Optimize linkonce_odr unnamed_addr functions during LTO.

Generalize the API so we can distinguish symbols that are needed just for a DSO
symbol table from those that are used from some native .o.

The symbols that are only wanted for the dso symbol table can be dropped if
llvm can prove every other dso has a copy (linkonce_odr) and the address is not
important (unnamed_addr).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@191922 91177308-0d34-0410-b5e6-96231b3b80d8

12 files changed, 124 insertions, 23 deletions

diff --git a/include/llvm-c/lto.h b/include/llvm-c/lto.h
index 58d58336b5..85d5439ec9 100644
--- a/include/llvm-c/lto.h
+++ b/include/llvm-c/lto.h
@@ -29,7 +29,7 @@
  * @{
  */
 
-#define LTO_API_VERSION 4
+#define LTO_API_VERSION 5
 
 typedef enum {
     LTO_SYMBOL_ALIGNMENT_MASK              = 0x0000001F, /* log2 of alignment */
@@ -253,13 +253,21 @@ lto_codegen_set_assembler_args(lto_code_gen_t cg, const char **args,
                                int nargs);
 
 /**
- * Adds to a list of all global symbols that must exist in the final
- * generated code.  If a function is not listed, it might be
- * inlined into every usage and optimized away.
+ * Tells LTO optimization passes that this symbol must be preserved
+ * because it is referenced by native code or a command line option.
  */
 extern void
 lto_codegen_add_must_preserve_symbol(lto_code_gen_t cg, const char* symbol);
 
+
+/**
+ * Tells LTO optimization passes that a dynamic shared library is being
+ * built and this symbol may be exported. Unless IR semantics allow the symbol
+ * to be made local to the library, it should remain so it can be exported by
+ * the shared library.
+ */
+extern void lto_codegen_add_dso_symbol(lto_code_gen_t cg, const char *symbol);
+
 /**
  * Writes a new object file at the specified path that contains the
  * merged contents of all modules added so far.
diff --git a/include/llvm/LTO/LTOCodeGenerator.h b/include/llvm/LTO/LTOCodeGenerator.h
index 97a5066339..9f50770dcd 100644
--- a/include/llvm/LTO/LTOCodeGenerator.h
+++ b/include/llvm/LTO/LTOCodeGenerator.h
@@ -73,6 +73,10 @@ struct LTOCodeGenerator {
 
   void addMustPreserveSymbol(const char *sym) { MustPreserveSymbols[sym] = 1; }
 
+  void addDSOSymbol(const char* Sym) {
+    DSOSymbols[Sym] = 1;
+  }
+
   // To pass options to the driver and optimization passes. These options are
   // not necessarily for debugging purpose (The function name is misleading).
   // This function should be called before LTOCodeGenerator::compilexxx(),
@@ -126,6 +130,7 @@ private:
   void applyScopeRestrictions();
   void applyRestriction(llvm::GlobalValue &GV,
                         std::vector<const char*> &MustPreserveList,
+                        std::vector<const char*> &SymtabList,
                         llvm::SmallPtrSet<llvm::GlobalValue*, 8> &AsmUsed,
                         llvm::Mangler &Mangler);
   bool determineTarget(std::string &errMsg);
@@ -138,6 +143,7 @@ private:
   bool EmitDwarfDebugInfo;
   bool ScopeRestrictionsDone;
   lto_codegen_model CodeModel;
+  StringSet DSOSymbols;
   StringSet MustPreserveSymbols;
   StringSet AsmUndefinedRefs;
   llvm::MemoryBuffer *NativeObjectFile;
diff --git a/include/llvm/Transforms/IPO.h b/include/llvm/Transforms/IPO.h
index 6ebb6c4658..ed2823098a 100644
--- a/include/llvm/Transforms/IPO.h
+++ b/include/llvm/Transforms/IPO.h
@@ -104,12 +104,32 @@ Pass *createPruneEHPass();
 
