[PATCH] D72853: [GlobalISel] Don't arbitrarily limit a mask to 64 bits

[Jay Foad via Phabricator via llvm-commits]

This revision was automatically updated to reflect the committed changes.
Closed by commit rG885260d5d805: [GlobalISel] Don't arbitrarily limit a mask to 64 bits (authored by foad).

Changed prior to commit:
  https://reviews.llvm.org/D72853?vs=238504&id=238510#toc

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72853/new/

https://reviews.llvm.org/D72853

Files:
  llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp


Index: llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
===================================================================
--- llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
+++ llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
@@ -1436,9 +1436,10 @@
     // Do the arithmetic in the larger type.
     auto NewOp = MIRBuilder.buildInstr(Opcode, {WideTy}, {LHSZext, RHSZext});
     LLT OrigTy = MRI.getType(MI.getOperand(0).getReg());
-    APInt Mask = APInt::getAllOnesValue(OrigTy.getSizeInBits());
+    APInt Mask =
+        APInt::getLowBitsSet(WideTy.getSizeInBits(), OrigTy.getSizeInBits());
     auto AndOp = MIRBuilder.buildAnd(
-        WideTy, NewOp, MIRBuilder.buildConstant(WideTy, Mask.getZExtValue()));
+        WideTy, NewOp, MIRBuilder.buildConstant(WideTy, Mask));
     // There is no overflow if the AndOp is the same as NewOp.
     MIRBuilder.buildICmp(CmpInst::ICMP_NE, MI.getOperand(1), NewOp, AndOp);
     // Now trunc the NewOp to the original result.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D72853.238510.patch
Type: text/x-patch
Size: 973 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200116/29b4732f/attachment.bin>