Merged main:0a369b06e34495966c6c9db427ea52f77a82a0bf into amd-gfx:24ba96a84828

Local branch amd-gfx 24ba96a Merged main:344228ebf45f9bd1f7626fdcd3c0fada0f0c8385 into amd-gfx:d6bc6d5dbcb2
Remote branch main 0a369b0 Reapply "[MachinePipeliner] Fix constraints arent considered in cert… (llvm#97259)

Author: SC llvm team

bolt/lib/Core/BinaryContext.cpp

@@ -20,7 +20,6 @@
 #include "llvm/DebugInfo/DWARF/DWARFCompileUnit.h"
 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
 #include "llvm/DebugInfo/DWARF/DWARFUnit.h"
-#include "llvm/MC/MCAsmLayout.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCDisassembler/MCDisassembler.h"
@@ -2416,8 +2415,7 @@ BinaryContext::calculateEmittedSize(BinaryFunction &BF, bool FixBranches) {
 
   MCAssembler &Assembler =
       static_cast<MCObjectStreamer *>(Streamer.get())->getAssembler();
-  MCAsmLayout Layout(Assembler);
-  Assembler.layout(Layout);
+  Assembler.layout();
 
   // Obtain fragment sizes.
   std::vector<uint64_t> FragmentSizes;

clang/docs/ReleaseNotes.rst

@@ -147,22 +147,6 @@ Clang Frontend Potentially Breaking Changes
 - The ``hasTypeLoc`` AST matcher will no longer match a ``classTemplateSpecializationDecl``;
   existing uses should switch to ``templateArgumentLoc`` or ``hasAnyTemplateArgumentLoc`` instead.
 
-- The comment parser now matches comments to declarations even if there is a
-  preprocessor macro in between the comment and declaration. This change is
-  intended to improve Clang's support for parsing documentation comments and
-  to better conform to Doxygen's behavior.
-
-  This has the potential to cause ``-Wdocumentation`` warnings, especially in
-  cases where a function-like macro has a documentation comment and is followed
-  immediately by a normal function. The function-like macro's documentation
-  comments will be attributed to the subsequent function and may cause
-  ``-Wdocumentation`` warnings such as mismatched parameter names, or invalid
-  return documentation comments.
-
-  In cases where the ``-Wdocumentation`` warnings are thrown, the suggested fix
-  is to document the declaration following the macro so that the warnings are
-  fixed.
-
 Clang Python Bindings Potentially Breaking Changes
 --------------------------------------------------
 - Renamed ``CursorKind`` variant 272 from ``OMP_TEAMS_DISTRIBUTE_DIRECTIVE``

clang/docs/tools/clang-formatted-files.txt

@@ -5357,7 +5357,6 @@ llvm/include/llvm/MC/MCAsmInfoELF.h
 llvm/include/llvm/MC/MCAsmInfoGOFF.h
 llvm/include/llvm/MC/MCAsmInfoWasm.h
 llvm/include/llvm/MC/MCAsmInfoXCOFF.h
-llvm/include/llvm/MC/MCAsmLayout.h
 llvm/include/llvm/MC/MCCodeView.h
 llvm/include/llvm/MC/MCContext.h
 llvm/include/llvm/MC/MCFixedLenDisassembler.h

clang/lib/AST/ASTContext.cpp

@@ -284,10 +284,9 @@ RawComment *ASTContext::getRawCommentForDeclNoCacheImpl(
   StringRef Text(Buffer + CommentEndOffset,
                  DeclLocDecomp.second - CommentEndOffset);
 