 //===----------------------------------------------------------------------===//
 /// createInternalizePass - This pass loops over all of the functions in the
-/// input module, internalizing all globals (functions and variables) not in the
-/// given exportList.
+/// input module, internalizing all globals (functions and variables) it can.
+////
+/// The symbols in \p ExportList are never internalized.
+///
+/// The symbol in DSOList are internalized if it is safe to drop them from
+/// the symbol table.
+///
+/// For example of the difference, consider a dynamic library being built from
+/// two translation units. The first one compiled to a native object
+/// (ELF/MachO/COFF) and second one compiled to IL. Translation unit A has a
+/// copy of linkonce_odr unnamed_addr function F. The translation unit B has a
+/// copy of the linkonce_odr unnamed_addr functions F and G.
+///
+/// Assume the linker decides to keep the copy of F in B. This means that LLVM
+/// must produce F in the object file it passes to the linker, otherwise we
+/// will have an undefined reference. For G the situation is different. The
+/// linker puts the function in the DSOList, since it is only wanted for the
+/// symbol table. With this information internalize can now reason that since
+/// the function is a linkonce_odr and its address is not important, it can be
+/// omitted. Any other shared library needing this function will have a copy of
+/// it.
 ///
 /// Note that commandline options that are used with the above function are not
 /// used now!
-ModulePass *createInternalizePass(ArrayRef<const char *> ExportList);
+ModulePass *createInternalizePass(ArrayRef<const char *> ExportList,
+                                  ArrayRef<const char *> DSOList);
 /// createInternalizePass - Same as above, but with an empty exportList.
 ModulePass *createInternalizePass();
 
diff --git a/lib/LTO/LTOCodeGenerator.cpp b/lib/LTO/LTOCodeGenerator.cpp
index 3abb6238b3..e35e33666e 100644
--- a/lib/LTO/LTOCodeGenerator.cpp
+++ b/lib/LTO/LTOCodeGenerator.cpp
@@ -310,6 +310,7 @@ bool LTOCodeGenerator::determineTarget(std::string &errMsg) {
 void LTOCodeGenerator::
 applyRestriction(GlobalValue &GV,
                  std::vector<const char*> &MustPreserveList,
+                 std::vector<const char*> &DSOList,
                  SmallPtrSet<GlobalValue*, 8> &AsmUsed,
                  Mangler &Mangler) {
   SmallString<64> Buffer;
@@ -319,6 +320,8 @@ applyRestriction(GlobalValue &GV,
     return;
   if (MustPreserveSymbols.count(Buffer))
     MustPreserveList.push_back(GV.getName().data());
+  if (DSOSymbols.count(Buffer))
+    DSOList.push_back(GV.getName().data());
   if (AsmUndefinedRefs.count(Buffer))
     AsmUsed.insert(&GV);
 }
@@ -348,17 +351,18 @@ void LTOCodeGenerator::applyScopeRestrictions() {
                      NULL);
   Mangler Mangler(MContext, TargetMach);
   std::vector<const char*> MustPreserveList;
+  std::vector<const char*> DSOList;
   SmallPtrSet<GlobalValue*, 8> AsmUsed;
 
   for (Module::iterator f = mergedModule->begin(),
          e = mergedModule->end(); f != e; ++f)
-    applyRestriction(*f, MustPreserveList, AsmUsed, Mangler);
+    applyRestriction(*f, MustPreserveList, DSOList, AsmUsed, Mangler);
   for (Module::global_iterator v = mergedModule->global_begin(),
          e = mergedModule->global_end(); v !=  e; ++v)
-    applyRestriction(*v, MustPreserveList, AsmUsed, Mangler);
+    applyRestriction(*v, MustPreserveList, DSOList, AsmUsed, Mangler);
   for (Module::alias_iterator a = mergedModule->alias_begin(),
          e = mergedModule->alias_end(); a != e; ++a)
-    applyRestriction(*a, MustPreserveList, AsmUsed, Mangler);
+    applyRestriction(*a, MustPreserveList, DSOList, AsmUsed, Mangler);
 
   GlobalVariable *LLVMCompilerUsed =
     mergedModule->getGlobalVariable("llvm.compiler.used");
@@ -386,7 +390,7 @@ void LTOCodeGenerator::applyScopeRestrictions() {
     LLVMCompilerUsed->setSection("llvm.metadata");
   }
 
-  passes.add(createInternalizePass(MustPreserveList));
+  passes.add(createInternalizePass(MustPreserveList, DSOList));
 
   // apply scope restrictions
   passes.run(*mergedModule);
diff --git a/lib/Transforms/IPO/IPO.cpp b/lib/Transforms/IPO/IPO.cpp
index 5d563d8bbf..5f26bac2c8 100644
--- a/lib/Transforms/IPO/IPO.cpp
+++ b/lib/Transforms/IPO/IPO.cpp
@@ -98,7 +98,7 @@ void LLVMAddInternalizePass(LLVMPassManagerRef PM, unsigned AllButMain) {
   std::vector<const char *> Export;
   if (AllButMain)
     Export.push_back("main");
-  unwrap(PM)->add(createInternalizePass(Export));
+  unwrap(PM)->add(createInternalizePass(Export, None));
 }
 
 void LLVMAddStripDeadPrototypesPass(LLVMPassManagerRef PM) {
diff --git a/lib/Transforms/IPO/Internalize.cpp b/lib/Transforms/IPO/Internalize.cpp
index f2feacc0b6..f20a7bd92f 100644
--- a/lib/Transforms/IPO/Internalize.cpp
+++ b/lib/Transforms/IPO/Internalize.cpp
@@ -44,13 +44,20 @@ APIList("internalize-public-api-list", cl::value_desc("list"),
         cl::desc("A list of symbol names to preserve"),
         cl::CommaSeparated);
 
+static cl::list<std::string>
+DSOList("internalize-dso-list", cl::value_desc("list"),
+        cl::desc("A list of symbol names need for a dso symbol table"),
+        cl::CommaSeparated);
+
 namespace {
   class InternalizePass : public ModulePass {
     std::set<std::string> ExternalNames;
+    std::set<std::string> DSONames;
   public:
     static char ID; // Pass identification, replacement for typeid
     explicit InternalizePass();
-    explicit InternalizePass(ArrayRef<const char *> ExportList);
+    explicit InternalizePass(ArrayRef<const char *> ExportList,
+                             ArrayRef<const char *> DSOList);
     void LoadFile(const char *Filename);
     virtual bool runOnModule(Module &M);
 
@@ -71,15 +78,21 @@ InternalizePass::InternalizePass()
   if (!APIFile.empty())           // If a filename is specified, use it.
     LoadFile(APIFile.c_str());
   ExternalNames.insert(APIList.begin(), APIList.end());
+  DSONames.insert(DSOList.begin(), DSOList.end());
 }
 
-InternalizePass::InternalizePass(ArrayRef<const char *> ExportList)
+InternalizePass::InternalizePass(ArrayRef<const char *> ExportList,
+                                 ArrayRef<const char *> DSOList)
   : ModulePass(ID){
   initializeInternalizePassPass(*PassRegistry::getPassRegistry());
   for(ArrayRef<const char *>::const_iterator itr = ExportList.begin();
         itr != ExportList.end(); itr++) {
     ExternalNames.insert(*itr);
   }
+  for(ArrayRef<const char *>::const_iterator itr = DSOList.begin();
+        itr != DSOList.end(); itr++) {
+    DSONames.insert(*itr);
+  }
 }
 
 void InternalizePass::LoadFile(const char *Filename) {
@@ -99,7 +112,8 @@ void InternalizePass::LoadFile(const char *Filename) {
 }
 
 static bool shouldInternalize(const GlobalValue &GV,
-                              const std::set<std::string> &ExternalNames) {
+                              const std::set<std::string> &ExternalNames,
+                              const std::set<std::string> &DSONames) {
   // Function must be defined here
   if (GV.isDeclaration())
     return false;
@@ -116,7 +130,20 @@ static bool shouldInternalize(const GlobalValue &GV,
   if (ExternalNames.count(GV.getName()))
     return false;
 
-  return true;
+  // Not needed for the symbol table?
+  if (!DSONames.count(GV.getName()))
+    return true;
+
+  // Not a linkonce. Someone can depend on it being on the symbol table.
+  if (!GV.hasLinkOnceLinkage())
+    return false;
+
+  // The address is not important, we can hide it.
+  if (GV.hasUnnamedAddr())
+    return true;
+
+  // FIXME: Check if the address is used.
+  return false;
 }
 
 bool InternalizePass::runOnModule(Module &M) {
@@ -145,7 +172,7 @@ bool InternalizePass::runOnModule(Module &M) {
   // Mark all functions not in the api as internal.
   // FIXME: maybe use private linkage?
   for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
-    if (!shouldInternalize(*I, ExternalNames))
+    if (!shouldInternalize(*I, ExternalNames, DSONames))
       continue;
 
