Merged main:f590963db836 into amd-gfx:23e4c3a6e3aa

Local branch amd-gfx 23e4c3a Merged main:9abcca5e2529 into amd-gfx:387d39f6081c
Remote branch main f590963 [RISCV] Implement RISCVTTIImpl::getPreferredAddressingMode for HasVendorXCVmem (llvm#120533)

Author: SC llvm team

clang-tools-extra/clangd/tool/ClangdMain.cpp

@@ -242,13 +242,13 @@ opt<std::string> FallbackStyle{
     init(clang::format::DefaultFallbackStyle),
 };
 
-opt<int> EnableFunctionArgSnippets{
+opt<std::string> EnableFunctionArgSnippets{
     "function-arg-placeholders",
     cat(Features),
     desc("When disabled (0), completions contain only parentheses for "
          "function calls. When enabled (1), completions also contain "
          "placeholders for method parameters"),
-    init(-1),
+    init("-1"),
 };
 
 opt<CodeCompleteOptions::IncludeInsertion> HeaderInsertion{
@@ -636,6 +636,22 @@ loadExternalIndex(const Config::ExternalIndexSpec &External,
   llvm_unreachable("Invalid ExternalIndexKind.");
 }
 
+std::optional<bool> shouldEnableFunctionArgSnippets() {
+  std::string Val = EnableFunctionArgSnippets;
+  // Accept the same values that a bool option parser would, but also accept
+  // -1 to indicate "unspecified", in which case the ArgumentListsPolicy
+  // config option will be respected.
+  if (Val == "1" || Val == "true" || Val == "True" || Val == "TRUE")
+    return true;
+  if (Val == "0" || Val == "false" || Val == "False" || Val == "FALSE")
+    return false;
+  if (Val != "-1")
+    elog("Value specified by --function-arg-placeholders is invalid. Provide a "
+         "boolean value or leave unspecified to use ArgumentListsPolicy from "
+         "config instead.");
+  return std::nullopt;
+}
+
 class FlagsConfigProvider : public config::Provider {
 private:
   config::CompiledFragment Frag;
@@ -696,10 +712,9 @@ class FlagsConfigProvider : public config::Provider {
       BGPolicy = Config::BackgroundPolicy::Skip;
     }
 
-    if (EnableFunctionArgSnippets >= 0) {
-      ArgumentLists = EnableFunctionArgSnippets
-                          ? Config::ArgumentListsPolicy::FullPlaceholders
-                          : Config::ArgumentListsPolicy::Delimiters;
+    if (std::optional<bool> Enable = shouldEnableFunctionArgSnippets()) {
+      ArgumentLists = *Enable ? Config::ArgumentListsPolicy::FullPlaceholders
+                              : Config::ArgumentListsPolicy::Delimiters;
     }
 
     Frag = [=](const config::Params &, Config &C) {

clang/lib/Sema/SemaOverload.cpp

@@ -6977,11 +6977,26 @@ void Sema::AddOverloadCandidate(
     /// have linkage. So that all entities of the same should share one
     /// linkage. But in clang, different entities of the same could have
     /// different linkage.
-    NamedDecl *ND = Function;
-    if (auto *SpecInfo = Function->getTemplateSpecializationInfo())
+    const NamedDecl *ND = Function;
+    bool IsImplicitlyInstantiated = false;
+    if (auto *SpecInfo = Function->getTemplateSpecializationInfo()) {
       ND = SpecInfo->getTemplate();
-
-    if (ND->getFormalLinkage() == Linkage::Internal) {
+      IsImplicitlyInstantiated = SpecInfo->getTemplateSpecializationKind() ==
+                                 TSK_ImplicitInstantiation;
+    }
+
+    /// Don't remove inline functions with internal linkage from the overload
+    /// set if they are declared in a GMF, in violation of C++ [basic.link]p17.
+    /// However:
+    /// - Inline functions with internal linkage are a common pattern in
+    ///   headers to avoid ODR issues.
+    /// - The global module is meant to be a transition mechanism for C and C++
+    ///   headers, and the current rules as written work against that goal.
+    const bool IsInlineFunctionInGMF =
+        Function->isFromGlobalModule() &&
+        (IsImplicitlyInstantiated || Function->isInlined());
+
+    if (ND->getFormalLinkage() == Linkage::Internal && !IsInlineFunctionInGMF) {
       Candidate.Viable = false;
       Candidate.FailureKind = ovl_fail_module_mismatched;
       return;