-  // There should be no other declarations between comment and declaration.
-  // Preprocessor directives are implicitly allowed to be between a comment and
-  // its associated decl.
-  if (Text.find_last_of(";{}@") != StringRef::npos)
+  // There should be no other declarations or preprocessor directives between
+  // comment and declaration.
+  if (Text.find_last_of(";{}#@") != StringRef::npos)
     return nullptr;
 
   return CommentBeforeDecl;

clang/lib/CodeGen/BackendUtil.cpp

@@ -76,6 +76,7 @@
 #include "llvm/Transforms/Instrumentation/LowerAllowCheckPass.h"
 #include "llvm/Transforms/Instrumentation/MemProfiler.h"
 #include "llvm/Transforms/Instrumentation/MemorySanitizer.h"
+#include "llvm/Transforms/Instrumentation/NumericalStabilitySanitizer.h"
 #include "llvm/Transforms/Instrumentation/PGOInstrumentation.h"
 #include "llvm/Transforms/Instrumentation/SanitizerBinaryMetadata.h"
 #include "llvm/Transforms/Instrumentation/SanitizerCoverage.h"
@@ -707,6 +708,9 @@ static void addSanitizers(const Triple &TargetTriple,
       MPM.addPass(createModuleToFunctionPassAdaptor(ThreadSanitizerPass()));
     }
 
+    if (LangOpts.Sanitize.has(SanitizerKind::NumericalStability))
+      MPM.addPass(NumericalStabilitySanitizerPass());
+
     auto ASanPass = [&](SanitizerMask Mask, bool CompileKernel) {
       if (LangOpts.Sanitize.has(Mask)) {
         bool UseGlobalGC = asanUseGlobalsGC(TargetTriple, CodeGenOpts);

clang/lib/Headers/amxcomplexintrin.h

@@ -107,24 +107,6 @@
 ///    The 2nd source tile. Max size is 1024 Bytes.
 #define _tile_cmmrlfp16ps(dst, a, b) __builtin_ia32_tcmmrlfp16ps(dst, a, b)
 
-/// Perform matrix multiplication of two tiles containing complex elements and
-///    accumulate the results into a packed single precision tile.
-///
-/// \param m
-///    The number of rows in the first tile and the number of rows in the result
-///    tile.
-/// \param n
-///    The number of columns in the second tile and the number of columns in the
-///    result tile.
-/// \param k
-///    The number of columns in the first tile and the number of rows in the
-///    second tile.
-/// \param dst
-///    Pointer to the destination tile where the result will be stored.
-/// \param src1
-///    Pointer to the first source tile.
-/// \param src2
-///    Pointer to the second source tile.
 static __inline__ _tile1024i __DEFAULT_FN_ATTRS_COMPLEX
 _tile_cmmimfp16ps_internal(unsigned short m, unsigned short n, unsigned short k,
                            _tile1024i dst, _tile1024i src1, _tile1024i src2) {

clang/lib/Headers/ia32intrin.h

@@ -533,13 +533,6 @@ __rdtscp(unsigned int *__A) {
 /// \see __rdpmc
 #define _rdpmc(A) __rdpmc(A)
 
-/// Invalidates the contents of the processor's internal caches.
-///    This function writes back and invalidates all modified cache lines in
-///    the processor.
-///
-/// \headerfile <x86intrin.h>
-///
-/// This intrinsic corresponds to the \c WBINVD instruction.
 static __inline__ void __DEFAULT_FN_ATTRS
 _wbinvd(void) {
   __builtin_ia32_wbinvd();

clang/lib/Sema/SemaHLSL.cpp

@@ -671,30 +671,58 @@ void DiagnoseHLSLAvailability::HandleFunctionOrMethodRef(FunctionDecl *FD,
 
 void DiagnoseHLSLAvailability::RunOnTranslationUnit(
     const TranslationUnitDecl *TU) {
+
   // Iterate over all shader entry functions and library exports, and for those
   // that have a body (definiton), run diag scan on each, setting appropriate
   // shader environment context based on whether it is a shader entry function
-  // or an exported function.
-  for (auto &D : TU->decls()) {
-    const FunctionDecl *FD = llvm::dyn_cast<FunctionDecl>(D);
-    if (!FD || !FD->isThisDeclarationADefinition())
-      continue;
+  // or an exported function. Exported functions can be in namespaces and in
+  // export declarations so we need to scan those declaration contexts as well.
+  llvm::SmallVector<const DeclContext *, 8> DeclContextsToScan;
+  DeclContextsToScan.push_back(TU);
+
+  while (!DeclContextsToScan.empty()) {
+    const DeclContext *DC = DeclContextsToScan.pop_back_val();
+    for (auto &D : DC->decls()) {
+      // do not scan implicit declaration generated by the implementation
+      if (D->isImplicit())
+        continue;
+
+      // for namespace or export declaration add the context to the list to be
+      // scanned later
+      if (llvm::dyn_cast<NamespaceDecl>(D) || llvm::dyn_cast<ExportDecl>(D)) {
+        DeclContextsToScan.push_back(llvm::dyn_cast<DeclContext>(D));
+        continue;
+      }
 
