Merge pull request #325 from undingen/llvm_rebase

Rebase llvm to r230300

CMakeLists.txt

@@ -91,7 +91,7 @@ endif()
 add_subdirectory(${DEPS_DIR}/llvm-trunk ${CMAKE_BINARY_DIR}/llvm EXCLUDE_FROM_ALL)
 set(CMAKE_MODULE_PATH "${CMAKE_BINARY_DIR}/llvm/share/llvm/cmake/")
 include(LLVMConfig)
-llvm_map_components_to_libnames(LLVM_LIBS core mcjit native bitreader ipo irreader debuginfo instrumentation ${INTEL_JIT_EVENTS_LIB})
+llvm_map_components_to_libnames(LLVM_LIBS core mcjit native bitreader ipo irreader debuginfodwarf instrumentation ${INTEL_JIT_EVENTS_LIB})
 
 # libunwind
 if("${CMAKE_BUILD_TYPE}" STREQUAL "Debug")

Makefile

@@ -94,7 +94,7 @@ else
 	LLVM_BIN := $(LLVM_BUILD)/Release/bin
 endif
 
-LLVM_LINK_LIBS := core mcjit native bitreader ipo irreader debuginfo instrumentation
+LLVM_LINK_LIBS := core mcjit native bitreader ipo irreader debuginfodwarf instrumentation
 ifneq ($(ENABLE_INTEL_JIT_EVENTS),0)
 LLVM_LINK_LIBS += inteljitevents
 endif

llvm_patches/0008-Add-llvm.experimental.patchpoint.double.patch

-From 3b888292dab18580ad6f234fd6bc4b92a7f54c46 Mon Sep 17 00:00:00 2001
+From b1fd8c81830c14e712b3c2da5379a506a11dc59f Mon Sep 17 00:00:00 2001
 From: Kevin Modzelewski <kevmod@gmail.com>
 Date: Thu, 9 Oct 2014 04:37:33 +0000
 Subject: [PATCH] Add llvm.experimental.patchpoint.double
@@ -7,18 +7,18 @@ Differential Revision: http://reviews.llvm.org/D5696
 ---
  docs/StackMaps.rst                                |  3 +++
  include/llvm/IR/Intrinsics.td                     |  5 +++++
- lib/CodeGen/SelectionDAG/FastISel.cpp             |    8 ++++++-
+ lib/CodeGen/SelectionDAG/FastISel.cpp             |  8 +++++++-
  lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp  |  4 +++-
  lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp     |  1 +
- lib/IR/Verifier.cpp                               |    3 ++-
+ lib/IR/Verifier.cpp                               |  1 +
  lib/Target/AArch64/AArch64TargetTransformInfo.cpp |  1 +
  lib/Target/X86/X86TargetTransformInfo.cpp         |  1 +
- test/CodeGen/X86/anyregcc.ll                      |   24 +++++++++++++++++++--
- test/CodeGen/X86/patchpoint.ll                    |   18 ++++++++++++++++
- 10 files changed, 63 insertions(+), 5 deletions(-)
+ test/CodeGen/X86/anyregcc.ll                      | 24 +++++++++++++++++++++--
+ test/CodeGen/X86/patchpoint.ll                    | 18 +++++++++++++++++
+ 10 files changed, 62 insertions(+), 4 deletions(-)
 