     I->setLinkage(GlobalValue::InternalLinkage);
@@ -182,7 +209,7 @@ bool InternalizePass::runOnModule(Module &M) {
   // FIXME: maybe use private linkage?
   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
        I != E; ++I) {
-    if (!shouldInternalize(*I, ExternalNames))
+    if (!shouldInternalize(*I, ExternalNames, DSONames))
       continue;
 
     I->setLinkage(GlobalValue::InternalLinkage);
@@ -194,7 +221,7 @@ bool InternalizePass::runOnModule(Module &M) {
   // Mark all aliases that are not in the api as internal as well.
   for (Module::alias_iterator I = M.alias_begin(), E = M.alias_end();
        I != E; ++I) {
-    if (!shouldInternalize(*I, ExternalNames))
+    if (!shouldInternalize(*I, ExternalNames, DSONames))
       continue;
 
     I->setLinkage(GlobalValue::InternalLinkage);
@@ -210,6 +237,7 @@ ModulePass *llvm::createInternalizePass() {
   return new InternalizePass();
 }
 
-ModulePass *llvm::createInternalizePass(ArrayRef<const char *> ExportList) {
-  return new InternalizePass(ExportList);
+ModulePass *llvm::createInternalizePass(ArrayRef<const char *> ExportList,
+                                        ArrayRef<const char *> DSOList) {
+  return new InternalizePass(ExportList, DSOList);
 }
diff --git a/lib/Transforms/IPO/PassManagerBuilder.cpp b/lib/Transforms/IPO/PassManagerBuilder.cpp
index 2008c5d65e..b9660fab78 100644
--- a/lib/Transforms/IPO/PassManagerBuilder.cpp
+++ b/lib/Transforms/IPO/PassManagerBuilder.cpp
@@ -277,7 +277,7 @@ void PassManagerBuilder::populateLTOPassManager(PassManagerBase &PM,
   // for a main function.  If main is defined, mark all other functions
   // internal.
   if (Internalize)
-    PM.add(createInternalizePass("main"));
+    PM.add(createInternalizePass("main", None));
 
   // Propagate constants at call sites into the functions they call.  This
   // opens opportunities for globalopt (and inlining) by substituting function
diff --git a/test/LTO/cfi_endproc.ll b/test/LTO/cfi_endproc.ll
index 4ea53e7be7..acfaf5a800 100644
--- a/test/LTO/cfi_endproc.ll
+++ b/test/LTO/cfi_endproc.ll
@@ -27,3 +27,11 @@ define i32 @main(i32 %argc, i8** %argv) {
   call void @PR14512()
   ret i32 0
 }
+
+; RUN: llvm-lto -o %t -dso-symbol=zed1 -dso-symbol=zed2 %t1 -disable-opt
+; RUN: llvm-nm %t | FileCheck %s -check-prefix=ZED1_AND_ZED2
+; ZED1_AND_ZED2: V zed1
+@zed1 = linkonce_odr global i32 42
+
+; ZED1_AND_ZED2: d zed2
+@zed2 = linkonce_odr unnamed_addr global i32 42
diff --git a/test/Transforms/Internalize/lists.ll b/test/Transforms/Internalize/lists.ll
index 83e441a2df..59fe0737d3 100644
--- a/test/Transforms/Internalize/lists.ll
+++ b/test/Transforms/Internalize/lists.ll
@@ -13,6 +13,10 @@
 ; -file and -list options should be merged, the apifile contains foo and j
 ; RUN: opt < %s -internalize -internalize-public-api-list bar -internalize-public-api-file %S/apifile -S | FileCheck --check-prefix=FOO_J_AND_BAR %s
 
