[ThinLTO][Legacy] Compute PreservedGUID based on IRName in Symtab

Instead of computing GUID based on some assumption about symbol mangling
rule from IRName to symbol name, lookup the IRName from all the symtabs
from all the input files to see if there are any matching symbols entry
provides the IRName for GUID computation.

rdar://65853754

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D84803

diff --git a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp b/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
index d0a1e18..4adc9a2 100644 (file)
--- a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
+++ b/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
@@ -269,16 +269,26 @@ addUsedSymbolToPreservedGUID(const lto::InputFile &File,
 }
 
 // Convert the PreservedSymbols map from "Name" based to "GUID" based.
+static void computeGUIDPreservedSymbols(const lto::InputFile &File,
+                                        const StringSet<> &PreservedSymbols,
+                                        const Triple &TheTriple,
+                                        DenseSet<GlobalValue::GUID> &GUIDs) {
+  // Iterate the symbols in the input file and if the input has preserved symbol
+  // compute the GUID for the symbol.
+  for (const auto &Sym : File.symbols()) {
+    if (PreservedSymbols.count(Sym.getName()))
+      GUIDs.insert(GlobalValue::getGUID(GlobalValue::getGlobalIdentifier(
+          Sym.getIRName(), GlobalValue::ExternalLinkage, "")));
+  }
+}
+
 static DenseSet<GlobalValue::GUID>
-computeGUIDPreservedSymbols(const StringSet<> &PreservedSymbols,
+computeGUIDPreservedSymbols(const lto::InputFile &File,
+                            const StringSet<> &PreservedSymbols,
                             const Triple &TheTriple) {
   DenseSet<GlobalValue::GUID> GUIDPreservedSymbols(PreservedSymbols.size());
-  for (auto &Entry : PreservedSymbols) {
-    StringRef Name = Entry.first();
-    if (TheTriple.isOSBinFormatMachO() && Name.size() > 0 && Name[0] == '_')
-      Name = Name.drop_front();
-    GUIDPreservedSymbols.insert(GlobalValue::getGUID(Name));
-  }
+  computeGUIDPreservedSymbols(File, PreservedSymbols, TheTriple,
+                              GUIDPreservedSymbols);
   return GUIDPreservedSymbols;
 }
 
@@ -652,7 +662,7 @@ void ThinLTOCodeGenerator::promote(Module &TheModule, ModuleSummaryIndex &Index,
 
   // Convert the preserved symbols set from string to GUID
   auto GUIDPreservedSymbols = computeGUIDPreservedSymbols(
-      PreservedSymbols, Triple(TheModule.getTargetTriple()));
+      File, PreservedSymbols, Triple(TheModule.getTargetTriple()));
 
   // Add used symbol to the preserved symbols.
   addUsedSymbolToPreservedGUID(File, GUIDPreservedSymbols);
@@ -702,7 +712,7 @@ void ThinLTOCodeGenerator::crossModuleImport(Module &TheModule,
 
   // Convert the preserved symbols set from string to GUID
   auto GUIDPreservedSymbols = computeGUIDPreservedSymbols(
-      PreservedSymbols, Triple(TheModule.getTargetTriple()));
+      File, PreservedSymbols, Triple(TheModule.getTargetTriple()));
 
   addUsedSymbolToPreservedGUID(File, GUIDPreservedSymbols);
 
@@ -737,7 +747,7 @@ void ThinLTOCodeGenerator::gatherImportedSummariesForModule(
 
   // Convert the preserved symbols set from string to GUID
   auto GUIDPreservedSymbols = computeGUIDPreservedSymbols(
-      PreservedSymbols, Triple(TheModule.getTargetTriple()));
+      File, PreservedSymbols, Triple(TheModule.getTargetTriple()));
 
   addUsedSymbolToPreservedGUID(File, GUIDPreservedSymbols);
 
@@ -770,7 +780,7 @@ void ThinLTOCodeGenerator::emitImports(Module &TheModule, StringRef OutputName,
 
   // Convert the preserved symbols set from string to GUID
   auto GUIDPreservedSymbols = computeGUIDPreservedSymbols(
-      PreservedSymbols, Triple(TheModule.getTargetTriple()));
+      File, PreservedSymbols, Triple(TheModule.getTargetTriple()));
 
   addUsedSymbolToPreservedGUID(File, GUIDPreservedSymbols);
 
@@ -808,7 +818,7 @@ void ThinLTOCodeGenerator::internalize(Module &TheModule,
 
   // Convert the preserved symbols set from string to GUID
   auto GUIDPreservedSymbols =
-      computeGUIDPreservedSymbols(PreservedSymbols, TMBuilder.TheTriple);
+      computeGUIDPreservedSymbols(File, PreservedSymbols, TMBuilder.TheTriple);
 
   addUsedSymbolToPreservedGUID(File, GUIDPreservedSymbols);
 
@@ -972,8 +982,10 @@ void ThinLTOCodeGenerator::run() {
 
   // Convert the preserved symbols set from string to GUID, this is needed for
   // computing the caching hash and the internalization.
-  auto GUIDPreservedSymbols =
-      computeGUIDPreservedSymbols(PreservedSymbols, TMBuilder.TheTriple);
+  DenseSet<GlobalValue::GUID> GUIDPreservedSymbols;
+  for (const auto &M : Modules)
+    computeGUIDPreservedSymbols(*M, PreservedSymbols, TMBuilder.TheTriple,
+                                GUIDPreservedSymbols);
 
