ARM: support stack probe emission for Windows on ARM

This introduces the stack lowering emission of the stack probe function for
Windows on ARM. The stack on Windows on ARM is a dynamically paged stack where
any page allocation which crosses a page boundary of the following guard page
will cause a page fault. This page fault must be handled by the kernel to
ensure that the page is faulted in. If this does not occur and a write access
any memory beyond that, the page fault will go unserviced, resulting in an
abnormal program termination.

The watermark for the stack probe appears to be at 4080 bytes (for
accommodating the stack guard canaries and stack alignment) when SSP is
enabled.  Otherwise, the stack probe is emitted on the page size boundary of
4096 bytes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207615 91177308-0d34-0410-b5e6-96231b3b80d8

3 files changed, 114 insertions, 4 deletions

diff --git a/docs/Extensions.rst b/docs/Extensions.rst
index 7d8c521ec4..9dddab5b5e 100644
--- a/docs/Extensions.rst
+++ b/docs/Extensions.rst
@@ -159,3 +159,34 @@ different COMDATs:
   .globl Symbol2
   Symbol2:
   .long 1
+
+Target Specific Behaviour
+=========================
+
+Windows on ARM
+--------------
+
+Stack Probe Emission
+^^^^^^^^^^^^^^^^^^^^
+
+The reference implementation (Microsoft Visual Studio 2012) emits stack probes
+in the following fashion:
+
+.. code-block:: gas
+
+  movw r4, #constant
+  bl __chkstk
+  sub.w sp, sp, r4
+
+However, this has the limitation of 32 MiB (±16MiB).  In order to accomodate
+larger binaries, LLVM supports the use of ``-mcode-model=large`` to allow a 4GiB
+range via a slight deviation.  It will generate an indirect jump as follows:
+
+.. code-block:: gas
+
+  movw r4, #constant
+  movw r12, :lower16:__chkstk
+  movt r12, :upper16:__chkstk
+  blx r12
+  sub.w sp, sp, r4
+
diff --git a/lib/Target/ARM/ARMFrameLowering.cpp b/lib/Target/ARM/ARMFrameLowering.cpp
index aae7aa45e4..af4c97db68 100644
--- a/lib/Target/ARM/ARMFrameLowering.cpp
+++ b/lib/Target/ARM/ARMFrameLowering.cpp
@@ -25,6 +25,7 @@
 #include "llvm/CodeGen/RegisterScavenging.h"
 #include "llvm/IR/CallingConv.h"
 #include "llvm/IR/Function.h"
+#include "llvm/IR/Module.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetOptions.h"
@@ -142,6 +143,14 @@ static int sizeOfSPAdjustment(const MachineInstr *MI) {
   return count;
 }
 
+static bool WindowsRequiresStackProbe(const MachineFunction &MF,
+                                      size_t StackSizeInBytes) {
+  const MachineFrameInfo *MFI = MF.getFrameInfo();
+  if (MFI->getStackProtectorIndex() > 0)
+    return StackSizeInBytes >= 4080;
+  return StackSizeInBytes >= 4096;
+}
+
 void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   MachineBasicBlock &MBB = MF.front();
   MachineBasicBlock::iterator MBBI = MBB.begin();
@@ -149,15 +158,16 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
   MachineModuleInfo &MMI = MF.getMMI();
   MCContext &Context = MMI.getContext();
+  const TargetMachine &TM = MF.getTarget();
   const MCRegisterInfo *MRI = Context.getRegisterInfo();
   const ARMBaseRegisterInfo *RegInfo =
-    static_cast<const ARMBaseRegisterInfo*>(MF.getTarget().getRegisterInfo());
+    static_cast<const ARMBaseRegisterInfo*>(TM.getRegisterInfo());
   const ARMBaseInstrInfo &TII =
-    *static_cast<const ARMBaseInstrInfo*>(MF.getTarget().getInstrInfo());
+    *static_cast<const ARMBaseInstrInfo*>(TM.getInstrInfo());
   assert(!AFI->isThumb1OnlyFunction() &&
          "This emitPrologue does not support Thumb1!");
   bool isARM = !AFI->isThumbFunction();
-  unsigned Align = MF.getTarget().getFrameLowering()->getStackAlignment();
+  unsigned Align = TM.getFrameLowering()->getStackAlignment();
   unsigned ArgRegsSaveSize = AFI->getArgRegsSaveSize(Align);
   unsigned NumBytes = MFI->getStackSize();
   const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo();
@@ -187,7 +197,8 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
         .addCFIIndex(CFIIndex);
   }
 