clang/test/Modules/expose-static-inline-from-gmf-1.cppm

@@ -0,0 +1,37 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm \
+// RUN:   -DTEST_INLINE
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify \
+// RUN:   -DTEST_INLINE
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify
+
+//--- a.h
+#ifdef TEST_INLINE
+#define INLINE inline
+#else
+#define INLINE
+#endif
+static INLINE void func(long) {}
+template <typename T = long> void a() { func(T{}); }
+
+//--- a.cppm
+module;
+#include "a.h"
+export module a;
+export using ::a;
+
+//--- test.cc
+import a;
+auto m = (a(), 0);
+
+#ifdef TEST_INLINE
+// expected-no-diagnostics
+#else
+// [email protected]:7 {{no matching function for call to 'func'}}
+// [email protected]:2 {{in instantiation of function template specialization 'a<long>' requested here}}
+#endif

clang/test/Modules/expose-static-inline-from-gmf-2.cppm

@@ -0,0 +1,22 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify
+
+//--- a.h
+template <typename G> static inline void func() {}
+template <typename T = long> void a() { func<T>(); }
+
+//--- a.cppm
+module;
+#include "a.h"
+export module a;
+export using ::a;
+
+//--- test.cc
+import a;
+auto m = (a(), 0);
+
+// expected-no-diagnostics

clang/test/Modules/expose-static-inline-from-gmf-3.cppm

@@ -0,0 +1,24 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify
+
+//--- a.h
+namespace ns {
+template <typename G> static void func() {}
+template <typename T = long> void a() { func<T>(); }
+}
+
+//--- a.cppm
+module;
+#include "a.h"
+export module a;
+export using ns::a;
+
+//--- test.cc
+import a;
+auto m = (a(), 0);
+
+// expected-no-diagnostics

clang/test/Modules/expose-static-inline-from-gmf-4.cppm

@@ -0,0 +1,40 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm \
+// RUN:   -DTEST_INLINE
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify \
+// RUN:   -DTEST_INLINE
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify
+
+//--- a.h
+#ifdef TEST_INLINE
+#define INLINE inline
+#else
+#define INLINE
+#endif
+namespace ns {
+template <typename G> static void func() {}
+template <> INLINE void func<long>() {}
+template <typename T = long> void a() { func<T>(); }
+}
+
+//--- a.cppm
+module;
+#include "a.h"
+export module a;
+export using ns::a;
+
+//--- test.cc
+import a;
+auto m = (a(), 0);
+
+#ifdef TEST_INLINE
+// expected-no-diagnostics
+#else
+// [email protected]:9 {{no matching function for call to 'func'}}
+// [email protected]:2 {{in instantiation of function template specialization 'ns::a<long>' requested here}}
+#endif

clang/test/Modules/expose-static-inline-from-gmf-5.cppm

@@ -0,0 +1,26 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang -std=c++20 %t/a.cppm --precompile -o %t/a.pcm
+// RUN: %clang -std=c++20 %t/test.cc -fprebuilt-module-path=%t -fsyntax-only -Xclang -verify
+
+//--- a.h
+namespace ns {
+namespace {
+template <typename G> void func() {}
+}
+template <typename T = long> void a() { func<T>(); }
+}
+
+//--- a.cppm
+module;
+#include "a.h"
+export module a;
+export using ns::a;
+
+//--- test.cc
+import a;
+auto m = (a(), 0);
+
+// expected-no-diagnostics

clang/tools/include-mapping/cppreference_parser.py

@@ -139,7 +139,7 @@ def _ParseIndexPage(index_page_html):
 
 
 def _ReadSymbolPage(path, name, qual_name):
-    with open(path) as f:
+    with open(path, encoding="utf-8") as f:
         return _ParseSymbolPage(f.read(), name, qual_name)
 
 
@@ -156,7 +156,7 @@ def _GetSymbols(pool, root_dir, index_page_name, namespace, variants_to_accept):
     #      contains the defined header.
     #   2. Parse the symbol page to get the defined header.
     index_page_path = os.path.join(root_dir, index_page_name)
-    with open(index_page_path, "r") as f:
+    with open(index_page_path, "r", encoding="utf-8") as f:
         # Read each symbol page in parallel.
         results = []  # (symbol_name, promise of [header...])
         for symbol_name, symbol_page_path, variant in _ParseIndexPage(f.read()):

libc/utils/hdrgen/class_implementation/classes/enumeration.py renamed to libc/utils/hdrgen/enumeration.py