-    // shader entry point
-    auto ShaderAttr = FD->getAttr<HLSLShaderAttr>();
-    if (ShaderAttr) {
-      SetShaderStageContext(ShaderAttr->getType());
-      RunOnFunction(FD);
-      continue;
-    }
-    // exported library function with definition
-    // FIXME: tracking issue #92073
-#if 0
-    if (FD->getFormalLinkage() == Linkage::External) {
-      SetUnknownShaderStageContext();
-      RunOnFunction(FD);
+      // skip over other decls or function decls without body
+      const FunctionDecl *FD = llvm::dyn_cast<FunctionDecl>(D);
+      if (!FD || !FD->isThisDeclarationADefinition())
+        continue;
+
+      // shader entry point
+      if (HLSLShaderAttr *ShaderAttr = FD->getAttr<HLSLShaderAttr>()) {
+        SetShaderStageContext(ShaderAttr->getType());
+        RunOnFunction(FD);
+        continue;
+      }
+      // exported library function
+      // FIXME: replace this loop with external linkage check once issue #92071
+      // is resolved
+      bool isExport = FD->isInExportDeclContext();
+      if (!isExport) {
+        for (const auto *Redecl : FD->redecls()) {
+          if (Redecl->isInExportDeclContext()) {
+            isExport = true;
+            break;
+          }
+        }
+      }
+      if (isExport) {
+        SetUnknownShaderStageContext();
+        RunOnFunction(FD);
+        continue;
+      }
     }
-#endif
   }
 }
 
@@ -707,8 +735,7 @@ void DiagnoseHLSLAvailability::RunOnFunction(const FunctionDecl *FD) {
     // For any CallExpr found during the traversal add it's callee to the top of
     // the stack to be processed next. Functions already processed are stored in
     // ScannedDecls.
-    const FunctionDecl *FD = DeclsToScan.back();
-    DeclsToScan.pop_back();
+    const FunctionDecl *FD = DeclsToScan.pop_back_val();
 
     // Decl was already scanned
     const unsigned ScannedStages = GetScannedStages(FD);

clang/test/CodeGen/no-skipped-passes-O0-opt-bisect.c

@@ -9,6 +9,7 @@
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -O0 %s -fdebug-pass-manager -emit-llvm -o /dev/null -fsanitize=local-bounds 2>&1 | FileCheck %s
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -O0 %s -fdebug-pass-manager -emit-llvm -o /dev/null -fsanitize=dataflow 2>&1 | FileCheck %s
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -O0 %s -fdebug-pass-manager -emit-llvm -o /dev/null -fsanitize-coverage-trace-pc-guard 2>&1 | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-linux-gnu -O0 %s -fdebug-pass-manager -emit-llvm -o /dev/null -fsanitize=numerical 2>&1 | FileCheck %s
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -O0 %s -fdebug-pass-manager -emit-llvm -o /dev/null -fmemory-profile 2>&1 | FileCheck %s
 