-  if (!AFI->hasStackFrame()) {
+  if (!AFI->hasStackFrame() &&
+      (!STI.isTargetWindows() || !WindowsRequiresStackProbe(MF, NumBytes))) {
     if (NumBytes - ArgRegsSaveSize != 0) {
       emitSPUpdate(isARM, MBB, MBBI, dl, TII, -(NumBytes - ArgRegsSaveSize),
                    MachineInstr::FrameSetup);
@@ -284,6 +295,50 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   } else
     NumBytes = DPRCSOffset;
 
+  if (STI.isTargetWindows() && WindowsRequiresStackProbe(MF, NumBytes)) {
+    uint32_t NumWords = NumBytes >> 2;
+
+    if (NumWords < 65536)
+      AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi16), ARM::R4)
+                     .addImm(NumWords));
+    else
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi32imm), ARM::R4)
+        .addImm(NumWords);
+
+    switch (TM.getCodeModel()) {
+    case CodeModel::Small:
+    case CodeModel::Medium:
+    case CodeModel::Default:
+    case CodeModel::Kernel:
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::tBL))
+        .addImm((unsigned)ARMCC::AL).addReg(0)
+        .addExternalSymbol("__chkstk")
+        .addReg(ARM::R4, RegState::Implicit);
+      break;
+    case CodeModel::Large:
+    case CodeModel::JITDefault: {
+      LLVMContext &Ctx = MF.getMMI().getModule()->getContext();
+      const GlobalValue *F =
+        Function::Create(FunctionType::get(Type::getVoidTy(Ctx), false),
+                         GlobalValue::AvailableExternallyLinkage, "__chkstk");
+
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi32imm), ARM::R12)
+        .addGlobalAddress(F);
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::BLX))
+        .addReg(ARM::R12, RegState::Kill)
+        .addReg(ARM::R4, RegState::Implicit);
+      break;
+    }
+    }
+
+    AddDefaultCC(AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::t2SUBrr),
+                                        ARM::SP)
+                                .addReg(ARM::SP, RegState::Define)
+                                .addReg(ARM::R4, RegState::Kill)
+                                .setMIFlags(MachineInstr::FrameSetup)));
+    NumBytes = 0;
+  }
+
   unsigned adjustedGPRCS1Size = GPRCS1Size;
   if (NumBytes) {
     // Adjust SP after all the callee-save spills.
diff --git a/test/CodeGen/ARM/Windows/chkstk.ll b/test/CodeGen/ARM/Windows/chkstk.ll
new file mode 100644
index 0000000000..9c58fa08d3
--- /dev/null
+++ b/test/CodeGen/ARM/Windows/chkstk.ll
@@ -0,0 +1,24 @@
+; RUN: llc -mtriple=thumbv7-windows -mcpu=cortex-a9 %s -o - \
+; RUN:  | FileCheck -check-prefix CHECK-DEFAULT-CODE-MODEL %s
+
+; RUN: llc -mtriple=thumbv7-windows -mcpu=cortex-a9 -code-model=large %s -o - \
+; RUN:  | FileCheck -check-prefix CHECK-LARGE-CODE-MODEL %s
+
+define arm_aapcs_vfpcc void @check_watermark() {
+entry:
+  %buffer = alloca [4096 x i8], align 1
+  ret void
+}
+
+; CHECK-DEFAULT-CODE-MODEL: check_watermark:
+; CHECK-DEFAULT-CODE-MODEL: 	movw r4, #1024
+; CHECK-DEFAULT-CODE-MODEL: 	bl __chkstk
+; CHECK-DEFAULT-CODE-MODEL: 	sub.w sp, sp, r4
+
+; CHECK-LARGE-CODE-MODEL: check_watermark:
+; CHECK-LARGE-CODE-MODEL: 	movw r4, #1024
+; CHECK-LARGE-CODE-MODEL: 	movw r12, :lower16:__chkstk
+; CHECK-LARGE-CODE-MODEL: 	movt r12, :upper16:__chkstk
+; CHECK-LARGE-CODE-MODEL: 	blx r12
+; CHECK-LARGE-CODE-MODEL: 	sub.w sp, sp, r4
+