Merge branch 'main' into fix/112208

Author: a-tarasyuk

clang/lib/Driver/ToolChains/Cuda.cpp

@@ -634,8 +634,8 @@ void NVPTX::Linker::ConstructJob(Compilation &C, const JobAction &JA,
   std::vector<StringRef> Features;
   getNVPTXTargetFeatures(C.getDriver(), getToolChain().getTriple(), Args,
                          Features);
-  for (StringRef Feature : Features)
-    CmdArgs.append({"--feature", Args.MakeArgString(Feature)});
+  CmdArgs.push_back(
+      Args.MakeArgString("--plugin-opt=-mattr=" + llvm::join(Features, ",")));
 
   // Add paths for the default clang library path.
   SmallString<256> DefaultLibPath =

clang/test/Driver/cuda-cross-compiling.c

@@ -104,4 +104,4 @@
 // RUN: %clang -target nvptx64-nvidia-cuda --cuda-feature=+ptx63 -march=sm_52 -### %s 2>&1 \
 // RUN:   | FileCheck -check-prefix=FEATURE %s
 
-// FEATURE: clang-nvlink-wrapper{{.*}}"--feature" "+ptx63"
+// FEATURE: clang-nvlink-wrapper{{.*}}"--plugin-opt=-mattr=+ptx63"

clang/test/Format/dry-run-warning.cpp

@@ -0,0 +1,22 @@
+// RUN: echo '{' > %t.json
+// RUN: echo '  "married": true' >> %t.json
+// RUN: echo '}' >> %t.json
+
+// RUN: clang-format -n -style=LLVM %t.json 2>&1 | FileCheck %s -allow-empty
+
+// RUN: clang-format -n -style=LLVM < %t.json 2>&1 \
+// RUN:   | FileCheck %s -check-prefix=CHECK2 -strict-whitespace
+
+// RUN: echo '{' > %t.json
+// RUN: echo '  "married" : true' >> %t.json
+// RUN: echo '}' >> %t.json
+
+// RUN: clang-format -n -style=LLVM < %t.json 2>&1 | FileCheck %s -allow-empty
+
+// RUN: clang-format -n -style=LLVM %t.json 2>&1 \
+// RUN:   | FileCheck %s -check-prefix=CHECK2 -strict-whitespace
+
+// RUN: rm %t.json
+
+// CHECK-NOT: warning
+// CHECK2: warning: code should be clang-formatted

clang/tools/clang-format/ClangFormat.cpp

@@ -351,9 +351,6 @@ static void outputReplacementsXML(const Replacements &Replaces) {
 static bool
 emitReplacementWarnings(const Replacements &Replaces, StringRef AssumedFileName,
                         const std::unique_ptr<llvm::MemoryBuffer> &Code) {
-  if (Replaces.empty())
-    return false;
-
   unsigned Errors = 0;
   if (WarnFormat && !NoWarnFormat) {
     SourceMgr Mgr;
@@ -490,9 +487,11 @@ static bool format(StringRef FileName, bool ErrorOnIncompleteFormat = false) {
   Replacements Replaces = sortIncludes(*FormatStyle, Code->getBuffer(), Ranges,
                                        AssumedFileName, &CursorPosition);
 
+  const bool IsJson = FormatStyle->isJson();
+
   // To format JSON insert a variable to trick the code into thinking its
   // JavaScript.
-  if (FormatStyle->isJson() && !FormatStyle->DisableFormat) {
+  if (IsJson && !FormatStyle->DisableFormat) {
     auto Err = Replaces.add(tooling::Replacement(
         tooling::Replacement(AssumedFileName, 0, 0, "x = ")));
     if (Err)
@@ -510,9 +509,11 @@ static bool format(StringRef FileName, bool ErrorOnIncompleteFormat = false) {
   Replacements FormatChanges =
       reformat(*FormatStyle, *ChangedCode, Ranges, AssumedFileName, &Status);
   Replaces = Replaces.merge(FormatChanges);
-  if (OutputXML || DryRun) {
-    if (DryRun)
-      return emitReplacementWarnings(Replaces, AssumedFileName, Code);
+  if (DryRun) {
+    return Replaces.size() > (IsJson ? 1 : 0) &&
+           emitReplacementWarnings(Replaces, AssumedFileName, Code);
+  }
+  if (OutputXML) {
     outputXML(Replaces, FormatChanges, Status, Cursor, CursorPosition);
   } else {
     IntrusiveRefCntPtr<llvm::vfs::InMemoryFileSystem> InMemoryFileSystem(

clang/tools/clang-nvlink-wrapper/ClangNVLinkWrapper.cpp

@@ -344,7 +344,7 @@ Expected<std::unique_ptr<lto::LTO>> createLTO(const ArgList &Args) {
   Conf.RemarksHotnessThreshold = RemarksHotnessThreshold;
   Conf.RemarksFormat = RemarksFormat;
 
