merge main into amd-staging (llvm#2967)

Author: z1_cciauto

clang/include/clang/Basic/SourceManager.h

@@ -767,6 +767,8 @@ class SourceManager : public RefCountedBase<SourceManager> {
   /// LastFileIDLookup records the last FileID looked up or created, because it
   /// is very common to look up many tokens from the same file.
   mutable FileID LastFileIDLookup;
+  mutable SourceLocation::UIntTy LastLookupStartOffset;
+  mutable SourceLocation::UIntTy LastLookupEndOffset; // exclude
 
   /// Holds information for \#line directives.
   ///
@@ -1901,9 +1903,8 @@ class SourceManager : public RefCountedBase<SourceManager> {
 
   FileID getFileID(SourceLocation::UIntTy SLocOffset) const {
     // If our one-entry cache covers this offset, just return it.
-    if (isOffsetInFileID(LastFileIDLookup, SLocOffset))
+    if (SLocOffset >= LastLookupStartOffset && SLocOffset < LastLookupEndOffset)
       return LastFileIDLookup;
-
     return getFileIDSlow(SLocOffset);
   }
 

clang/include/clang/Driver/Options.td

@@ -1143,6 +1143,10 @@ def fno_convergent_functions : Flag<["-"], "fno-convergent-functions">,
 
 // Common offloading options
 let Group = offload_Group in {
+def offload_targets_EQ : CommaJoined<["--"], "offload-targets=">,
+  Flags<[NoXarchOption]>, Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>,
+  HelpText<"Specify a list of target architectures to use for offloading.">;
+
 def offload_arch_EQ : CommaJoined<["--"], "offload-arch=">,
   Visibility<[ClangOption, FlangOption]>,
   HelpText<"Specify an offloading device architecture for CUDA, HIP, or OpenMP. (e.g. sm_35). "
@@ -3693,7 +3697,7 @@ def fopenmp_use_tls : Flag<["-"], "fopenmp-use-tls">, Group<f_Group>,
   Flags<[NoArgumentUnused, HelpHidden]>;
 def fnoopenmp_use_tls : Flag<["-"], "fnoopenmp-use-tls">, Group<f_Group>,
   Flags<[NoArgumentUnused, HelpHidden]>, Visibility<[ClangOption, CC1Option]>;
-def fopenmp_targets_EQ : CommaJoined<["-"], "fopenmp-targets=">,
+def fopenmp_targets_EQ : CommaJoined<["-"], "fopenmp-targets=">, Alias<offload_targets_EQ>,
   Flags<[NoXarchOption]>, Visibility<[ClangOption, CC1Option, FlangOption, FC1Option]>,
   HelpText<"Specify comma-separated list of triples OpenMP offloading targets to be supported">;
 def fopenmp_relocatable_target : Flag<["-"], "fopenmp-relocatable-target">,

clang/lib/Basic/SourceManager.cpp

@@ -335,6 +335,7 @@ void SourceManager::clearIDTables() {
   LastLineNoFileIDQuery = FileID();
   LastLineNoContentCache = nullptr;
   LastFileIDLookup = FileID();
+  LastLookupStartOffset = LastLookupEndOffset = 0;
 
   IncludedLocMap.clear();
   if (LineTable)
@@ -630,9 +631,11 @@ FileID SourceManager::createFileIDImpl(ContentCache &File, StringRef Filename,
   LocalSLocEntryTable.push_back(
       SLocEntry::get(NextLocalOffset,
                      FileInfo::get(IncludePos, File, FileCharacter, Filename)));
+  LastLookupStartOffset = NextLocalOffset;
   // We do a +1 here because we want a SourceLocation that means "the end of the
   // file", e.g. for the "no newline at the end of the file" diagnostic.
   NextLocalOffset += FileSize + 1;
+  LastLookupEndOffset = NextLocalOffset;
   updateSlocUsageStats();
 
   // Set LastFileIDLookup to the newly created file.  The next getFileID call is
@@ -802,8 +805,9 @@ FileID SourceManager::getFileIDSlow(SourceLocation::UIntTy SLocOffset) const {
 /// loaded one.
 FileID SourceManager::getFileIDLocal(SourceLocation::UIntTy SLocOffset) const {
   assert(SLocOffset < NextLocalOffset && "Bad function choice");
-  assert(SLocOffset >= LocalSLocEntryTable[0].getOffset() &&
+  assert(SLocOffset >= LocalSLocEntryTable[0].getOffset() && SLocOffset > 0 &&
          "Invalid SLocOffset");
+  assert(LastFileIDLookup.ID >= 0 && "Only cache local file sloc entry");
 