+; Put zed1 and zed2 in the symbol table. If the address is not relevant, we
+; internalize them.
+; RUN: opt < %s -internalize -internalize-dso-list zed1,zed2 -S | FileCheck --check-prefix=ZED1_AND_ZED2 %s
+
 ; ALL: @i = internal global
 ; FOO_AND_J: @i = internal global
 ; FOO_AND_BAR: @i = internal global
@@ -25,6 +29,12 @@
 ; FOO_J_AND_BAR: @j = global
 @j = global i32 0
 
+; ZED1_AND_ZED2: @zed1 = linkonce_odr global i32 42
+@zed1 = linkonce_odr global i32 42
+
+; ZED1_AND_ZED2: @zed2 = internal unnamed_addr global i32 42
+@zed2 = linkonce_odr unnamed_addr global i32 42
+
 ; ALL: define internal void @main() {
 ; FOO_AND_J: define internal void @main() {
 ; FOO_AND_BAR: define internal void @main() {
diff --git a/tools/gold/gold-plugin.cpp b/tools/gold/gold-plugin.cpp
index bd11d1b555..119631cfa7 100644
--- a/tools/gold/gold-plugin.cpp
+++ b/tools/gold/gold-plugin.cpp
@@ -197,7 +197,7 @@ ld_plugin_status onload(ld_plugin_tv *tv) {
       case LDPT_ADD_SYMBOLS:
         add_symbols = tv->tv_u.tv_add_symbols;
         break;
-      case LDPT_GET_SYMBOLS:
+      case LDPT_GET_SYMBOLS_V2:
         get_symbols = tv->tv_u.tv_get_symbols;
         break;
       case LDPT_ADD_INPUT_FILE:
@@ -386,6 +386,11 @@ static ld_plugin_status all_symbols_read_hook(void) {
 
         if (options::generate_api_file)
           api_file << I->syms[i].name << "\n";
+      } else if (I->syms[i].resolution == LDPR_PREVAILING_DEF_IRONLY_EXP) {
+        lto_codegen_add_dso_symbol(code_gen, I->syms[i].name);
+
+        if (options::generate_api_file)
+          api_file << I->syms[i].name << " dso only\n";
       }
     }
   }
diff --git a/tools/llvm-lto/llvm-lto.cpp b/tools/llvm-lto/llvm-lto.cpp
index 3ecd13f5c7..1d03fa62a8 100644
--- a/tools/llvm-lto/llvm-lto.cpp
+++ b/tools/llvm-lto/llvm-lto.cpp
@@ -50,6 +50,10 @@ ExportedSymbols("exported-symbol",
   cl::desc("Symbol to export from the resulting object file"),
   cl::ZeroOrMore);
 
+static cl::list<std::string>
+DSOSymbols("dso-symbol",
+  cl::desc("Symbol to put in the symtab in the resulting dso"),
+  cl::ZeroOrMore);
 
 int main(int argc, char **argv) {
   // Print a stack trace if we signal out.
@@ -117,6 +121,10 @@ int main(int argc, char **argv) {
   for (unsigned i = 0; i < ExportedSymbols.size(); ++i)
     CodeGen.addMustPreserveSymbol(ExportedSymbols[i].c_str());
 
+  // Add all the dso symbols to the table of symbols to expose.
+  for (unsigned i = 0; i < DSOSymbols.size(); ++i)
+    CodeGen.addDSOSymbol(DSOSymbols[i].c_str());
+
   if (!OutputFilename.empty()) {
     size_t len = 0;
     std::string ErrorInfo;
diff --git a/tools/lto/lto.cpp b/tools/lto/lto.cpp
index 7bfddcd9ec..a3acd4c950 100644
--- a/tools/lto/lto.cpp
+++ b/tools/lto/lto.cpp
@@ -260,6 +260,10 @@ void lto_codegen_add_must_preserve_symbol(lto_code_gen_t cg,
   cg->addMustPreserveSymbol(symbol);
 }
 
+void lto_codegen_add_dso_symbol(lto_code_gen_t cg, const char *symbol) {
+  cg->addDSOSymbol(symbol);
+}
+
 /// lto_codegen_write_merged_modules - Writes a new file at the specified path
 /// that contains the merged contents of all modules added so far. Returns true
 /// on error (check lto_get_error_message() for details).