Merged main:2579b411a13799534c8b8a22246134b88ba7785d into amd-gfx:6cde321d5030

Local branch amd-gfx 6cde321 Merged main:b10303730d7cd009ad505da307b417aaecc91f51 into amd-gfx:c1c160c3b024
Remote branch main 2579b41 [clang-cl] [AST] Fix MS 1920+ placeholder return type mangling for lambdas (llvm#105999)

Author: SC llvm team

clang/docs/ReleaseNotes.rst

@@ -249,6 +249,11 @@ Improvements to Clang's diagnostics
   a warning which defaults to being an error, is enabled by default, and is
   also controlled by the now-deprecated ``-fheinous-gnu-extensions`` flag.
 
+- Added the ``-Wdecls-in-multiple-modules`` option to assist users to identify
+  multiple declarations in different modules, which is the major reason of the slow
+  compilation speed with modules. This warning is disabled by default and it needs
+  to be explicitly enabled or by ``-Weverything``.
+
 Improvements to Clang's time-trace
 ----------------------------------
 

clang/docs/StandardCPlusPlusModules.rst

@@ -894,6 +894,9 @@ approach:
 Reducing the duplication from textual includes is what improves compile-time
 performance.
 
+To help users to identify such issues, we add a warning ``-Wdecls-in-multiple-modules``.
+This warning is disabled by default and it needs to be explicitly enabled or by ``-Weverything``.
+
 Transitioning to modules
 ------------------------
 

clang/lib/AST/MicrosoftMangle.cpp

@@ -2967,13 +2967,12 @@ void MicrosoftCXXNameMangler::mangleFunctionType(const FunctionType *T,
       mangleType(ResultType, Range, QMM_Result);
     } else if (IsInLambda) {
       if (const auto *AT = ResultType->getContainedAutoType()) {
-        assert(AT->getKeyword() == AutoTypeKeyword::Auto &&
-               "should only need to mangle auto!");
-        (void)AT;
+        assert(AT->getKeyword() != AutoTypeKeyword::GNUAutoType &&
+               "shouldn't need to mangle __auto_type!");
         Out << '?';
         mangleQualifiers(ResultType.getLocalQualifiers(), /*IsMember=*/false);
         Out << '?';
-        mangleSourceName("<auto>");
+        mangleSourceName(AT->isDecltypeAuto() ? "<decltype-auto>" : "<auto>");
         Out << '@';
       } else {
         Out << '@';

clang/test/CodeGenCXX/mangle-ms-auto-return.cpp

@@ -215,6 +215,8 @@ void test_template_decltypeauto() {
 // Still want to use clang's custom mangling for lambdas to keep backwards compatibility until
 // MSVC lambda name mangling has been deciphered.
 void test_lambda() {
+  int i = 0;
+
   auto lambdaIntRetAuto = []() { return 0; };
   lambdaIntRetAuto();
   // CHECK: call {{.*}} @"??R<lambda_1>@?0??test_lambda@@YAXXZ@QEBA?A?<auto>@@XZ"
@@ -226,6 +228,18 @@ void test_lambda() {
   auto lambdaGenericIntIntRetAuto = [](auto a) { return a; };
   lambdaGenericIntIntRetAuto(0);
   // CHECK: call {{.*}} @"??$?RH@<lambda_0>@?0??test_lambda@@YAXXZ@QEBA?A?<auto>@@H@Z"
+
+  auto lambdaRetTrailingAuto = []() -> auto { return 0; };
+  lambdaRetTrailingAuto();
+  // CHECK: call {{.*}} @"??R<lambda_3>@?0??test_lambda@@YAXXZ@QEBA?A?<auto>@@XZ"
+
+  auto lambdaRetTrailingDecltypeAuto = []() -> decltype(auto) { return 0; };
+  lambdaRetTrailingDecltypeAuto();
+  // CHECK: call {{.*}} @"??R<lambda_4>@?0??test_lambda@@YAXXZ@QEBA?A?<decltype-auto>@@XZ"
+
+  auto lambdaRetTrailingRefCollapse = [](int x) -> auto&& { return x; };
+  lambdaRetTrailingRefCollapse(i);
+  // CHECK: call {{.*}} @"??R<lambda_5>@?0??test_lambda@@YAXXZ@QEBA?A?<auto>@@H@Z"
 }
 
 auto TestTrailingInt() -> int {

llvm/include/llvm/Config/llvm-config.h.cmake

@@ -16,7 +16,7 @@
 
 /* Indicate that this is LLVM compiled from the amd-gfx branch. */
 #define LLVM_HAVE_BRANCH_AMD_GFX
-#define LLVM_MAIN_REVISION 509571
+#define LLVM_MAIN_REVISION 509576
 
 /* Define if LLVM_ENABLE_DUMP is enabled */
 #cmakedefine LLVM_ENABLE_DUMP

llvm/lib/Target/AMDGPU/AMDGPU.h

@@ -279,6 +279,17 @@ struct AMDGPUAlwaysInlinePass : PassInfoMixin<AMDGPUAlwaysInlinePass> {
   bool GlobalOpt;
 };
 
+void initializeAMDGPUSwLowerLDSLegacyPass(PassRegistry &);
+extern char &AMDGPUSwLowerLDSLegacyPassID;
+ModulePass *
+createAMDGPUSwLowerLDSLegacyPass(const AMDGPUTargetMachine *TM = nullptr);
+
+struct AMDGPUSwLowerLDSPass : PassInfoMixin<AMDGPUSwLowerLDSPass> {
+  const AMDGPUTargetMachine &TM;
+  AMDGPUSwLowerLDSPass(const AMDGPUTargetMachine &TM_) : TM(TM_) {}
+  PreservedAnalyses run(Module &M, ModuleAnalysisManager &AM);
+};
+
 class AMDGPUCodeGenPreparePass
     : public PassInfoMixin<AMDGPUCodeGenPreparePass> {
 private:

llvm/lib/Target/AMDGPU/AMDGPUPassRegistry.def

@@ -20,6 +20,7 @@ MODULE_PASS("amdgpu-always-inline", AMDGPUAlwaysInlinePass())
 MODULE_PASS("amdgpu-lower-buffer-fat-pointers",
             AMDGPULowerBufferFatPointersPass(*this))
 MODULE_PASS("amdgpu-lower-ctor-dtor", AMDGPUCtorDtorLoweringPass())
+MODULE_PASS("amdgpu-sw-lower-lds", AMDGPUSwLowerLDSPass(*this))
 MODULE_PASS("amdgpu-lower-module-lds", AMDGPULowerModuleLDSPass(*this))
 MODULE_PASS("amdgpu-perf-hint",
             AMDGPUPerfHintAnalysisPass(