@@ -11,13 +11,14 @@
 import yaml
 import argparse
 from pathlib import Path
-from header import HeaderFile
+
+from enumeration import Enumeration
+from function import Function
 from gpu_headers import GpuHeaderFile as GpuHeader
-from class_implementation.classes.macro import Macro
-from class_implementation.classes.type import Type
-from class_implementation.classes.function import Function
-from class_implementation.classes.enumeration import Enumeration
-from class_implementation.classes.object import Object
+from header import HeaderFile
+from macro import Macro
+from object import Object
+from type import Type
 
 
 def yaml_to_classes(yaml_data, header_class, entry_points=None):

libc/utils/hdrgen/class_implementation/classes/function.py renamed to libc/utils/hdrgen/function.py

@@ -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 522529
+#define LLVM_MAIN_REVISION 522534
 
 /* Define if LLVM_ENABLE_DUMP is enabled */
 #cmakedefine LLVM_ENABLE_DUMP

libc/utils/hdrgen/class_implementation/classes/macro.py renamed to libc/utils/hdrgen/macro.py

@@ -2329,6 +2329,15 @@ unsigned RISCVTTIImpl::getMaximumVF(unsigned ElemWidth, unsigned Opcode) const {
   return std::max<unsigned>(1U, RegWidth.getFixedValue() / ElemWidth);
 }
 
+TTI::AddressingModeKind
+RISCVTTIImpl::getPreferredAddressingMode(const Loop *L,
+                                         ScalarEvolution *SE) const {
+  if (ST->hasVendorXCVmem() && !ST->is64Bit())
+    return TTI::AMK_PostIndexed;
+
+  return BasicTTIImplBase::getPreferredAddressingMode(L, SE);
+}
+
 bool RISCVTTIImpl::isLSRCostLess(const TargetTransformInfo::LSRCost &C1,
                                  const TargetTransformInfo::LSRCost &C2) {
   // RISC-V specific here are "instruction number 1st priority".

libc/utils/hdrgen/class_implementation/classes/object.py renamed to libc/utils/hdrgen/object.py

@@ -388,6 +388,9 @@ class RISCVTTIImpl : public BasicTTIImplBase<RISCVTTIImpl> {
     llvm_unreachable("unknown register class");
   }
 
+  TTI::AddressingModeKind getPreferredAddressingMode(const Loop *L,
+                                                     ScalarEvolution *SE) const;
+
   unsigned getRegisterClassForType(bool Vector, Type *Ty = nullptr) const {
     if (Vector)
       return RISCVRegisterClass::VRRC;

libc/utils/hdrgen/class_implementation/classes/type.py renamed to libc/utils/hdrgen/type.py

@@ -0,0 +1,34 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -O3 -mtriple=riscv32 -mattr=+m,+xcvmem -verify-machineinstrs < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK
+
+define i32 @test_heuristic(ptr %b, i32 %e, i1 %0) {
+; CHECK-LABEL: test_heuristic:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    add a3, a0, a1
+; CHECK-NEXT:    andi a2, a2, 1
+; CHECK-NEXT:  .LBB0_1: # %loop
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    cv.lbu a1, (a3), 1
+; CHECK-NEXT:    addi a0, a0, 1
+; CHECK-NEXT:    beqz a2, .LBB0_1
+; CHECK-NEXT:  # %bb.2: # %exit
+; CHECK-NEXT:    mv a0, a1
+; CHECK-NEXT:    ret
+entry:
+  %1 = getelementptr i8, ptr %b, i32 %e
+  br label %loop
+
+loop:                                             ; preds = %loop, %entry
+  %2 = phi ptr [ %b, %entry ], [ %7, %loop ]
+  %3 = phi ptr [ %1, %entry ], [ %8, %loop ]
+  %4 = load i8, ptr %2, align 1
+  %5 = load i8, ptr %3, align 1
+  %6 = zext i8 %5 to i32
+  %7 = getelementptr i8, ptr %2, i32 1
+  %8 = getelementptr i8, ptr %3, i32 1
+  br i1 %0, label %exit, label %loop
+
+exit:                                             ; preds = %loop
+  ret i32 %6
+}