[ThinLTO] Disable importing and other cross-module optis at -O0

Summary:
There is no point to importing at -O0, since we won't inline. We should
also disable other cross-module optimizations.

(Plan to backport this fix to the 3.9 branch to fix PR30774)

Reviewers: pcc

Subscribers: johanengelen, mehdi_amini

Differential Revision: https://reviews.llvm.org/D25918

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@285648 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/LTO/LTO.cpp b/lib/LTO/LTO.cpp
index 654b86a0dfd57f13352103bfc13e4566963a571a..20da6d9f018cf8a898546135aad7d47a334f6cec 100644 (file)
--- a/lib/LTO/LTO.cpp
+++ b/lib/LTO/LTO.cpp
@@ -770,36 +770,40 @@ Error LTO::runThinLTO(AddStreamFn AddStream, NativeObjectCache Cache,
       ThinLTO.ModuleMap.size());
   StringMap<FunctionImporter::ExportSetTy> ExportLists(
       ThinLTO.ModuleMap.size());
-  ComputeCrossModuleImport(ThinLTO.CombinedIndex, ModuleToDefinedGVSummaries,
-                           ImportLists, ExportLists);
-
-  std::set<GlobalValue::GUID> ExportedGUIDs;
-  for (auto &Res : GlobalResolutions) {
-    if (!Res.second.IRName.empty() &&
-        Res.second.Partition == GlobalResolution::External)
-      ExportedGUIDs.insert(GlobalValue::getGUID(Res.second.IRName));
-  }
+  StringMap<std::map<GlobalValue::GUID, GlobalValue::LinkageTypes>> ResolvedODR;
 
-  auto isPrevailing = [&](GlobalValue::GUID GUID, const GlobalValueSummary *S) {
-    return ThinLTO.PrevailingModuleForGUID[GUID] == S->modulePath();
-  };
-  auto isExported = [&](StringRef ModuleIdentifier, GlobalValue::GUID GUID) {
-    const auto &ExportList = ExportLists.find(ModuleIdentifier);
-    return (ExportList != ExportLists.end() &&
-            ExportList->second.count(GUID)) ||
-           ExportedGUIDs.count(GUID);
-  };
-  thinLTOInternalizeAndPromoteInIndex(ThinLTO.CombinedIndex, isExported);
+  if (Conf.OptLevel > 0) {
+    ComputeCrossModuleImport(ThinLTO.CombinedIndex, ModuleToDefinedGVSummaries,
+                             ImportLists, ExportLists);
 
-  StringMap<std::map<GlobalValue::GUID, GlobalValue::LinkageTypes>> ResolvedODR;
-  auto recordNewLinkage = [&](StringRef ModuleIdentifier,
-                              GlobalValue::GUID GUID,
-                              GlobalValue::LinkageTypes NewLinkage) {
-    ResolvedODR[ModuleIdentifier][GUID] = NewLinkage;
-  };
+    std::set<GlobalValue::GUID> ExportedGUIDs;
+    for (auto &Res : GlobalResolutions) {
+      if (!Res.second.IRName.empty() &&
+          Res.second.Partition == GlobalResolution::External)
+        ExportedGUIDs.insert(GlobalValue::getGUID(Res.second.IRName));
+    }
 
-  thinLTOResolveWeakForLinkerInIndex(ThinLTO.CombinedIndex, isPrevailing,
-                                     recordNewLinkage);
+    auto isPrevailing = [&](GlobalValue::GUID GUID,
+                            const GlobalValueSummary *S) {
+      return ThinLTO.PrevailingModuleForGUID[GUID] == S->modulePath();
+    };
+    auto isExported = [&](StringRef ModuleIdentifier, GlobalValue::GUID GUID) {
+      const auto &ExportList = ExportLists.find(ModuleIdentifier);
+      return (ExportList != ExportLists.end() &&
+              ExportList->second.count(GUID)) ||
+             ExportedGUIDs.count(GUID);
+    };
+    thinLTOInternalizeAndPromoteInIndex(ThinLTO.CombinedIndex, isExported);
+
+    auto recordNewLinkage = [&](StringRef ModuleIdentifier,
+                                GlobalValue::GUID GUID,
+                                GlobalValue::LinkageTypes NewLinkage) {
+      ResolvedODR[ModuleIdentifier][GUID] = NewLinkage;
+    };
+
+    thinLTOResolveWeakForLinkerInIndex(ThinLTO.CombinedIndex, isPrevailing,
+                                       recordNewLinkage);
+  }
 
   std::unique_ptr<ThinBackendProc> BackendProc =
       ThinLTO.Backend(Conf, ThinLTO.CombinedIndex, ModuleToDefinedGVSummaries,

diff --git a/test/ThinLTO/X86/Inputs/funcimport2.ll b/test/ThinLTO/X86/Inputs/funcimport2.ll
new file mode 100644 (file)
index 0000000..216ef0f
--- /dev/null
+++ b/test/ThinLTO/X86/Inputs/funcimport2.ll
@@ -0,0 +1,11 @@
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.11.0"
+
+
+define i32 @main() #0 {
+entry:
+  call void (...) @foo()
+  ret i32 0
+}
+
+declare void @foo(...) #1