   // Add used symbol from inputs to the preserved symbols.
   for (const auto &M : Modules)

diff --git a/llvm/test/ThinLTO/X86/internalize.ll b/llvm/test/ThinLTO/X86/internalize.ll
index edd5abe..5d80a4f 100644 (file)
--- a/llvm/test/ThinLTO/X86/internalize.ll
+++ b/llvm/test/ThinLTO/X86/internalize.ll
@@ -4,13 +4,13 @@
 ; prevailing the %t1.bc copy as non-prevailing.
 ; RUN: llvm-lto -thinlto-action=thinlink -o %t.index.bc %t2.bc %t1.bc
 ; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc -o - | llvm-dis -o - | FileCheck %s --check-prefix=REGULAR
-; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc -o -  --exported-symbol=foo | llvm-dis -o - | FileCheck %s --check-prefix=INTERNALIZE
+; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc -o -  --exported-symbol=_foo | llvm-dis -o - | FileCheck %s --check-prefix=INTERNALIZE
 
 ; Test the enable-lto-internalization option by setting it to false.
 ; This makes sure indices are not marked as internallinkage and therefore
 ; internalization does not happen.
 ; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc \
-; RUN:          -enable-lto-internalization=false --exported-symbol=foo
+; RUN:          -enable-lto-internalization=false --exported-symbol=_foo
 ; RUN: llvm-dis < %t1.bc.thinlto.internalized.bc | FileCheck %s --check-prefix=INTERNALIZE-OPTION-DISABLE
 
 ; RUN: llvm-lto2 run %t1.bc -o %t.o -save-temps \

diff --git a/llvm/test/ThinLTO/X86/mangled_symbol.ll b/llvm/test/ThinLTO/X86/mangled_symbol.ll
new file mode 100644 (file)
index 0000000..ffdefe3
--- /dev/null
+++ b/llvm/test/ThinLTO/X86/mangled_symbol.ll
@@ -0,0 +1,26 @@
+; RUN: opt -module-summary %s -o %t1.bc
+; RUN: llvm-lto -thinlto-action=thinlink -o %t.index.bc %t1.bc
+;; Check baseline when both of them internalized when not exported.
+; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc -o - --exported-symbol=_exported | llvm-dis -o - | FileCheck %s --check-prefix=INTERNALIZED
+;; Check symbols are exported, including the ones with `\01` prefix.
+; RUN: llvm-lto -thinlto-action=internalize -thinlto-index %t.index.bc %t1.bc -o - --exported-symbol=_exported --exported-symbol=_extern_not_mangled --exported-symbol=_extern_mangled | llvm-dis -o - | FileCheck %s --check-prefix=EXPORTED
+
+; INTERNALIZED: define internal void @extern_not_mangled
+; INTERNALIZED: define internal void @"\01_extern_mangled"
+; EXPORTED: define void @extern_not_mangled
+; EXPORTED: define void @"\01_extern_mangled"
+
+target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.11.0"
+
+define void @exported() {
+  ret void
+}
+
+define void @extern_not_mangled() {
+  ret void
+}
+
+define void @"\01_extern_mangled"() {
+  ret void
+}

diff --git a/llvm/test/ThinLTO/X86/weak_resolution.ll b/llvm/test/ThinLTO/X86/weak_resolution.ll
index b9f10af..ccebb56 100644 (file)
--- a/llvm/test/ThinLTO/X86/weak_resolution.ll
+++ b/llvm/test/ThinLTO/X86/weak_resolution.ll
@@ -7,10 +7,10 @@
 ; non-prevailing ODR are not kept when possible, but non-ODR non-prevailing
 ; are not affected.
 ; RUN: llvm-lto -thinlto-action=promote %t.bc -thinlto-index=%t3.bc -o - | llvm-dis -o - | FileCheck %s --check-prefix=MOD1
-; RUN: llvm-lto -thinlto-action=internalize %t.bc -thinlto-index=%t3.bc -exported-symbol=linkoncefunc -o - | llvm-dis -o - | FileCheck %s --check-prefix=MOD1-INT
+; RUN: llvm-lto -thinlto-action=internalize %t.bc -thinlto-index=%t3.bc -exported-symbol=_linkoncefunc -o - | llvm-dis -o - | FileCheck %s --check-prefix=MOD1-INT
 ; RUN: llvm-lto -thinlto-action=promote %t2.bc -thinlto-index=%t3.bc -o - | llvm-dis -o - | FileCheck %s --check-prefix=MOD2
 ; When exported, we always preserve a linkonce
-; RUN: llvm-lto -thinlto-action=promote %t.bc -thinlto-index=%t3.bc -o - --exported-symbol=linkonceodrfuncInSingleModule | llvm-dis -o - | FileCheck %s --check-prefix=EXPORTED
+; RUN: llvm-lto -thinlto-action=promote %t.bc -thinlto-index=%t3.bc -o - --exported-symbol=_linkonceodrfuncInSingleModule | llvm-dis -o - | FileCheck %s --check-prefix=EXPORTED
 
 target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-apple-macosx10.11.0"

diff --git a/llvm/test/ThinLTO/X86/weak_resolution_single.ll b/llvm/test/ThinLTO/X86/weak_resolution_single.ll
index 779b967..203f60f 100644 (file)
--- a/llvm/test/ThinLTO/X86/weak_resolution_single.ll
+++ b/llvm/test/ThinLTO/X86/weak_resolution_single.ll
@@ -1,7 +1,7 @@
 ; RUN: opt -module-summary %s -o %t.bc
 ; RUN: llvm-lto -thinlto-action=thinlink -o %t2.bc %t.bc
 
-; RUN: llvm-lto -thinlto-action=internalize %t.bc -thinlto-index=%t2.bc -exported-symbol=foo -o - | llvm-dis -o - | FileCheck %s
+; RUN: llvm-lto -thinlto-action=internalize %t.bc -thinlto-index=%t2.bc -exported-symbol=_foo -o - | llvm-dis -o - | FileCheck %s
 
 ; CHECK: define weak_odr void @foo()
 target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"