[PATCH] D43833: [InstCombine] Split the FP constant code out of lookThroughFPExtensions and use nullptr as a sentinel.

Craig Topper via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Feb 27 14:29:27 PST 2018


craig.topper created this revision.
craig.topper added a reviewer: spatel.

[InstCombine] Split the FP constant shrinking code out of lookThroughFPExtensions and turn it into a helper function that returns nullptr on failure.

Currently this code's control flow very much assumes that there are no meaningful checks after determining that it's a ConstantFP. So whenever it wants to stop it just does "return V".

This patch just moves all the code into a helper function and returns nullptr when it wants to stop.

I've split this from https://reviews.llvm.org/D43774 while I try to figure out how to best handle the vector case there. But this change by itself at least seemed like a readability improvement.


https://reviews.llvm.org/D43833

Files:
  lib/Transforms/InstCombine/InstCombineCasts.cpp


Index: lib/Transforms/InstCombine/InstCombineCasts.cpp
===================================================================
--- lib/Transforms/InstCombine/InstCombineCasts.cpp
+++ lib/Transforms/InstCombine/InstCombineCasts.cpp
@@ -1420,29 +1420,34 @@
   return nullptr;
 }
 
+static Constant *shrinkFPConstant(ConstantFP *CFP) {
+  if (CFP->getType() == Type::getPPC_FP128Ty(CFP->getContext()))
+    return nullptr;  // No constant folding of this.
+  // See if the value can be truncated to half and then reextended.
+  if (Constant *NewCFP = fitsInFPType(CFP, APFloat::IEEEhalf()))
+    return NewCFP;
+  // See if the value can be truncated to float and then reextended.
+  if (Constant *NewCFP = fitsInFPType(CFP, APFloat::IEEEsingle()))
+    return NewCFP;
+  if (CFP->getType()->isDoubleTy())
+    return nullptr;  // Won't shrink.
+  if (Constant *NewCFP = fitsInFPType(CFP, APFloat::IEEEdouble()))
+    return NewCFP;
+  // Don't try to shrink to various long double types.
+  return nullptr;
+}
+
 /// Look through floating-point extensions until we get the source value.
 static Value *lookThroughFPExtensions(Value *V) {
   while (auto *FPExt = dyn_cast<FPExtInst>(V))
     V = FPExt->getOperand(0);
 
   // If this value is a constant, return the constant in the smallest FP type
   // that can accurately represent it.  This allows us to turn
   // (float)((double)X+2.0) into x+2.0f.
-  if (auto *CFP = dyn_cast<ConstantFP>(V)) {
-    if (CFP->getType() == Type::getPPC_FP128Ty(V->getContext()))
-      return V;  // No constant folding of this.
-    // See if the value can be truncated to half and then reextended.
-    if (Value *V = fitsInFPType(CFP, APFloat::IEEEhalf()))
-      return V;
-    // See if the value can be truncated to float and then reextended.
-    if (Value *V = fitsInFPType(CFP, APFloat::IEEEsingle()))
-      return V;
-    if (CFP->getType()->isDoubleTy())
-      return V;  // Won't shrink.
-    if (Value *V = fitsInFPType(CFP, APFloat::IEEEdouble()))
-      return V;
-    // Don't try to shrink to various long double types.
-  }
+  if (auto *CFP = dyn_cast<ConstantFP>(V))
+    if (Constant *NewCFP = shrinkFPConstant(CFP))
+      return NewCFP;
 
   return V;
 }


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D43833.136158.patch
Type: text/x-patch
Size: 2210 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180227/c2cb785f/attachment.bin>


More information about the llvm-commits mailing list