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -mllvm -opt-bisect-limit=0 %s -fdebug-pass-manager -emit-llvm -o /dev/null 2>&1 | FileCheck %s

clang/test/CodeGen/nsan-basic.c

@@ -0,0 +1,7 @@
+// RUN: %clang_cc1 -triple x86_64 -emit-llvm -o - -fsanitize=numerical %s | FileCheck %s
+
+// CHECK: Function Attrs: noinline nounwind optnone sanitize_numerical_stability
+float add(float x, float y) {
+  float z = x + y;
+  return z;
+}

clang/test/CodeGen/sanitizer-module-constructor.c

@@ -1,6 +1,7 @@
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -fsanitize=address -O3 -emit-llvm -fdebug-pass-manager -o - %s 2>&1 | FileCheck %s
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -fsanitize=thread -O3 -emit-llvm -fdebug-pass-manager -o - %s 2>&1 | FileCheck %s
 // RUN: %clang_cc1 -triple x86_64-linux-gnu -fsanitize=memory -O3 -emit-llvm -fdebug-pass-manager -o - %s 2>&1 | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-linux-gnu -fsanitize=numerical -O3 -emit-llvm -fdebug-pass-manager -o - %s 2>&1 | FileCheck %s
 
 // This is regression test for PR42877
 

clang/test/Index/annotate-comments.cpp

@@ -204,9 +204,9 @@ void isdoxy45(void);
 /// Ggg. IS_DOXYGEN_END
 void isdoxy46(void);
 
-/// isdoxy47 IS_DOXYGEN_SINGLE
+/// IS_DOXYGEN_NOT_ATTACHED
 #define FOO
-void isdoxy47(void);
+void notdoxy47(void);
 
 /// IS_DOXYGEN_START Aaa bbb
 /// \param ccc
@@ -330,7 +330,6 @@ void isdoxy54(int);
 // CHECK: annotate-comments.cpp:185:6: FunctionDecl=isdoxy44:{{.*}} BriefComment=[IS_DOXYGEN_START Aaa bbb ccc.]
 // CHECK: annotate-comments.cpp:195:6: FunctionDecl=isdoxy45:{{.*}} BriefComment=[Ddd eee. Fff.]
 // CHECK: annotate-comments.cpp:205:6: FunctionDecl=isdoxy46:{{.*}} BriefComment=[Ddd eee. Fff.]
-// CHECK: annotate-comments.cpp:209:6: FunctionDecl=isdoxy47:{{.*}} isdoxy47 IS_DOXYGEN_SINGLE
 // CHECK: annotate-comments.cpp:214:6: FunctionDecl=isdoxy48:{{.*}} BriefComment=[IS_DOXYGEN_START Aaa bbb]
 // CHECK: annotate-comments.cpp:218:6: FunctionDecl=isdoxy49:{{.*}} BriefComment=[IS_DOXYGEN_START Aaa]
 // CHECK: annotate-comments.cpp:222:6: FunctionDecl=isdoxy50:{{.*}} BriefComment=[Returns ddd IS_DOXYGEN_END]

clang/test/SemaHLSL/Availability/avail-diag-default-lib.hlsl

@@ -110,6 +110,55 @@ class MyClass
   }
 };
 