 diff --git a/docs/StackMaps.rst b/docs/StackMaps.rst
-index bd0fb94..5f6d83b 100644
+index 43c60c9..c900c97 100644
 --- a/docs/StackMaps.rst
 +++ b/docs/StackMaps.rst
 @@ -192,6 +192,9 @@ Syntax:
@@ -32,10 +32,10 @@ index bd0fb94..5f6d83b 100644
  Overview:
  """""""""
 diff --git a/include/llvm/IR/Intrinsics.td b/include/llvm/IR/Intrinsics.td
-index 1b9339a..45eecca 100644
+index e89e65d..1603067 100644
 --- a/include/llvm/IR/Intrinsics.td
 +++ b/include/llvm/IR/Intrinsics.td
-@@ -490,4 +490,9 @@ def int_experimental_patchpoint_i64 : Intrinsic<[llvm_i64_ty],
+@@ -514,6 +514,11 @@ def int_experimental_patchpoint_i64 : Intrinsic<[llvm_i64_ty],
                                                   llvm_ptr_ty, llvm_i32_ty,
                                                   llvm_vararg_ty],
                                                   [Throws]>;
@@ -45,11 +45,13 @@ index 1b9339a..45eecca 100644
 +                                                 llvm_vararg_ty],
 +                                                 [Throws]>;
  
+ 
+ //===------------------------ Garbage Collection Intrinsics ---------------===//
 diff --git a/lib/CodeGen/SelectionDAG/FastISel.cpp b/lib/CodeGen/SelectionDAG/FastISel.cpp
-index 72390a9..b624a63 100644
+index 1df4a1d..54af94b 100644
 --- a/lib/CodeGen/SelectionDAG/FastISel.cpp
 +++ b/lib/CodeGen/SelectionDAG/FastISel.cpp
-@@ -738,7 +738,12 @@ bool FastISel::selectPatchpoint(const CallInst *I) {
+@@ -739,7 +739,12 @@ bool FastISel::selectPatchpoint(const CallInst *I) {
    // Add an explicit result reg if we use the anyreg calling convention.
    if (IsAnyRegCC && HasDef) {
      assert(CLI.NumResultRegs == 0 && "Unexpected result register.");
@@ -63,7 +65,7 @@ index 72390a9..b624a63 100644
      CLI.NumResultRegs = 1;
      Ops.push_back(MachineOperand::CreateReg(CLI.ResultReg, /*IsDef=*/true));
    }
-@@ -1205,6 +1210,7 @@ bool FastISel::selectIntrinsicCall(const IntrinsicInst *II) {
+@@ -1206,6 +1211,7 @@ bool FastISel::selectIntrinsicCall(const IntrinsicInst *II) {
      return selectStackmap(II);
    case Intrinsic::experimental_patchpoint_void:
    case Intrinsic::experimental_patchpoint_i64:
@@ -72,10 +74,10 @@ index 72390a9..b624a63 100644
    }
  
 diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
-index a63c90a..dc45bf7 100644
+index 29c0360..c0a8299 100644
 --- a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
 +++ b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
-@@ -2026,6 +2026,7 @@ void SelectionDAGBuilder::visitInvoke(const InvokeInst &I) {
+@@ -2025,6 +2025,7 @@ void SelectionDAGBuilder::visitInvoke(const InvokeInst &I) {
        break;
      case Intrinsic::experimental_patchpoint_void:
      case Intrinsic::experimental_patchpoint_i64:
@@ -83,7 +85,7 @@ index a63c90a..dc45bf7 100644
        visitPatchpoint(&I, LandingPad);
        break;
      }
-@@ -5536,7 +5537,8 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
+@@ -5596,7 +5597,8 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
      return nullptr;
    }
    case Intrinsic::experimental_patchpoint_void:
@@ -94,10 +96,10 @@ index a63c90a..dc45bf7 100644
      return nullptr;
    }
 diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
-index 31f4b06..ac0e635 100644
+index 5e867cf..cb3f995 100644
 --- a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
 +++ b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
-@@ -1014,6 +1014,7 @@ static void collectFailStats(const Instruction *I) {
+@@ -1101,6 +1101,7 @@ static void collectFailStats(const Instruction *I) {
          NumFastIselFailStackMap++; return;
        case Intrinsic::experimental_patchpoint_void: // fall-through
        case Intrinsic::experimental_patchpoint_i64:
@@ -106,42 +108,40 @@ index 31f4b06..ac0e635 100644
        }
      }
 diff --git a/lib/IR/Verifier.cpp b/lib/IR/Verifier.cpp
-index 5519cac..cec9229 100644
+index d01e138..be969eb 100644
 --- a/lib/IR/Verifier.cpp
 +++ b/lib/IR/Verifier.cpp
-@@ -2218,7 +2218,8 @@ void Verifier::visitInstruction(Instruction &I) {
-       Assert1(!F->isIntrinsic() || isa<CallInst>(I) ||
+@@ -2487,6 +2487,7 @@ void Verifier::visitInstruction(Instruction &I) {
                F->getIntrinsicID() == Intrinsic::donothing ||
                F->getIntrinsicID() == Intrinsic::experimental_patchpoint_void ||
--              F->getIntrinsicID() == Intrinsic::experimental_patchpoint_i64,
-+              F->getIntrinsicID() == Intrinsic::experimental_patchpoint_i64 ||
-+              F->getIntrinsicID() == Intrinsic::experimental_patchpoint_double,
+               F->getIntrinsicID() == Intrinsic::experimental_patchpoint_i64 ||
++              F->getIntrinsicID() == Intrinsic::experimental_patchpoint_double ||
+               F->getIntrinsicID() == Intrinsic::experimental_gc_statepoint,
                "Cannot invoke an intrinsinc other than"
                " donothing or patchpoint", &I);
-       Assert1(F->getParent() == M, "Referencing function in another module!",
 diff --git a/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
-index dbdf199..17ea7d0 100644
+index 0646d85..40f2c85 100644
 --- a/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
 +++ b/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
-@@ -280,6 +280,7 @@ unsigned AArch64TTI::getIntImmCost(Intrinsic::ID IID, unsigned Idx,
+@@ -159,6 +159,7 @@ unsigned AArch64TTIImpl::getIntImmCost(Intrinsic::ID IID, unsigned Idx,
      break;
    case Intrinsic::experimental_patchpoint_void:
    case Intrinsic::experimental_patchpoint_i64:
 +  case Intrinsic::experimental_patchpoint_double:
      if ((Idx < 4) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue())))
-       return TCC_Free;
+       return TTI::TCC_Free;
      break;
 diff --git a/lib/Target/X86/X86TargetTransformInfo.cpp b/lib/Target/X86/X86TargetTransformInfo.cpp
-index 531e035..c274edf 100644
+index 5136619..9f455eb 100644
 --- a/lib/Target/X86/X86TargetTransformInfo.cpp
 +++ b/lib/Target/X86/X86TargetTransformInfo.cpp
-@@ -1141,6 +1141,7 @@ unsigned X86TTI::getIntImmCost(Intrinsic::ID IID, unsigned Idx,
+@@ -1102,6 +1102,7 @@ unsigned X86TTIImpl::getIntImmCost(Intrinsic::ID IID, unsigned Idx,
      break;
    case Intrinsic::experimental_patchpoint_void:
    case Intrinsic::experimental_patchpoint_i64:
 +  case Intrinsic::experimental_patchpoint_double:
      if ((Idx < 4) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue())))
-       return TCC_Free;
+       return TTI::TCC_Free;
      break;
 diff --git a/test/CodeGen/X86/anyregcc.ll b/test/CodeGen/X86/anyregcc.ll
 index 98ba17c..b81a3ad 100644
@@ -233,5 +233,5 @@ index 07148f0..7e2a340 100644
  declare i64 @llvm.experimental.patchpoint.i64(i64, i32, i8*, i32, ...)
 +declare double @llvm.experimental.patchpoint.double(i64, i32, i8*, i32, ...)
 -- 
-1.7.9.5
+2.1.0
 

llvm_revision.txt

-225000
+230300

src/codegen/irgen.cpp

@@ -23,7 +23,11 @@
 #include "llvm/IR/DIBuilder.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Verifier.h"
+#if LLVMREV < 229094
 #include "llvm/PassManager.h"
+#else
+#include "llvm/IR/LegacyPassManager.h"
+#endif
 #include "llvm/Support/FileSystem.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
@@ -71,7 +75,12 @@ static void optimizeIR(llvm::Function* f, EffortLevel effort) {
 
