From 531f025361555e7a695eb559ec02645c054ee146 Mon Sep 17 00:00:00 2001
From: Michael Liao <michael.liao@intel.com>
Date: Wed, 23 Oct 2013 18:32:43 +0000
Subject: [PATCH] Fix PR17631

- Skip instructions added in prolog. For specific targets, prolog may
  insert helper function calls (e.g. _chkstk will be called when
  there're more than 4K bytes allocated on stack). However, these
  helpers don't use/def YMM/XMM registers.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193261 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/X86/X86VZeroUpper.cpp | 11 ++++++++++-
 test/CodeGen/X86/pr17631.ll      | 22 ++++++++++++++++++++++
 2 files changed, 32 insertions(+), 1 deletion(-)
 create mode 100644 test/CodeGen/X86/pr17631.ll

diff --git a/lib/Target/X86/X86VZeroUpper.cpp b/lib/Target/X86/X86VZeroUpper.cpp
index 477f75afef2..0d37a7d0e67 100644
--- a/lib/Target/X86/X86VZeroUpper.cpp
+++ b/lib/Target/X86/X86VZeroUpper.cpp
@@ -231,8 +231,17 @@ bool VZeroUpperInserter::processBasicBlock(MachineFunction &MF,
   bool BBHasCall = false;
 
   for (MachineBasicBlock::iterator I = BB.begin(); I != BB.end(); ++I) {
-    MachineInstr *MI = I;
     DebugLoc dl = I->getDebugLoc();
+    MachineInstr *MI = I;
+
+    // Don't need to check instructions added in prolog.
+    // In prolog, special function calls may be added for specific targets
+    // (e.g. on Windows, a prolog helper '_chkstk' is called when the local
+    // variables exceed 4K bytes on stack.) These helpers won't use/def YMM/XMM
+    // registers.
+    if (MI->getFlag(MachineInstr::FrameSetup))
+      continue;
+
     bool isControlFlow = MI->isCall() || MI->isReturn();
 
     // Shortcut: don't need to check regular instructions in dirty state.
diff --git a/test/CodeGen/X86/pr17631.ll b/test/CodeGen/X86/pr17631.ll
new file mode 100644
index 00000000000..a572ff2e3b3
--- /dev/null
+++ b/test/CodeGen/X86/pr17631.ll
@@ -0,0 +1,22 @@
+; RUN: llc < %s -mcpu=core-avx-i -mtriple=i386-pc-win32 | FileCheck %s
+ 
+%struct_type = type { [64 x <8 x float>], <8 x float> }
+ 
+; Function Attrs: nounwind readnone
+declare i32 @llvm.x86.avx.movmsk.ps.256(<8 x float>)
+ 
+; Function Attrs: nounwind
+define i32 @equal(<8 x i32> %A) {
+allocas:
+  %first_alloc  = alloca [64 x <8 x i32>]
+  %second_alloc = alloca %struct_type
+ 
+  %A1 = bitcast <8 x i32> %A to <8 x float>
+  %A2 = call i32 @llvm.x86.avx.movmsk.ps.256(<8 x float> %A1)
+  ret i32 %A2
+}
+
+; CHECK: equal
+; CHECK-NOT: vzeroupper
+; CHECK: _chkstk
+; CHECK: ret