+// Exported function without body, not used
+export void exportedFunctionUnused(float f);
+
+// Exported function with body, without export, not used
+void exportedFunctionUnused(float f) {
+  // expected-error@#exportedFunctionUnused_fx_call {{'fx' is only available on Shader Model 6.5 or newer}}
+  // expected-note@#fx {{'fx' has been marked as being introduced in Shader Model 6.5 here, but the deployment target is Shader Model 6.0}}
+  float A = fx(f); // #exportedFunctionUnused_fx_call
+
+  // API with shader-stage-specific availability in unused exported library function
+  // - no errors expected because the actual shader stage this function
+  // will be used in not known at this time
+  float B = fy(f);
+  float C = fz(f);
+}
+
+// Exported function with body - called from main() which is a compute shader entry point
+export void exportedFunctionUsed(float f) {
+  // expected-error@#exportedFunctionUsed_fx_call {{'fx' is only available on Shader Model 6.5 or newer}}
+  // expected-note@#fx {{'fx' has been marked as being introduced in Shader Model 6.5 here, but the deployment target is Shader Model 6.0}}
+  float A = fx(f); // #exportedFunctionUsed_fx_call
+
+  // expected-error@#exportedFunctionUsed_fy_call {{'fy' is only available in compute environment on Shader Model 6.5 or newer}}
+  // expected-note@#fy {{'fy' has been marked as being introduced in Shader Model 6.5 in compute environment here, but the deployment target is Shader Model 6.0 compute environment}}
+  float B = fy(f); // #exportedFunctionUsed_fy_call
+
+  // expected-error@#exportedFunctionUsed_fz_call {{'fz' is unavailable}}
+  // expected-note@#fz {{'fz' has been marked as being introduced in Shader Model 6.5 in mesh environment here, but the deployment target is Shader Model 6.0 compute environment}}
+  float C = fz(f); // #exportedFunctionUsed_fz_call
+}
+
+namespace A {
+  namespace B {
+    export {
+      void exportedFunctionInNS(float x) {
+        // expected-error@#exportedFunctionInNS_fx_call {{'fx' is only available on Shader Model 6.5 or newer}}
+        // expected-note@#fx {{'fx' has been marked as being introduced in Shader Model 6.5 here, but the deployment target is Shader Model 6.0}}
+        float A = fx(x); // #exportedFunctionInNS_fx_call
+
+        // API with shader-stage-specific availability in exported library function
+        // - no errors expected because the actual shader stage this function
+        // will be used in not known at this time
+        float B = fy(x);
+        float C = fz(x);
+      }
+    }
+  }
+}
+
 // Shader entry point without body
 [shader("compute")]
 [numthreads(4,1,1)]
@@ -126,5 +175,6 @@ float main() {
   float c = C.makeF();
   float d = test((float)1.0);
   float e = test((half)1.0);
+  exportedFunctionUsed(1.0f);
   return a * b * c;
 }

clang/test/SemaHLSL/Availability/avail-diag-relaxed-lib.hlsl

@@ -110,6 +110,37 @@ class MyClass
   }
 };
 
+// Exported function without body, not used
+export void exportedFunctionUnused(float f);
+
+// Exported function with body, without export, not used
+void exportedFunctionUnused(float f) {
+  // expected-warning@#exportedFunctionUnused_fx_call {{'fx' is only available on Shader Model 6.5 or newer}}
+  // expected-note@#fx {{'fx' has been marked as being introduced in Shader Model 6.5 here, but the deployment target is Shader Model 6.0}}
+  float A = fx(f); // #exportedFunctionUnused_fx_call
+
+  // API with shader-stage-specific availability in unused exported library function
+  // - no errors expected because the actual shader stage this function
+  // will be used in not known at this time
+  float B = fy(f);
+  float C = fz(f);
+}
+
+// Exported function with body - called from main() which is a compute shader entry point
+export void exportedFunctionUsed(float f) {
+  // expected-warning@#exportedFunctionUsed_fx_call {{'fx' is only available on Shader Model 6.5 or newer}}
+  // expected-note@#fx {{'fx' has been marked as being introduced in Shader Model 6.5 here, but the deployment target is Shader Model 6.0}}
+  float A = fx(f); // #exportedFunctionUsed_fx_call
+
+  // expected-warning@#exportedFunctionUsed_fy_call {{'fy' is only available in compute environment on Shader Model 6.5 or newer}}
+  // expected-note@#fy {{'fy' has been marked as being introduced in Shader Model 6.5 in compute environment here, but the deployment target is Shader Model 6.0 compute environment}}
+  float B = fy(f); // #exportedFunctionUsed_fy_call
+
+  // expected-warning@#exportedFunctionUsed_fz_call {{'fz' is unavailable}}
+  // expected-note@#fz {{'fz' has been marked as being introduced in Shader Model 6.5 in mesh environment here, but the deployment target is Shader Model 6.0 compute environment}}
+  float C = fz(f); // #exportedFunctionUsed_fz_call
+}
+
 // Shader entry point without body
 [shader("compute")]
 [numthreads(4,1,1)]
@@ -126,5 +157,6 @@ float main() {
   float c = C.makeF();
   float d = test((float)1.0);
   float e = test((half)1.0);
+  exportedFunctionUsed(1.0f);
   return a * b * c;
 }