   // After the first and second level caches, I see two common sorts of
   // behavior: 1) a lot of searched FileID's are "near" the cached file
@@ -823,13 +827,11 @@ FileID SourceManager::getFileIDLocal(SourceLocation::UIntTy SLocOffset) const {
   unsigned LessIndex = 0;
   // upper bound of the search range.
   unsigned GreaterIndex = LocalSLocEntryTable.size();
-  if (LastFileIDLookup.ID >= 0) {
-    // Use the LastFileIDLookup to prune the search space.
-    if (LocalSLocEntryTable[LastFileIDLookup.ID].getOffset() < SLocOffset)
-      LessIndex = LastFileIDLookup.ID;
-    else
-      GreaterIndex = LastFileIDLookup.ID;
-  }
+  // Use the LastFileIDLookup to prune the search space.
+  if (LastLookupStartOffset < SLocOffset)
+    LessIndex = LastFileIDLookup.ID;
+  else
+    GreaterIndex = LastFileIDLookup.ID;
 
   // Find the FileID that contains this.
   unsigned NumProbes = 0;
@@ -840,7 +842,12 @@ FileID SourceManager::getFileIDLocal(SourceLocation::UIntTy SLocOffset) const {
       FileID Res = FileID::get(int(GreaterIndex));
       // Remember it.  We have good locality across FileID lookups.
       LastFileIDLookup = Res;
-      NumLinearScans += NumProbes+1;
+      LastLookupStartOffset = LocalSLocEntryTable[GreaterIndex].getOffset();
+      LastLookupEndOffset =
+          GreaterIndex + 1 >= LocalSLocEntryTable.size()
+              ? NextLocalOffset
+              : LocalSLocEntryTable[GreaterIndex + 1].getOffset();
+      NumLinearScans += NumProbes + 1;
       return Res;
     }
     if (++NumProbes == 8)
@@ -864,6 +871,8 @@ FileID SourceManager::getFileIDLocal(SourceLocation::UIntTy SLocOffset) const {
   // At this point, LessIndex is the index of the *first element greater than*
   // SLocOffset. The element we are actually looking for is the one immediately
   // before it.
+  LastLookupStartOffset = LocalSLocEntryTable[LessIndex - 1].getOffset();
+  LastLookupEndOffset = LocalSLocEntryTable[LessIndex].getOffset();
   return LastFileIDLookup = FileID::get(LessIndex - 1);
 }
 

clang/lib/Driver/Driver.cpp

@@ -999,7 +999,7 @@ void Driver::CreateOffloadingDeviceToolChains(Compilation &C,
                                         /*SpecificToolchain=*/true);
   } else if (IsHIP && !UseLLVMOffload) {
     if (auto *OMPTargetArg =
-            C.getInputArgs().getLastArg(options::OPT_fopenmp_targets_EQ)) {
+            C.getInputArgs().getLastArg(options::OPT_offload_targets_EQ)) {
       Diag(clang::diag::err_drv_unsupported_opt_for_language_mode)
           << OMPTargetArg->getSpelling() << "HIP";
       return;
@@ -1033,7 +1033,7 @@ void Driver::CreateOffloadingDeviceToolChains(Compilation &C,
       ((IsCuda || IsHIP) && UseLLVMOffload) ||
       (C.getInputArgs().hasFlag(options::OPT_fopenmp, options::OPT_fopenmp_EQ,
                                 options::OPT_fno_openmp, false) &&
-       (C.getInputArgs().hasArg(options::OPT_fopenmp_targets_EQ) ||
+       (C.getInputArgs().hasArg(options::OPT_offload_targets_EQ) ||
         C.getInputArgs().hasArg(options::OPT_offload_arch_EQ)));
   if (IsOpenMPOffloading) {
     // We expect that -fopenmp-targets is always used in conjunction with the
@@ -1049,7 +1049,7 @@ void Driver::CreateOffloadingDeviceToolChains(Compilation &C,
     // valid triple. Otherwise, if only --offload-arch= was specified we instead
     // attempt to derive the appropriate toolchains from the arguments.
     if (Arg *OpenMPTargets =
-            C.getInputArgs().getLastArg(options::OPT_fopenmp_targets_EQ)) {
+            C.getInputArgs().getLastArg(options::OPT_offload_targets_EQ)) {
       if (OpenMPTargets && !OpenMPTargets->getNumValues()) {
         Diag(clang::diag::warn_drv_empty_joined_argument)
             << OpenMPTargets->getAsString(C.getInputArgs());
@@ -1127,7 +1127,7 @@ void Driver::CreateOffloadingDeviceToolChains(Compilation &C,
         Diag(clang::diag::err_drv_failed_to_deduce_target_from_arch)
             << "native";
     }
-  } else if (C.getInputArgs().hasArg(options::OPT_fopenmp_targets_EQ)) {
+  } else if (C.getInputArgs().hasArg(options::OPT_offload_targets_EQ)) {
     Diag(clang::diag::err_drv_expecting_fopenmp_with_fopenmp_targets);
     return;
   }

clang/lib/Driver/ToolChain.cpp

@@ -1742,7 +1742,7 @@ llvm::opt::DerivedArgList *ToolChain::TranslateOpenMPTargetArgs(
       continue;
     }
     if (XOpenMPTargetNoTriple && XOpenMPTargetArg &&
-        Args.getAllArgValues(options::OPT_fopenmp_targets_EQ).size() != 1) {
+        Args.getAllArgValues(options::OPT_offload_targets_EQ).size() != 1) {
       getDriver().Diag(diag::err_drv_Xopenmp_target_missing_triple);
       continue;
     }

clang/lib/Driver/ToolChains/CommonArgs.cpp

@@ -1542,7 +1542,7 @@ void tools::addOpenMPHostOffloadingArgs(const Compilation &C,
 
   // For all the host OpenMP offloading compile jobs we need to pass the targets
   // information using -fopenmp-targets= option.
-  constexpr llvm::StringLiteral Targets("-fopenmp-targets=");
+  constexpr llvm::StringLiteral Targets("--offload-targets=");
 
   SmallVector<std::string> Triples;
   auto TCRange = C.getOffloadToolChains<Action::OFK_OpenMP>();

clang/lib/Frontend/CompilerInvocation.cpp

@@ -3962,7 +3962,7 @@ void CompilerInvocationBase::GenerateLangArgs(const LangOptions &Opts,
     llvm::interleave(
         Opts.OMPTargetTriples, OS,
         [&OS](const llvm::Triple &T) { OS << T.str(); }, ",");
-    GenerateArg(Consumer, OPT_fopenmp_targets_EQ, Targets);
+    GenerateArg(Consumer, OPT_offload_targets_EQ, Targets);
   }
 
   if (!Opts.OMPHostIRFile.empty())
@@ -4393,7 +4393,8 @@ bool CompilerInvocation::ParseLangArgs(LangOptions &Opts, ArgList &Args,
   Opts.OpenMPIRBuilder =
       Opts.OpenMP && Args.hasArg(options::OPT_fopenmp_enable_irbuilder);
   bool IsTargetSpecified =
-      Opts.OpenMPIsTargetDevice || Args.hasArg(options::OPT_fopenmp_targets_EQ);
+      Opts.OpenMPIsTargetDevice || Args.hasArg(options::OPT_offload_targets_EQ);
+
   if (Opts.OpenMP || Opts.OpenMPSimd) {
     if (int Version = getLastArgIntValue(
             Args, OPT_fopenmp_version_EQ,
@@ -4510,7 +4511,7 @@ bool CompilerInvocation::ParseLangArgs(LangOptions &Opts, ArgList &Args,
       options::OPT_fno_openmp_assume_no_nested_parallelism, isOFastUsed(Args));
 
