[PATCH] D60256: [InstCombine][X86] Expand MOVMSK to generic IR (PR39927)

[Sanjay Patel via Phabricator via llvm-commits]

spatel added inline comments.


================
Comment at: lib/Transforms/InstCombine/InstCombineCalls.cpp:716
+  // %cmp = icmp slt <16 x i8> %x, zeroinitializer
+  // %int = bitcast <16 x i8> %cmp to i16
+  // %res = zext i16 %int to i32
----------------
<16 x i8> should be <16 x i1> as the source type for the bitcast


================
Comment at: test/Transforms/InstCombine/X86/x86-movmsk.ll:399
 
 ; Negative test - bitcast from scalar.
 
----------------
Stale comment. Could mark this a TODO, but it might cause more harm than good:
  %fourbits = sext i1 %x to i4
  %r = zext i4 %fourbits to i32 


================
Comment at: test/Transforms/InstCombine/X86/x86-movmsk.ll:416
 
 ; Negative test - bitcast from vector type with more elements.
 
----------------
Stale comment


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D60256/new/

https://reviews.llvm.org/D60256