From: Craig Topper <craig.topper@intel.com>
Date: Sat, 28 Jul 2018 18:21:46 +0000 (+0000)
Subject: [X86] Correct the immediate cost for 'add/sub i64 %x, 0x80000000'.
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=e192af49a170ac54aadc8cb75b9e88ac2d54e937;p=llvm

[X86] Correct the immediate cost for 'add/sub i64 %x, 0x80000000'.

X86 normally requires immediates to be a signed 32-bit value which would exclude i64 0x80000000. But for add/sub we can negate the constant and use the opposite instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338204 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/X86/X86TargetTransformInfo.cpp b/lib/Target/X86/X86TargetTransformInfo.cpp
index 1b260d8d45d..0257c42def2 100644
--- a/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -2332,9 +2332,15 @@ int X86TTIImpl::getIntImmCost(unsigned Opcode, unsigned Idx, const APInt &Imm,
     // immediates here as the normal path expects bit 31 to be sign extended.
     if (Idx == 1 && Imm.getBitWidth() == 64 && isUInt<32>(Imm.getZExtValue()))
       return TTI::TCC_Free;
-    LLVM_FALLTHROUGH;
+    ImmIdx = 1;
+    break;
   case Instruction::Add:
   case Instruction::Sub:
+    // For add/sub, we can use the opposite instruction for INT32_MIN.
+    if (Idx == 1 && Imm.getBitWidth() == 64 && isInt<32>(-Imm.getSExtValue()))
+      return TTI::TCC_Free;
+    ImmIdx = 1;
+    break;
   case Instruction::Mul:
   case Instruction::UDiv:
   case Instruction::SDiv: