Merge pull request #218 from AMD-Lightning-Internal/upstream_merge_202501201507

merge main into amd-staging

Author: ronlieb

.github/workflows/build-ci-container.yml

@@ -36,19 +36,22 @@ jobs:
           tag=`date +%s`
           container_name="ghcr.io/$GITHUB_REPOSITORY_OWNER/ci-ubuntu-22.04"
           echo "container-name=$container_name" >> $GITHUB_OUTPUT
+          echo "container-name-agent=$container_name-agent" >> $GITHUB_OUTPUT
           echo "container-name-tag=$container_name:$tag" >> $GITHUB_OUTPUT
+          echo "container-name-agent-tag=$container_name-agent:$tag" >> $GITHUB_OUTPUT
           echo "container-filename=$(echo $container_name:$tag  | sed -e 's/\//-/g' -e 's/:/-/g').tar" >> $GITHUB_OUTPUT
       - name: Build container
         working-directory: ./.github/workflows/containers/github-action-ci/
         run: |
-          podman build -t ${{ steps.vars.outputs.container-name-tag }} .
+          podman build --target ci-container -t ${{ steps.vars.outputs.container-name-tag }} .
+          podman build --target ci-container-agent -t ${{ steps.vars.outputs.container-name-agent-tag }} .
 
       # Save the container so we have it in case the push fails.  This also
       # allows us to separate the push step into a different job so we can
       # maintain minimal permissions while building the container.
       - name: Save container image
         run: |
-          podman save  ${{ steps.vars.outputs.container-name-tag }} >  ${{ steps.vars.outputs.container-filename }}
+          podman save  ${{ steps.vars.outputs.container-name-tag }} ${{ steps.vars.outputs.container-name-agent-tag }} >  ${{ steps.vars.outputs.container-filename }}
 
       - name: Upload container image
         uses: actions/upload-artifact@v4
@@ -86,3 +89,7 @@ jobs:
           podman login -u ${{ github.actor }} -p $GITHUB_TOKEN ghcr.io
           podman push ${{ needs.build-ci-container.outputs.container-name-tag }}
           podman push ${{ needs.build-ci-container.outputs.container-name }}:latest
+
+          podman tag ${{ needs.build-ci-container.outpus.container-name-agent-tag }} ${{ needs.build-ci-container.outputs.container-name-agent }}:latest
+          podman push ${{ needs.build-ci-container.outputs.container-name-agent-tag }}
+          podman push ${{ needs.build-ci-container.outputs.container-name-agent }}:latest

.github/workflows/containers/github-action-ci/Dockerfile

@@ -13,7 +13,8 @@ RUN apt-get update && \
     ninja-build \
     python3 \
     git \
-    curl
+    curl \
+    zlib1g-dev
 
 RUN curl -O -L https://github.com/llvm/llvm-project/archive/refs/tags/llvmorg-$LLVM_VERSION.tar.gz && tar -xf llvmorg-$LLVM_VERSION.tar.gz
 
@@ -38,7 +39,7 @@ RUN cmake -B ./build -G Ninja ./llvm \
 
 RUN ninja -C ./build stage2-clang-bolt stage2-install-distribution && ninja -C ./build install-distribution
 
-FROM base
+FROM base as ci-container
 
 COPY --from=stage1-toolchain $LLVM_SYSROOT $LLVM_SYSROOT
 
@@ -91,4 +92,15 @@ RUN adduser gha sudo
 RUN echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
 
 USER gha
+WORKDIR /home/gha
+
+FROM ci-container as ci-container-agent
+
+ENV GITHUB_RUNNER_VERSION=2.321.0
+
+RUN mkdir actions-runner && \
+    cd actions-runner && \
+    curl -O -L https://github.com/actions/runner/releases/download/v$GITHUB_RUNNER_VERSION/actions-runner-linux-x64-$GITHUB_RUNNER_VERSION.tar.gz && \
+    tar xzf ./actions-runner-linux-x64-$GITHUB_RUNNER_VERSION.tar.gz && \
+    rm ./actions-runner-linux-x64-$GITHUB_RUNNER_VERSION.tar.gz
 

clang/docs/LanguageExtensions.rst

@@ -434,6 +434,114 @@ __datasizeof
 ``__datasizeof`` behaves like ``sizeof``, except that it returns the size of the
 type ignoring tail padding.
 
+_BitInt, _ExtInt
+----------------
+
+Clang supports the C23 ``_BitInt(N)`` feature as an extension in older C modes
+and in C++. This type was previously implemented in Clang with the same
+semantics, but spelled ``_ExtInt(N)``. This spelling has been deprecated in
+favor of the standard type.
+
+Note: the ABI for ``_BitInt(N)`` is still in the process of being stabilized,
+so this type should not yet be used in interfaces that require ABI stability.
+
+C keywords supported in all language modes
+------------------------------------------
+
+Clang supports ``_Alignas``, ``_Alignof``, ``_Atomic``, ``_Complex``,
+``_Generic``, ``_Imaginary``, ``_Noreturn``, ``_Static_assert``,
+``_Thread_local``, and ``_Float16`` in all language modes with the C semantics.
+
+__alignof, __alignof__
+----------------------
+
+``__alignof`` and ``__alignof__`` return, in contrast to ``_Alignof`` and
+``alignof``, the preferred alignment of a type. This may be larger than the
+required alignment for improved performance.
+
+__extension__
+-------------
+
+``__extension__`` suppresses extension diagnostics in the statement it is
+prepended to.
+
+__auto_type
+-----------
+
+``__auto_type`` behaves the same as ``auto`` in C++11 but is available in all
+language modes.
+
+__imag, __imag__
+----------------
+
+``__imag`` and ``__imag__`` can be used to get the imaginary part of a complex
+value.
+
+__real, __real__
+----------------
+
+``__real`` and ``__real__`` can be used to get the real part of a complex value.
+
+__asm, __asm__
+--------------
+
+``__asm`` and ``__asm__`` are alternate spellings for ``asm``, but available in
+all language modes.
+
+__complex, __complex__
+----------------------
+
+``__complex`` and ``__complex__`` are alternate spellings for ``_Complex``.
+
+__const, __const__, __volatile, __volatile__, __restrict, __restrict__
+----------------------------------------------------------------------
+
+These are alternate spellings for their non-underscore counterparts, but are
+available in all langauge modes.
+
+__decltype
+----------
+
+``__decltype`` is an alternate spelling for ``decltype``, but is also available
+in C++ modes before C++11.
+
+__inline, __inline__
+--------------------
+
+``__inline`` and ``__inline__`` are alternate spellings for ``inline``, but are
+available in all language modes.
+
+__nullptr
+---------
+
+``__nullptr`` is an alternate spelling for ``nullptr``, but is also available in
+C++ modes prior to C++11. Note that it's currently not availbale in C despite
+C23 having support for ``nullptr``.
+
+__signed, __signed__
+--------------------
+
+``__signed`` and ``__signed__`` are alternate spellings for ``signed``.
+``__unsigned`` and ``__unsigned__`` are **not** supported.
+
+__typeof, __typeof__, __typeof_unqual, __typeof_unqual__
+--------------------------------------------------------
+
+``__typeof`` and ``__typeof__`` are alternate spellings for ``typeof``, but are
+available in all langauge modes. These spellings result in the operand,
+retaining all qualifiers.
+
+``__typeof_unqual`` and ``__typeof_unqual__`` are alternate spellings for the
+C23 ``typeof_unqual`` type specifier, but are available in all language modes.
+These spellings result in the type of the operand, stripping all qualifiers.
+
+__char16_t, __char32_t
+----------------------
+
+``__char16_t`` and ``__char32_t`` are alternate spellings for ``char16_t`` and
+``char32_t`` respectively, but are also available in C++ modes before C++11.
+They are only supported in C++. ``__char8_t`` is not available.
+
 ..
   FIXME: This should list all the keyword extensions
 
@@ -5785,17 +5893,6 @@ Examples are:
    # 60 "" 2 // return to "main.c"
    # 1 "/usr/ancient/header.h" 1 4 // Enter an implicit extern "C" header
 
-Extended Integer Types
-======================
-
-Clang supports the C23 ``_BitInt(N)`` feature as an extension in older C modes
-and in C++. This type was previously implemented in Clang with the same
-semantics, but spelled ``_ExtInt(N)``. This spelling has been deprecated in
-favor of the standard type.
-
-Note: the ABI for ``_BitInt(N)`` is still in the process of being stabilized,
-so this type should not yet be used in interfaces that require ABI stability.
-
 Intrinsics Support within Constant Expressions
 ==============================================
 

clang/include/clang/Basic/AttrDocs.td

@@ -121,11 +121,12 @@ def InitPriorityDocs : Documentation {
 In C++, the order in which global variables are initialized across translation
 units is unspecified, unlike the ordering within a single translation unit. The
 ``init_priority`` attribute allows you to specify a relative ordering for the
-initialization of objects declared at namespace scope in C++. The priority is
-given as an integer constant expression between 101 and 65535 (inclusive).
-Priorities outside of that range are reserved for use by the implementation. A
-lower value indicates a higher priority of initialization. Note that only the
-relative ordering of values is important. For example:
+initialization of objects declared at namespace scope in C++ within a single
+linked image on supported platforms. The priority is given as an integer constant
+expression between 101 and 65535 (inclusive). Priorities outside of that range are
+reserved for use by the implementation. A lower value indicates a higher priority
+of initialization. Note that only the relative ordering of values is important.
+For example:
 
 .. code-block:: c++
 
@@ -136,10 +137,16 @@ relative ordering of values is important. For example:
 ``Obj2`` will be initialized *before* ``Obj1`` despite the usual order of
 initialization being the opposite.
 
+Note that this attribute does not control the initialization order of objects
+across final linked image boundaries like shared objects and executables.
+
 On Windows, ``init_seg(compiler)`` is represented with a priority of 200 and
 ``init_seg(library)`` is represented with a priority of 400. ``init_seg(user)``
 uses the default 65535 priority.
 
+On MachO platforms, this attribute also does not control the order of initialization
+across translation units, where it only affects the order within a single TU.
+
 This attribute is only supported for C++ and Objective-C++ and is ignored in
 other language modes. Currently, this attribute is not implemented on z/OS.
   }];

clang/include/clang/Basic/BuiltinsSystemZ.def

@@ -286,6 +286,7 @@ TARGET_BUILTIN(__builtin_s390_vstrszf, "V16UcV4UiV4UiV16Uci*", "nc", "vector-enh
 TARGET_BUILTIN(__builtin_s390_vlbrh, "V8UsV8Us", "nc", "vector")
 TARGET_BUILTIN(__builtin_s390_vlbrf, "V4UiV4Ui", "nc", "vector")
 TARGET_BUILTIN(__builtin_s390_vlbrg, "V2ULLiV2ULLi", "nc", "vector")
+TARGET_BUILTIN(__builtin_s390_vlbrq, "ULLLiULLLi", "nc", "vector")
 
 // NNP-assist facility intrinsics.
 TARGET_BUILTIN(__builtin_s390_vclfnhs, "V4fV8UsIi", "nc", "nnp-assist")
@@ -294,5 +295,44 @@ TARGET_BUILTIN(__builtin_s390_vcrnfs, "V8UsV4fV4fIi", "nc", "nnp-assist")
 TARGET_BUILTIN(__builtin_s390_vcfn, "V8UsV8UsIi", "nc", "nnp-assist")
 TARGET_BUILTIN(__builtin_s390_vcnf, "V8UsV8UsIi", "nc", "nnp-assist")
 
+// Miscellaneous instruction extensions facility 4 intrinsics.
+TARGET_BUILTIN(__builtin_s390_bdepg, "ULiULiULi", "nc", "miscellaneous-extensions-4")
+TARGET_BUILTIN(__builtin_s390_bextg, "ULiULiULi", "nc", "miscellaneous-extensions-4")
+
+// Vector-enhancements facility 3 intrinsics.
+TARGET_BUILTIN(__builtin_s390_vgemb, "V16UcV8Us", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vgemh, "V8UsV16Uc", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vgemf, "V4UiV16Uc", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vgemg, "V2ULLiV16Uc", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vgemq, "ULLLiV16Uc", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vuplg, "SLLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vupllg, "ULLLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vuphg, "SLLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vuplhg, "ULLLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vavgq, "SLLLiSLLLiSLLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vavglq, "ULLLiULLLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_veval, "V16UcV16UcV16UcV16UcIi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmahg, "V2SLLiV2SLLiV2SLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmahq, "SLLLiSLLLiSLLLiSLLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmalhg, "V2ULLiV2ULLiV2ULLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmalhq, "ULLLiULLLiULLLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmaeg, "SLLLiV2SLLiV2SLLiSLLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmaleg, "ULLLiV2ULLiV2ULLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmaog, "SLLLiV2SLLiV2SLLiSLLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmalog, "ULLLiV2ULLiV2ULLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmhg, "V2SLLiV2SLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmhq, "SLLLiSLLLiSLLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmlhg, "V2ULLiV2ULLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmlhq, "ULLLiULLLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmeg, "SLLLiV2SLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmleg, "ULLLiV2ULLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmog, "SLLLiV2SLLiV2SLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vmlog, "ULLLiV2ULLiV2ULLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vceqqs, "SLLLiULLLiULLLii*", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vchqs, "SLLLiSLLLiSLLLii*", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vchlqs, "SLLLiULLLiULLLii*", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vclzq, "ULLLiULLLi", "nc", "vector-enhancements-3")
+TARGET_BUILTIN(__builtin_s390_vctzq, "ULLLiULLLi", "nc", "vector-enhancements-3")
+
 #undef BUILTIN
 #undef TARGET_BUILTIN

clang/lib/AST/ByteCode/ByteCodeEmitter.cpp

@@ -332,6 +332,12 @@ void emit(Program &P, std::vector<std::byte> &Code, const IntegralAP<true> &Val,
   emitSerialized(Code, Val, Success);
 }
 
+template <>
+void emit(Program &P, std::vector<std::byte> &Code, const FixedPoint &Val,
+          bool &Success) {
+  emitSerialized(Code, Val, Success);
+}
+
 template <typename... Tys>
 bool ByteCodeEmitter::emitOp(Opcode Op, const Tys &...Args,
                              const SourceInfo &SI) {

clang/lib/AST/ByteCode/Disasm.cpp

@@ -62,6 +62,12 @@ inline IntegralAP<true> ReadArg<IntegralAP<true>>(Program &P, CodePtr &OpPC) {
   return I;
 }
 
+template <> inline FixedPoint ReadArg<FixedPoint>(Program &P, CodePtr &OpPC) {
+  FixedPoint I = FixedPoint::deserialize(*OpPC);
+  OpPC += align(I.bytesToSerialize());
+  return I;
+}
+
 LLVM_DUMP_METHOD void Function::dump() const { dump(llvm::errs()); }
 
 LLVM_DUMP_METHOD void Function::dump(llvm::raw_ostream &OS) const {

clang/lib/AST/ByteCode/FixedPoint.h

@@ -91,6 +91,32 @@ class FixedPoint final {
     return ComparisonCategoryResult::Greater;
   }
 
