x86_64: Fix calls to __morestack under the large code model.

Under the large code model, we cannot assume that __morestack lives within
2^31 bytes of the call site, so we cannot use pc-relative addressing. We
cannot perform the call via a temporary register, as the rax register may
be used to store the static chain, and all other suitable registers may be
either callee-save or used for parameter passing. We cannot use the stack
at this point either because __morestack manipulates the stack directly.

To avoid these issues, perform an indirect call via a read-only memory
location containing the address.

This solution is not perfect, as it assumes that the .rodata section
is laid out within 2^31 bytes of each function body, but this seems to
be sufficient for JIT.

Differential Revision: http://reviews.llvm.org/D6787

llvm-svn: 225003

5 files changed, 78 insertions, 7 deletions

diff --git a/llvm/include/llvm/CodeGen/MachineModuleInfo.h b/llvm/include/llvm/CodeGen/MachineModuleInfo.h
index 287b3ebb0e2..584ce65c9d1 100644
--- a/llvm/include/llvm/CodeGen/MachineModuleInfo.h
+++ b/llvm/include/llvm/CodeGen/MachineModuleInfo.h
@@ -161,6 +161,13 @@ class MachineModuleInfo : public ImmutablePass {
   /// to _fltused on Windows targets.
   bool UsesVAFloatArgument;
 
+  /// UsesMorestackAddr - True if the module calls the __morestack function
+  /// indirectly, as is required under the large code model on x86. This is used
+  /// to emit a definition of a symbol, __morestack_addr, containing the
+  /// address. See comments in lib/Target/X86/X86FrameLowering.cpp for more
+  /// details.
+  bool UsesMorestackAddr;
+
 public:
   static char ID; // Pass identification, replacement for typeid
 
@@ -234,6 +241,14 @@ public:
     UsesVAFloatArgument = b;
   }
 
+  bool usesMorestackAddr() const {
+    return UsesMorestackAddr;
+  }
+
+  void setUsesMorestackAddr(bool b) {
+    UsesMorestackAddr = b;
+  }
+
   /// \brief Returns a reference to a list of cfi instructions in the current
   /// function's prologue.  Used to construct frame maps for debug and exception
   /// handling comsumers.
diff --git a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
index 97ef12c8328..ff8fc60e603 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -1011,6 +1011,23 @@ bool AsmPrinter::doFinalization(Module &M) {
   // Emit llvm.ident metadata in an '.ident' directive.
   EmitModuleIdents(M);
 
+  // Emit __morestack address if needed for indirect calls.
+  if (MMI->usesMorestackAddr()) {
+    const MCSection *ReadOnlySection =
+        getObjFileLowering().getSectionForConstant(SectionKind::getReadOnly(),
+                                                   /*C=*/nullptr);
+    OutStreamer.SwitchSection(ReadOnlySection);
+
+    MCSymbol *AddrSymbol =
+        OutContext.GetOrCreateSymbol(StringRef("__morestack_addr"));
+    OutStreamer.EmitLabel(AddrSymbol);
+
+    const DataLayout &DL = *TM.getSubtargetImpl()->getDataLayout();
+    unsigned PtrSize = DL.getPointerSize(0);
+    OutStreamer.EmitSymbolValue(GetExternalSymbolSymbol("__morestack"),
+                                PtrSize);
+  }
+
   // If we don't have any trampolines, then we don't require stack memory
   // to be executable. Some targets have a directive to declare this.
   Function *InitTrampolineIntrinsic = M.getFunction("llvm.init.trampoline");
diff --git a/llvm/lib/CodeGen/MachineModuleInfo.cpp b/llvm/lib/CodeGen/MachineModuleInfo.cpp
index eb3c0bf224d..baad411e2c5 100644
--- a/llvm/lib/CodeGen/MachineModuleInfo.cpp
+++ b/llvm/lib/CodeGen/MachineModuleInfo.cpp
@@ -273,7 +273,7 @@ bool MachineModuleInfo::doInitialization(Module &M) {
   CurCallSite = 0;
   CallsEHReturn = 0;
   CallsUnwindInit = 0;
-  DbgInfoAvailable = UsesVAFloatArgument = false; 
+  DbgInfoAvailable = UsesVAFloatArgument = UsesMorestackAddr = false;
   // Always emit some info, by default "no personality" info.
   Personalities.push_back(nullptr);
   AddrLabelSymbols = nullptr;
diff --git a/llvm/lib/Target/X86/X86FrameLowering.cpp b/llvm/lib/Target/X86/X86FrameLowering.cpp
index c408fddf890..95bca7d936f 100644
--- a/llvm/lib/Target/X86/X86FrameLowering.cpp
+++ b/llvm/lib/Target/X86/X86FrameLowering.cpp
@@ -1648,12 +1648,36 @@ X86FrameLowering::adjustForSegmentedStacks(MachineFunction &MF) const {
   }
 
   // __morestack is in libgcc
-  if (Is64Bit)
-    BuildMI(allocMBB, DL, TII.get(X86::CALL64pcrel32))
-      .addExternalSymbol("__morestack");
-  else
-    BuildMI(allocMBB, DL, TII.get(X86::CALLpcrel32))
-      .addExternalSymbol("__morestack");
+  if (Is64Bit && MF.getTarget().getCodeModel() == CodeModel::Large) {
+    // Under the large code model, we cannot assume that __morestack lives
+    // within 2^31 bytes of the call site, so we cannot use pc-relative
+    // addressing. We cannot perform the call via a temporary register,
+    // as the rax register may be used to store the static chain, and all
+    // other suitable registers may be either callee-save or used for
+    // parameter passing. We cannot use the stack at this point either
+    // because __morestack manipulates the stack directly.
+    //
+    // To avoid these issues, perform an indirect call via a read-only memory
+    // location containing the address.
+    //
+    // This solution is not perfect, as it assumes that the .rodata section
+    // is laid out within 2^31 bytes of each function body, but this seems
+    // to be sufficient for JIT.
+    BuildMI(allocMBB, DL, TII.get(X86::CALL64m))
+        .addReg(X86::RIP)
+        .addImm(0)
+        .addReg(0)
+        .addExternalSymbol("__morestack_addr")
+        .addReg(0);
+    MF.getMMI().setUsesMorestackAddr(true);
+  } else {
+    if (Is64Bit)
+      BuildMI(allocMBB, DL, TII.get(X86::CALL64pcrel32))
+        .addExternalSymbol("__morestack");
+    else
+      BuildMI(allocMBB, DL, TII.get(X86::CALLpcrel32))
+        .addExternalSymbol("__morestack");
+  }
 
   if (IsNested)
     BuildMI(allocMBB, DL, TII.get(X86::MORESTACK_RET_RESTORE_R10));
diff --git a/llvm/test/CodeGen/X86/segmented-stacks.ll b/llvm/test/CodeGen/X86/segmented-stacks.ll
index 45e157ff7c7..3e47121a380 100644
--- a/llvm/test/CodeGen/X86/segmented-stacks.ll
+++ b/llvm/test/CodeGen/X86/segmented-stacks.ll
@@ -1,5 +1,6 @@
 ; RUN: llc < %s -mcpu=generic -mtriple=i686-linux -verify-machineinstrs | FileCheck %s -check-prefix=X32-Linux
 ; RUN: llc < %s -mcpu=generic -mtriple=x86_64-linux  -verify-machineinstrs | FileCheck %s -check-prefix=X64-Linux
+; RUN: llc < %s -mcpu=generic -mtriple=x86_64-linux -code-model=large -verify-machineinstrs | FileCheck %s -check-prefix=X64-Linux-Large
 ; RUN: llc < %s -mcpu=generic -mtriple=x86_64-linux-gnux32 -verify-machineinstrs | FileCheck %s -check-prefix=X32ABI
 ; RUN: llc < %s -mcpu=generic -mtriple=i686-darwin -verify-machineinstrs | FileCheck %s -check-prefix=X32-Darwin
 ; RUN: llc < %s -mcpu=generic -mtriple=x86_64-darwin -verify-machineinstrs | FileCheck %s -check-prefix=X64-Darwin
@@ -57,6 +58,16 @@ define void @test_basic() #0 {
 ; X64-Linux-NEXT:  callq __morestack
 ; X64-Linux-NEXT:  ret
 
+; X64-Linux-Large-LABEL:       test_basic:
+
+; X64-Linux-Large:       cmpq %fs:112, %rsp
+; X64-Linux-Large-NEXT:  ja      .LBB0_2
+
+; X64-Linux-Large:       movabsq $40, %r10
+; X64-Linux-Large-NEXT:  movabsq $0, %r11
+; X64-Linux-Large-NEXT:  callq *__morestack_addr(%rip)
+; X64-Linux-Large-NEXT:  ret
+
 ; X32ABI-LABEL:       test_basic:
 
 ; X32ABI:       cmpl %fs:64, %esp
@@ -626,3 +637,7 @@ define void @test_nostack() #0 {
 }
 
 attributes #0 = { "split-stack" }
+
+; X64-Linux-Large: .rodata
+; X64-Linux-Large-NEXT: __morestack_addr:
+; X64-Linux-Large-NEXT: .quad	__morestack