[PATCH] D50518: [ARM] Disallow zexts in ARMCodeGenPrepare

[Sam Parker via Phabricator via llvm-commits]

samparker added inline comments.


================
Comment at: lib/Target/ARM/ARMCodeGenPrepare.cpp:517-518
 bool ARMCodeGenPrepare::isLegalToPromote(Value *V) {
-  if (!isSupportedType(V))
-    return false;
-
-  unsigned VSize = 0;
-  if (auto *Ld = dyn_cast<LoadInst>(V)) {
-    auto *PtrTy = cast<PointerType>(Ld->getPointerOperandType());
-    VSize = PtrTy->getElementType()->getPrimitiveSizeInBits();
-  } else if (auto *ZExt = dyn_cast<ZExtInst>(V)) {
-    VSize = ZExt->getOperand(0)->getType()->getPrimitiveSizeInBits();
-  } else {
-    VSize = V->getType()->getPrimitiveSizeInBits();
-  }
-
-  if (VSize > TypeSize)
-    return false;
+  //if (!isSupportedType(V))
+    //return false;
 
----------------
rnk wrote:
> Commented out code not intended for commit?
ah yep.


================
Comment at: test/CodeGen/ARM/arm-cgp-zext-truncs.ll:267-270
+  switch i9 %trunc, label %cleanup.fold.split [
+    i9 0, label %cleanup
+    i9 -256, label %if.then7
+  ]
----------------
rnk wrote:
> I take it this is a reduction of the test case I sent?
yes, its the exact IR just with the function renamed to something half readable.


https://reviews.llvm.org/D50518