-  Conf.MAttrs = {Args.getLastArgValue(OPT_feature, "").str()};
+  Conf.MAttrs = llvm::codegen::getMAttrs();
   std::optional<CodeGenOptLevel> CGOptLevelOrNone =
       CodeGenOpt::parseLevel(Args.getLastArgValue(OPT_O, "2")[0]);
   assert(CGOptLevelOrNone && "Invalid optimization level");

clang/tools/clang-nvlink-wrapper/NVLinkOpts.td

@@ -47,9 +47,6 @@ def arch : Separate<["--", "-"], "arch">,
 def : Joined<["--", "-"], "plugin-opt=mcpu=">,
   Flags<[HelpHidden, WrapperOnlyOption]>, Alias<arch>;
 
-def feature : Separate<["--", "-"], "feature">, Flags<[WrapperOnlyOption]>,
-  HelpText<"Specify the '+ptx' freature to use for LTO.">;
-
 def g : Flag<["-"], "g">, HelpText<"Specify that this was a debug compile.">;
 def debug : Flag<["--"], "debug">, Alias<g>;
 

compiler-rt/lib/lsan/lsan_common.cpp

@@ -293,6 +293,27 @@ struct DirectMemoryAccessor {
   void Init(uptr begin, uptr end) {};
   void *LoadPtr(uptr p) const { return *reinterpret_cast<void **>(p); }
 };
+
+struct CopyMemoryAccessor {
+  void Init(uptr begin, uptr end) {
+    this->begin = begin;
+    buffer.clear();
+    buffer.resize(end - begin);
+    MemCpyAccessible(buffer.data(), reinterpret_cast<void *>(begin),
+                     buffer.size());
+  };
+
+  void *LoadPtr(uptr p) const {
+    uptr offset = p - begin;
+    CHECK_LE(offset + sizeof(void *), reinterpret_cast<uptr>(buffer.size()));
+    return *reinterpret_cast<void **>(offset +
+                                      reinterpret_cast<uptr>(buffer.data()));
+  }
+
+ private:
+  uptr begin;
+  InternalMmapVector<char> buffer;
+};
 }  // namespace
 
 // Scans the memory range, looking for byte patterns that point into allocator
@@ -373,18 +394,18 @@ void ScanGlobalRange(uptr begin, uptr end, Frontier *frontier) {
 }
 
 template <class Accessor>
-void ScanExtraStack(const InternalMmapVector<Range> &ranges, Frontier *frontier,
-                    Accessor &accessor) {
+void ScanRanges(const InternalMmapVector<Range> &ranges, Frontier *frontier,
+                const char *region_type, Accessor &accessor) {
   for (uptr i = 0; i < ranges.size(); i++) {
-    ScanForPointers(ranges[i].begin, ranges[i].end, frontier, "FAKE STACK",
+    ScanForPointers(ranges[i].begin, ranges[i].end, frontier, region_type,
                     kReachable, accessor);
   }
 }
 
 void ScanExtraStackRanges(const InternalMmapVector<Range> &ranges,
                           Frontier *frontier) {
   DirectMemoryAccessor accessor;
-  ScanExtraStack(ranges, frontier, accessor);
+  ScanRanges(ranges, frontier, "FAKE STACK", accessor);
 }
 
 #  if SANITIZER_FUCHSIA
@@ -478,7 +499,7 @@ static void ProcessThread(tid_t os_id, uptr sp,
     ScanForPointers(stack_begin, stack_end, frontier, "STACK", kReachable,
                     accessor);
     GetThreadExtraStackRangesLocked(os_id, &extra_ranges);
-    ScanExtraStack(extra_ranges, frontier, accessor);
+    ScanRanges(extra_ranges, frontier, "FAKE STACK", accessor);
   }
 
   if (flags()->use_tls) {
@@ -500,13 +521,14 @@ static void ProcessThread(tid_t os_id, uptr sp,
       }
     }
 #    if SANITIZER_ANDROID
+    extra_ranges.clear();
     auto *cb = +[](void *dtls_begin, void *dtls_end, uptr /*dso_idd*/,
                    void *arg) -> void {
-      ScanForPointers(
-          reinterpret_cast<uptr>(dtls_begin), reinterpret_cast<uptr>(dtls_end),
-          reinterpret_cast<Frontier *>(arg), "DTLS", kReachable, accessor);
+      reinterpret_cast<InternalMmapVector<Range> *>(arg)->push_back(
+          {reinterpret_cast<uptr>(dtls_begin),
+           reinterpret_cast<uptr>(dtls_end)});
     };
-
+    ScanRanges(extra_ranges, frontier, "DTLS", accessor);
     // FIXME: There might be a race-condition here (and in Bionic) if the
     // thread is suspended in the middle of updating its DTLS. IOWs, we
     // could scan already freed memory. (probably fine for now)
@@ -535,6 +557,7 @@ static void ProcessThread(tid_t os_id, uptr sp,
 static void ProcessThreads(SuspendedThreadsList const &suspended_threads,
                            Frontier *frontier, tid_t caller_tid,
                            uptr caller_sp) {
+  InternalMmapVector<tid_t> done_threads;
   InternalMmapVector<uptr> registers;
   InternalMmapVector<Range> extra_ranges;
   for (uptr i = 0; i < suspended_threads.ThreadCount(); i++) {
@@ -559,6 +582,25 @@ static void ProcessThreads(SuspendedThreadsList const &suspended_threads,
 