+  size_t bytesToSerialize() const {
+    return sizeof(uint32_t) + (V.getValue().getBitWidth() / CHAR_BIT);
+  }
+
+  void serialize(std::byte *Buff) const {
+    // Semantics followed by APInt.
+    uint32_t SemI = V.getSemantics().toOpaqueInt();
+    std::memcpy(Buff, &SemI, sizeof(SemI));
+
+    llvm::APInt API = V.getValue();
+    llvm::StoreIntToMemory(API, (uint8_t *)(Buff + sizeof(SemI)),
+                           bitWidth() / 8);
+  }
+
+  static FixedPoint deserialize(const std::byte *Buff) {
+    auto Sem = llvm::FixedPointSemantics::getFromOpaqueInt(
+        *reinterpret_cast<const uint32_t *>(Buff));
+    unsigned BitWidth = Sem.getWidth();
+    APInt I(BitWidth, 0ull, !Sem.isSigned());
+    llvm::LoadIntFromMemory(
+        I, reinterpret_cast<const uint8_t *>(Buff + sizeof(uint32_t)),
+        BitWidth / CHAR_BIT);
+
+    return FixedPoint(I, Sem);
+  }
+
   static bool neg(const FixedPoint &A, FixedPoint *R) {
     bool Overflow = false;
     *R = FixedPoint(A.V.negate(&Overflow));

clang/lib/AST/ByteCode/Interp.cpp

@@ -416,9 +416,11 @@ bool CheckRange(InterpState &S, CodePtr OpPC, const Pointer &Ptr,
                 AccessKinds AK) {
   if (!Ptr.isOnePastEnd())
     return true;
-  const SourceInfo &Loc = S.Current->getSource(OpPC);
-  S.FFDiag(Loc, diag::note_constexpr_access_past_end)
-      << AK << S.Current->getRange(OpPC);
+  if (S.getLangOpts().CPlusPlus) {
+    const SourceInfo &Loc = S.Current->getSource(OpPC);
+    S.FFDiag(Loc, diag::note_constexpr_access_past_end)
+        << AK << S.Current->getRange(OpPC);
+  }
   return false;
 }
 
@@ -538,7 +540,7 @@ bool CheckInitialized(InterpState &S, CodePtr OpPC, const Pointer &Ptr,
     return true;
 
   if (const auto *VD = Ptr.getDeclDesc()->asVarDecl();
-      VD && VD->hasGlobalStorage()) {
+      VD && (VD->isConstexpr() || VD->hasGlobalStorage())) {
     const SourceInfo &Loc = S.Current->getSource(OpPC);
     if (VD->getAnyInitializer()) {
       S.FFDiag(Loc, diag::note_constexpr_var_init_non_constant, 1) << VD;

clang/lib/AST/ByteCode/Interp.h

@@ -2693,6 +2693,10 @@ template <PrimType Name, class T = typename PrimConv<Name>::T>
 inline bool GetIntPtr(InterpState &S, CodePtr OpPC, const Descriptor *Desc) {
   const T &IntVal = S.Stk.pop<T>();
 
+  if (Desc)
+    S.CCEDiag(S.Current->getSource(OpPC), diag::note_constexpr_invalid_cast)
+        << 2 << S.getLangOpts().CPlusPlus;
+
   S.Stk.push<Pointer>(static_cast<uint64_t>(IntVal), Desc);
   return true;
 }
@@ -3070,6 +3074,13 @@ inline IntegralAP<true> ReadArg<IntegralAP<true>>(InterpState &S,
   return I;
 }
 
+template <>
+inline FixedPoint ReadArg<FixedPoint>(InterpState &S, CodePtr &OpPC) {
+  FixedPoint FP = FixedPoint::deserialize(*OpPC);
+  OpPC += align(FP.bytesToSerialize());
+  return FP;
+}
+
 } // namespace interp
 } // namespace clang