     Timer _t("optimizing");
 
+#if LLVMREV < 229094
     llvm::FunctionPassManager fpm(g.cur_module);
+#else
+    llvm::legacy::FunctionPassManager fpm(g.cur_module);
+#endif
+
 
 #if LLVMREV < 217548
     fpm.add(new llvm::DataLayoutPass(*g.tm->getDataLayout()));
@@ -898,8 +907,10 @@ CompiledFunction* doCompile(SourceInfo* source, ParamNames* param_names, const O
     g.cur_module = new llvm::Module(name, g.context);
 #if LLVMREV < 217070 // not sure if this is the right rev
     g.cur_module->setDataLayout(g.tm->getDataLayout()->getStringRepresentation());
-#else
+#elif LLVMREV < 227113
     g.cur_module->setDataLayout(g.tm->getSubtargetImpl()->getDataLayout());
+#else
+    g.cur_module->setDataLayout(g.tm->getDataLayout());
 #endif
     // g.engine->addModule(g.cur_module);
 

src/codegen/irgen/irgenerator.cpp

@@ -501,7 +501,7 @@ private:
                 ConcreteCompilerVariable* converted_obj = obj->makeConverted(emitter, obj->getBoxType());
                 obj->decvref(emitter);
 
-                llvm::Value* v = emitter.createCall(unw_info, g.funcs.getPystonIter, { converted_obj->getValue() });
+                llvm::Value* v = emitter.createCall(unw_info, g.funcs.getPystonIter, converted_obj->getValue());
                 assert(v->getType() == g.llvm_value_type_ptr);
 
                 return new ConcreteCompilerVariable(UNKNOWN, v, true);
@@ -1647,7 +1647,7 @@ private:
             // end code for handling of softspace
 
 
-            llvm::Value* v = emitter.createCall(unw_info, g.funcs.str, { converted->getValue() });
+            llvm::Value* v = emitter.createCall(unw_info, g.funcs.str, converted->getValue());
             v = emitter.getBuilder()->CreateBitCast(v, g.llvm_value_type_ptr);
             auto s = new ConcreteCompilerVariable(STR, v, true);
             r = dest->callattr(emitter, getOpInfoForNode(node, unw_info), &write_str, false, ArgPassSpec(1), { s },

src/codegen/opt/const_classes.cpp

@@ -99,8 +99,10 @@ private:
             APInt ap_offset(64, 0, true);
 #if LLVMREV < 214781
             bool success = gep->accumulateConstantOffset(*g.tm->getDataLayout(), ap_offset);
-#else
+#elif LLVMREV < 227113
             bool success = gep->accumulateConstantOffset(*g.tm->getSubtargetImpl()->getDataLayout(), ap_offset);
+#else
+            bool success = gep->accumulateConstantOffset(*g.tm->getDataLayout(), ap_offset);
 #endif
             assert(success);
             int64_t offset = ap_offset.getSExtValue();

src/codegen/opt/inliner.cpp

@@ -22,7 +22,11 @@
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Verifier.h"
+#if LLVMREV < 229094
 #include "llvm/PassManager.h"
+#else
+#include "llvm/IR/LegacyPassManager.h"
+#endif
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/IPO.h"
 #include "llvm/Transforms/IPO/PassManagerBuilder.h"
@@ -99,8 +103,11 @@ public:
         if (!initialized) {
             llvm::initializeInlineCostAnalysisPass(*llvm::PassRegistry::getPassRegistry());
             llvm::initializeSimpleInlinerPass(*llvm::PassRegistry::getPassRegistry());
+#if LLVMREV < 227669
             llvm::initializeTargetTransformInfoAnalysisGroup(*llvm::PassRegistry::getPassRegistry());
-
+#else
+            llvm::initializeTargetTransformInfoWrapperPassPass(*llvm::PassRegistry::getPassRegistry());
+#endif
             fake_module = new llvm::Module("fake", g.context);
 
             initialized = true;
@@ -119,7 +126,11 @@ public:
 
         llvm::Module* cur_module = f.getParent();
 