     DirectMemoryAccessor accessor;
     ProcessThread(os_id, sp, registers, extra_ranges, frontier, accessor);
+    if (flags()->use_detached)
+      done_threads.push_back(os_id);
+  }
+
+  if (flags()->use_detached) {
+    CopyMemoryAccessor accessor;
+    InternalMmapVector<tid_t> known_threads;
+    GetRunningThreadsLocked(&known_threads);
+    Sort(done_threads.data(), done_threads.size());
+    for (tid_t os_id : known_threads) {
+      registers.clear();
+      extra_ranges.clear();
+
+      uptr i = InternalLowerBound(done_threads, os_id);
+      if (i >= done_threads.size() || done_threads[i] != os_id) {
+        uptr sp = (os_id == caller_tid) ? caller_sp : 0;
+        ProcessThread(os_id, sp, registers, extra_ranges, frontier, accessor);
+      }
+    }
   }
 
   // Add pointers reachable from ThreadContexts

compiler-rt/lib/lsan/lsan_flags.inc

@@ -41,6 +41,8 @@ LSAN_FLAG(bool, use_ld_allocations, true,
 LSAN_FLAG(bool, use_unaligned, false, "Consider unaligned pointers valid.")
 LSAN_FLAG(bool, use_poisoned, false,
           "Consider pointers found in poisoned memory to be valid.")
+LSAN_FLAG(bool, use_detached, false,
+          "Scan threads even if attaching to them failed.")
 LSAN_FLAG(bool, log_pointers, false, "Debug logging")
 LSAN_FLAG(bool, log_threads, false, "Debug logging")
 LSAN_FLAG(int, tries, 1, "Debug option to repeat leak checking multiple times")

flang/include/flang/Optimizer/Dialect/CUF/CUFOps.h

@@ -12,6 +12,7 @@
 #include "flang/Optimizer/Dialect/CUF/Attributes/CUFAttr.h"
 #include "flang/Optimizer/Dialect/CUF/CUFDialect.h"
 #include "flang/Optimizer/Dialect/FIRType.h"
+#include "mlir/Dialect/LLVMIR/LLVMDialect.h"
 #include "mlir/IR/OpDefinition.h"
 
 #define GET_OP_CLASSES

flang/include/flang/Optimizer/Dialect/CUF/CUFOps.td

@@ -18,6 +18,7 @@ include "flang/Optimizer/Dialect/CUF/CUFDialect.td"
 include "flang/Optimizer/Dialect/CUF/Attributes/CUFAttr.td"
 include "flang/Optimizer/Dialect/FIRTypes.td"
 include "flang/Optimizer/Dialect/FIRAttr.td"
+include "mlir/Dialect/LLVMIR/LLVMOpBase.td"
 include "mlir/Interfaces/LoopLikeInterface.td"
 include "mlir/IR/BuiltinAttributes.td"
 
@@ -288,15 +289,30 @@ def cuf_KernelOp : cuf_Op<"kernel", [AttrSizedOperandSegments,
   let hasVerifier = 1;
 }
 
+def cuf_RegisterModuleOp : cuf_Op<"register_module", []> {
+  let summary = "Register a CUDA module";
+
+  let arguments = (ins
+    SymbolRefAttr:$name
+  );
+
+  let assemblyFormat = [{
+    $name attr-dict `->` type($modulePtr)
+  }];
+
+  let results = (outs LLVM_AnyPointer:$modulePtr);
+}
+
 def cuf_RegisterKernelOp : cuf_Op<"register_kernel", []> {
   let summary = "Register a CUDA kernel";
 
   let arguments = (ins
-    SymbolRefAttr:$name
+    SymbolRefAttr:$name,
+    LLVM_AnyPointer:$modulePtr
   );
 
   let assemblyFormat = [{
-    $name attr-dict
+    $name `(` $modulePtr `:` type($modulePtr) `)`attr-dict
   }];
 
   let hasVerifier = 1;