[PATCH] D18094: [x86, AVX] replace masked load with full vector load when possible

Sanjay Patel via llvm-commits llvm-commits at lists.llvm.org
Mon Mar 14 09:25:39 PDT 2016

spatel marked an inline comment as done.

Comment at: lib/Target/X86/X86ISelLowering.cpp:27371
@@ -27349,4 +27370,3 @@
   // cause an infinite loop because that's what we're about to create.
-  if (!ISD::isBuildVectorOfConstantSDNodes(ML->getMask().getNode()) ||
-      ML->getSrc0().getOpcode() == ISD::UNDEF)
+  if (ML->getSrc0().getOpcode() == ISD::UNDEF)
     return SDValue();
RKSimon wrote:
> if (ML->getSrc0().isUndef())
Good catch. I cut and pasted that. Looks like a mass edit is needed - a grep shows 194 of these scattered around the code base.


More information about the llvm-commits mailing list