true if optimization was applied
+// TODO: This optimization can be further improved to optimize branching on
+// overflow where the 'overflow_no' BB can branch directly to the false
+// successor of overflow, but that would add additional complexity so we leave
+// it for future work.
+bool CodeGenPrepare::optimizeMulWithOverflow(Instruction *I, bool IsSigned,
+                                             ModifyDT &ModifiedDT) {
+  // Check if target supports this optimization.
+  if (!TLI->shouldOptimizeMulOverflowWithZeroHighBits(
+          I->getContext(),
+          TLI->getValueType(*DL, I->getType()->getContainedType(0))))
+    return false;
+
+  ExtractValueInst *MulExtract = nullptr, *OverflowExtract = nullptr;
+  if (!matchOverflowPattern(I, MulExtract, OverflowExtract))
+    return false;
+
+  // Keep track of the instruction to stop reoptimizing it again.
+  InsertedInsts.insert(I);
+
+  Value *LHS = I->getOperand(0);
+  Value *RHS = I->getOperand(1);
+  Type *Ty = LHS->getType();
+  unsigned VTHalfBitWidth = Ty->getScalarSizeInBits() / 2;
+  Type *LegalTy = Ty->getWithNewBitWidth(VTHalfBitWidth);
+
+  // New BBs:
+  std::string OriginalBlockName = I->getParent()->getName().str();
----------------
hassnaaHamdi wrote:

I didn't know this. I use it, thanks.

https://github.com/llvm/llvm-project/pull/148343