+#if LLVMREV < 217548
         llvm::PassManager fake_pm;
+#else
+        llvm::legacy::PassManager fake_pm;
+#endif
         llvm::InlineCostAnalysis* cost_analysis = new llvm::InlineCostAnalysis();
         fake_pm.add(cost_analysis);
         // llvm::errs() << "doing fake run\n";

src/codegen/runtime_hooks.cpp

@@ -20,7 +20,11 @@
 #include "llvm/Analysis/Passes.h"
 #include "llvm/ExecutionEngine/ExecutionEngine.h"
 #include "llvm/IR/Module.h"
+#if LLVMREV < 229094
 #include "llvm/PassManager.h"
+#else
+#include "llvm/IR/LegacyPassManager.h"
+#endif
 #include "llvm/Support/DynamicLibrary.h"
 #include "llvm/Support/TargetSelect.h"
 #include "llvm/Transforms/Scalar.h"

src/codegen/unwinding.cpp

@@ -18,7 +18,11 @@
 #include <sys/types.h>
 #include <unistd.h>
 
+#if LLVMREV < 227586
 #include "llvm/DebugInfo/DIContext.h"
+#else
+#include "llvm/DebugInfo/DWARF/DIContext.h"
+#endif
 #include "llvm/ExecutionEngine/JITEventListener.h"
 #include "llvm/IR/DebugInfo.h"
 #include "llvm/Object/ObjectFile.h"

src/runtime/dict.cpp

@@ -257,7 +257,7 @@ extern "C" int PyDict_Next(PyObject* op, Py_ssize_t* ppos, PyObject** pkey, PyOb
     // Clients are supposed to zero-initialize *ppos:
     if (*it_ptr == NULL) {
         *it_ptr = (iterator*)malloc(sizeof(iterator));
-        ** it_ptr = self->d.begin();
+        **it_ptr = self->d.begin();
     }
 
     iterator* it = *it_ptr;

src/runtime/stacktrace.cpp

@@ -16,8 +16,6 @@
 #include <cstdarg>
 #include <dlfcn.h>
 
-#include "llvm/DebugInfo/DIContext.h"
-
 #include "codegen/unwinding.h"
 #include "core/options.h"
 #include "gc/collector.h"

tools/publicize.cpp

@@ -89,7 +89,23 @@ bool makeVisible(llvm::GlobalValue* gv) {
     if (visibility == llvm::GlobalValue::HiddenVisibility) {
         gv->setVisibility(llvm::GlobalValue::ProtectedVisibility);
         //gv->setDLLStorageClass(llvm::GlobalValue::DLLExportStorageClass);
-        gv->setName(gv->getName() + "_protected");
+        std::string old_name = gv->getName();
+        std::string new_name = (gv->getName() + "_protected").str();
+        gv->setName(new_name);
+
+#if LLVMREV >= 225705
+        // If this symbol has comdat set we have to remove the comdat for the old name
+        // and create a new comdat with the new name
+        if (gv->hasComdat() && isa<llvm::GlobalObject>(gv)) {
+            llvm::Module* mod = gv->getParent();
+            llvm::GlobalObject* go = cast<llvm::GlobalObject>(gv);
+            llvm::Comdat* new_com = mod->getOrInsertComdat(new_name);
+            new_com->setSelectionKind(go->getComdat()->getSelectionKind());
+            go->setComdat(new_com);
+            auto& Comdats = mod->getComdatSymbolTable();
+            Comdats.erase(Comdats.find(old_name));
+        }
+#endif
 
         changed = true;
     }