diff --git a/test/ThinLTO/X86/funcimport2.ll b/test/ThinLTO/X86/funcimport2.ll
new file mode 100644 (file)
index 0000000..c83370b
--- /dev/null
+++ b/test/ThinLTO/X86/funcimport2.ll
@@ -0,0 +1,28 @@
+; Do setup work for all below tests: generate bitcode and combined index
+; RUN: opt -module-summary %s -o %t1.bc
+; RUN: opt -module-summary %p/Inputs/funcimport2.ll -o %t2.bc
+
+; RUN: llvm-lto2 %t1.bc %t2.bc -o %t.o -save-temps \
+; RUN:     -r=%t1.bc,_foo,plx \
+; RUN:     -r=%t2.bc,_main,plx \
+; RUN:     -r=%t2.bc,_foo,l
+; RUN: llvm-dis %t.o.1.3.import.bc -o - | FileCheck %s
+; CHECK: define available_externally void @foo()
+
+; We shouldn't do any importing at -O0
+; rm -f %t.o.1.3.import.bc
+; RUN: llvm-lto2 %t1.bc %t2.bc -o %t.o -save-temps \
+; RUN:     -O0 \
+; RUN:     -r=%t1.bc,_foo,plx \
+; RUN:     -r=%t2.bc,_main,plx \
+; RUN:     -r=%t2.bc,_foo,l
+; RUN: llvm-dis %t.o.1.3.import.bc -o - | FileCheck %s --check-prefix=CHECKO0
+; CHECKO0: declare void @foo(...)
+
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.11.0"
+
+define void @foo() #0 {
+entry:
+  ret void
+}

diff --git a/test/tools/gold/X86/Inputs/thinlto_funcimport.ll b/test/tools/gold/X86/Inputs/thinlto_funcimport.ll
new file mode 100644 (file)
index 0000000..74246ab
--- /dev/null
+++ b/test/tools/gold/X86/Inputs/thinlto_funcimport.ll
@@ -0,0 +1,10 @@
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+define i32 @main() #0 {
+entry:
+  call void (...) @foo()
+  ret i32 0
+}
+
+declare void @foo(...) #1

diff --git a/test/tools/gold/X86/thinlto_funcimport.ll b/test/tools/gold/X86/thinlto_funcimport.ll
new file mode 100644 (file)
index 0000000..d00b42d
--- /dev/null
+++ b/test/tools/gold/X86/thinlto_funcimport.ll
@@ -0,0 +1,28 @@
+; Do setup work for all below tests: generate bitcode and combined index
+; RUN: opt -module-summary %s -o %t1.bc
+; RUN: opt -module-summary %p/Inputs/thinlto_funcimport.ll -o %t2.bc
+
+; RUN: %gold -plugin %llvmshlibdir/LLVMgold.so \
+; RUN:    --plugin-opt=save-temps \
+; RUN:    --plugin-opt=thinlto \
+; RUN:    -shared %t1.bc %t2.bc -o %t
+; RUN: llvm-dis %t2.bc.3.import.bc -o - | FileCheck %s
+; CHECK: define available_externally void @foo()
+
+; We shouldn't do any importing at -O0
+; rm -f %t2.bc.3.import.bc
+; RUN: %gold -plugin %llvmshlibdir/LLVMgold.so \
+; RUN:    --plugin-opt=save-temps \
+; RUN:    --plugin-opt=thinlto \
+; RUN:    --plugin-opt=O0 \
+; RUN:    -shared %t1.bc %t2.bc -o %t
+; RUN: llvm-dis %t2.bc.3.import.bc -o - | FileCheck %s --check-prefix=CHECKO0
+; CHECKO0: declare void @foo(...)
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+define void @foo() #0 {
+entry:
+  ret void
+}

diff --git a/tools/llvm-lto2/llvm-lto2.cpp b/tools/llvm-lto2/llvm-lto2.cpp
index 042f874596fb1edbe8384e6f4f22318ca0d9b51e..b48f6c1993e1d4dfd5d3c98e0f4e6d13799fd54c 100644 (file)
--- a/tools/llvm-lto2/llvm-lto2.cpp
+++ b/tools/llvm-lto2/llvm-lto2.cpp
@@ -26,6 +26,11 @@ using namespace llvm;
 using namespace lto;
 using namespace object;
 
+static cl::opt<char>
+    OptLevel("O", cl::desc("Optimization level. [-O0, -O1, -O2, or -O3] "
+                           "(default = '-O2')"),
+             cl::Prefix, cl::ZeroOrMore, cl::init('2'));
+
 static cl::list<std::string> InputFilenames(cl::Positional, cl::OneOrMore,
                                             cl::desc("<input bitcode files>"));
 
@@ -145,6 +150,8 @@ int main(int argc, char **argv) {
   Conf.OptPipeline = OptPipeline;
   Conf.AAPipeline = AAPipeline;
 
+  Conf.OptLevel = OptLevel - '0';
+
   ThinBackend Backend;
   if (ThinLTODistributedIndexes)
     Backend = createWriteIndexesThinBackend("", "", true, "");