[llvm] r244499 - fix minsize detection: minsize attribute implies optimizing for size

[Sanjay Patel via llvm-commits]

Author: spatel
Date: Mon Aug 10 15:45:44 2015
New Revision: 244499

URL: http://llvm.org/viewvc/llvm-project?rev=244499&view=rev
Log:
fix minsize detection: minsize attribute implies optimizing for size

Modified:
    llvm/trunk/lib/Target/X86/X86InstrInfo.cpp
    llvm/trunk/test/CodeGen/X86/sse_partial_update.ll

Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=244499&r1=244498&r2=244499&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Mon Aug 10 15:45:44 2015
@@ -5347,11 +5347,8 @@ MachineInstr *X86InstrInfo::foldMemoryOp
   // Check switch flag
   if (NoFusing) return nullptr;
 
-  // Unless optimizing for size, don't fold to avoid partial
-  // register update stalls
-  // FIXME: Use Function::optForSize().
-  if (!MF.getFunction()->hasFnAttribute(Attribute::OptimizeForSize) &&
-      hasPartialRegUpdate(MI->getOpcode()))
+  // Avoid partial register update stalls unless optimizing for size.
+  if (!MF.getFunction()->optForSize() && hasPartialRegUpdate(MI->getOpcode()))
     return nullptr;
 
   // Determine the alignment of the load.

Modified: llvm/trunk/test/CodeGen/X86/sse_partial_update.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sse_partial_update.ll?rev=244499&r1=244498&r2=244499&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sse_partial_update.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sse_partial_update.ll Mon Aug 10 15:45:44 2015
@@ -113,9 +113,8 @@ define <2 x double> @load_fold_cvtss2sd_
 
 define <2 x double> @load_fold_cvtss2sd_int_minsize(<4 x float> *%a) minsize {
 ; CHECK-LABEL: load_fold_cvtss2sd_int_minsize:
-; CHECK:       movaps   (%rdi), %xmm1
-; CHECK-NEXT:  xorps    %xmm0, %xmm0
-; CHECK-NEXT:  cvtss2sd %xmm1, %xmm0
+; CHECK:       xorps    %xmm0, %xmm0
+; CHECK-NEXT:  cvtss2sd (%rdi), %xmm0
 ; CHECK-NEXT:  retq
   %ld = load <4 x float>, <4 x float> *%a
   %x = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> <double 0x0, double 0x0>, <4 x float> %ld)