   // Get the OpenMP target triples if any.
-  if (Arg *A = Args.getLastArg(options::OPT_fopenmp_targets_EQ)) {
+  if (Arg *A = Args.getLastArg(options::OPT_offload_targets_EQ)) {
     enum ArchPtrSize { Arch16Bit, Arch32Bit, Arch64Bit };
     auto getArchPtrSize = [](const llvm::Triple &T) {
       if (T.isArch16Bit())

clang/test/Driver/amdgpu-openmp-sanitize-options.c

@@ -54,11 +54,11 @@
 // NOXNACK: warning: ignoring '-fsanitize=address' option for offload arch 'gfx908' as it is not currently supported there. Use it with an offload arch containing 'xnack+' instead
 // XNACKNEG: warning: ignoring '-fsanitize=address' option for offload arch 'gfx908:xnack-' as it is not currently supported there. Use it with an offload arch containing 'xnack+' instead
 
-// HOSTSAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "x86_64-unknown-linux-gnu".* "-fopenmp".* "-fsanitize=address".* "-fopenmp-targets=amdgcn-amd-amdhsa".* "-x" "c".*}}
+// HOSTSAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "x86_64-unknown-linux-gnu".* "-fopenmp".* "-fsanitize=address".* "--offload-targets=amdgcn-amd-amdhsa".* "-x" "c".*}}
 
 // GPUSAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "amdgcn-amd-amdhsa" "-aux-triple" "x86_64-unknown-linux-gnu".* "-emit-llvm-bc".* "-mlink-bitcode-file" "[^"]*asanrtl.bc".* "-mlink-bitcode-file" "[^"]*ockl.bc".* "-target-cpu" "(gfx908|gfx900)".* "-fopenmp".* "-fsanitize=address".* "-x" "c".*}}
 // NOGPUSAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "amdgcn-amd-amdhsa" "-aux-triple" "x86_64-unknown-linux-gnu".* "-emit-llvm-bc".* "-target-cpu" "(gfx908|gfx900)".* "-fopenmp".* "-x" "c".*}}
 
 // SAN: {{"[^"]*clang-offload-packager[^"]*" "-o".* "--image=file=.*.bc,triple=amdgcn-amd-amdhsa,arch=gfx908(:xnack\-|:xnack\+)?,kind=openmp(,feature=(\-xnack|\+xnack))?"}}
-// SAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "x86_64-unknown-linux-gnu".* "-fopenmp".* "-fsanitize=address".* "-fopenmp-targets=amdgcn-amd-amdhsa".* "-x" "ir".*}}
+// SAN: {{"[^"]*clang[^"]*" "-cc1" "-triple" "x86_64-unknown-linux-gnu".* "-fopenmp".* "-fsanitize=address".* "--offload-targets=amdgcn-amd-amdhsa".* "-x" "ir".*}}
 // SAN: {{"[^"]*clang-linker-wrapper[^"]*".* "--host-triple=x86_64-unknown-linux-gnu".* "--linker-path=[^"]*".* "--whole-archive" "[^"]*(libclang_rt.asan_static.a|libclang_rt.asan_static-x86_64.a)".* "--whole-archive" "[^"]*(libclang_rt.asan.a|libclang_rt.asan-x86_64.a)".*}}

clang/test/Driver/hip-options.hip

@@ -118,7 +118,7 @@
 // RUN: not %clang --target=x86_64-unknown-linux-gnu -nogpuinc -nogpulib \
 // RUN:   --offload-arch=gfx906 -fopenmp=libomp -fopenmp-targets=amdgcn %s 2>&1 \
 // RUN:   | FileCheck -check-prefix=OMPTGT %s
-// OMPTGT: unsupported option '-fopenmp-targets=' for language mode 'HIP'
+// OMPTGT: unsupported option '--offload-targets=' for language mode 'HIP'
 
 // Check -Xoffload-linker option is passed to lld.
 

flang/lib/Frontend/CompilerInvocation.cpp

@@ -1262,7 +1262,7 @@ static bool parseOpenMPArgs(CompilerInvocation &res, llvm::opt::ArgList &args,
 
   // Get the OpenMP target triples if any.
   if (auto *arg =
-          args.getLastArg(clang::driver::options::OPT_fopenmp_targets_EQ)) {
+          args.getLastArg(clang::driver::options::OPT_offload_targets_EQ)) {
     enum ArchPtrSize { Arch16Bit, Arch32Bit, Arch64Bit };
     auto getArchPtrSize = [](const llvm::Triple &triple) {
       if (triple.isArch16Bit())