Merge branch 'main' into users/chapuni/cov/merge/region_segment

Author: chapuni

.github/workflows/commit-access-review.py

@@ -67,39 +67,47 @@ def check_manual_requests(
 ) -> list[str]:
     """
     Return a list of users who have been asked since ``start_date`` if they
-    want to keep their commit access.
+    want to keep their commit access or if they have applied for commit
+    access since ``start_date``
     """
+
     query = """
-        query ($query: String!) {
-          search(query: $query, type: ISSUE, first: 100) {
+        query ($query: String!, $after: String) {
+          search(query: $query, type: ISSUE, first: 100, after: $after) {
             nodes {
               ... on Issue {
-                body
-                comments (first: 100) {
-                  nodes {
-                    author {
-                      login
-                    }
-                  }
+                author {
+                  login
                 }
+                body
               }
             }
+            pageInfo {
+              hasNextPage
+              endCursor
+            }
           }
         }
         """
     formatted_start_date = start_date.strftime("%Y-%m-%dT%H:%M:%S")
     variables = {
-        "query": f"type:issue created:>{formatted_start_date} org:llvm repo:llvm-project label:infra:commit-access"
+        "query": f"type:issue created:>{formatted_start_date} org:llvm repo:llvm-project label:infra:commit-access,infra:commit-access-request"
     }
 
-    res_header, res_data = gh._Github__requester.graphql_query(
-        query=query, variables=variables
-    )
-    data = res_data["data"]
+    has_next_page = True
     users = []
-    for issue in data["search"]["nodes"]:
-        users.extend([user[1:] for user in re.findall("@[^ ,\n]+", issue["body"])])
-
+    while has_next_page:
+        res_header, res_data = gh._Github__requester.graphql_query(
+            query=query, variables=variables
+        )
+        data = res_data["data"]
+        for issue in data["search"]["nodes"]:
+            users.extend([user[1:] for user in re.findall("@[^ ,\n]+", issue["body"])])
+            if issue["author"]:
+                users.append(issue["author"]["login"])
+        has_next_page = data["search"]["pageInfo"]["hasNextPage"]
+        if has_next_page:
+            variables["after"] = data["search"]["pageInfo"]["endCursor"]
     return users
 
 

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

@@ -57,6 +57,7 @@ RUN apt-get update && \
     nodejs \
     perl-modules \
     python3-psutil \
+    sudo \
 
     # These are needed by the premerge pipeline. Pip is used to install
     # dependent python packages and ccache is used for build caching. File and
@@ -66,12 +67,28 @@ RUN apt-get update && \
     file \
     tzdata
 
+# Install sccache as it is needed by most of the project test workflows and
+# cannot be installed by the ccache action when executing as a non-root user.
+# TODO(boomanaiden154): This should be switched to being installed with apt
+# once we bump to Ubuntu 24.04.
+RUN curl -L 'https://github.com/mozilla/sccache/releases/download/v0.7.6/sccache-v0.7.6-x86_64-unknown-linux-musl.tar.gz' > /tmp/sccache.tar.gz && \
+    echo "2902a5e44c3342132f07b62e70cca75d9b23252922faf3b924f449808cc1ae58 /tmp/sccache.tar.gz" | sha256sum -c && \
+    tar xzf /tmp/sccache.tar.gz -O --wildcards '*/sccache' > '/usr/local/bin/sccache' && \
+    rm /tmp/sccache.tar.gz && \
+    chmod +x /usr/local/bin/sccache
+
 ENV LLVM_SYSROOT=$LLVM_SYSROOT
 ENV PATH=${LLVM_SYSROOT}/bin:${PATH}
 
 # Create a new user to avoid test failures related to a lack of expected
 # permissions issues in some tests. Set the user id to 1001 as that is the
 # user id that Github Actions uses to perform the checkout action.
 RUN useradd gha -u 1001 -m -s /bin/bash
+
+# Also add the user to passwordless sudoers so that we can install software
+# later on without having to rebuild the container.
+RUN adduser gha sudo
+RUN echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
+
 USER gha
 

.github/workflows/pr-code-format.yml

@@ -60,7 +60,7 @@ jobs:
       - name: Install clang-format
         uses: aminya/setup-cpp@v1
         with:
-          clangformat: 18.1.7
+          clangformat: 19.1.6
 
       - name: Setup Python env
         uses: actions/setup-python@v5

.github/workflows/premerge.yaml

@@ -31,6 +31,12 @@ jobs:
       - name: Setup ccache
         uses: hendrikmuhs/[email protected]
       - name: Build and Test
+        # Mark the job as a success even if the step fails so that people do
+        # not get notified while the new premerge pipeline is in an
+        # experimental state.
+        # TODO(boomanaiden154): Remove this once the pipeline is stable and we
+        # are ready for people to start recieving notifications.
+        continue-on-error: true
         run: |
           git config --global --add safe.directory '*'
 

bolt/lib/Core/BinaryEmitter.cpp

@@ -47,12 +47,16 @@ BreakFunctionNames("break-funcs",
   cl::cat(BoltCategory));
 
 static cl::list<std::string>
-FunctionPadSpec("pad-funcs",
-  cl::CommaSeparated,
-  cl::desc("list of functions to pad with amount of bytes"),
-  cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."),
-  cl::Hidden,
-  cl::cat(BoltCategory));
+    FunctionPadSpec("pad-funcs", cl::CommaSeparated,
+                    cl::desc("list of functions to pad with amount of bytes"),
+                    cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."),
+                    cl::Hidden, cl::cat(BoltCategory));
+
+static cl::list<std::string> FunctionPadBeforeSpec(
+    "pad-funcs-before", cl::CommaSeparated,
+    cl::desc("list of functions to pad with amount of bytes"),
+    cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."), cl::Hidden,
+    cl::cat(BoltCategory));
 
 static cl::opt<bool> MarkFuncs(
     "mark-funcs",
@@ -70,11 +74,11 @@ X86AlignBranchBoundaryHotOnly("x86-align-branch-boundary-hot-only",
   cl::init(true),
   cl::cat(BoltOptCategory));
 
-size_t padFunction(const BinaryFunction &Function) {
-  static std::map<std::string, size_t> FunctionPadding;
-
-  if (FunctionPadding.empty() && !FunctionPadSpec.empty()) {
-    for (std::string &Spec : FunctionPadSpec) {
+size_t padFunction(std::map<std::string, size_t> &FunctionPadding,
+                   const cl::list<std::string> &Spec,
+                   const BinaryFunction &Function) {
+  if (FunctionPadding.empty() && !Spec.empty()) {
+    for (const std::string &Spec : Spec) {
       size_t N = Spec.find(':');
       if (N == std::string::npos)
         continue;
@@ -94,6 +98,15 @@ size_t padFunction(const BinaryFunction &Function) {
   return 0;
 }
 
+size_t padFunctionBefore(const BinaryFunction &Function) {
+  static std::map<std::string, size_t> CacheFunctionPadding;
+  return padFunction(CacheFunctionPadding, FunctionPadBeforeSpec, Function);
+}
+size_t padFunctionAfter(const BinaryFunction &Function) {
+  static std::map<std::string, size_t> CacheFunctionPadding;
+  return padFunction(CacheFunctionPadding, FunctionPadSpec, Function);
+}
+
 } // namespace opts
 
 namespace {
@@ -319,6 +332,31 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
     Streamer.emitCodeAlignment(Function.getAlign(), &*BC.STI);
   }
 
+  if (size_t Padding = opts::padFunctionBefore(Function)) {
+    // Handle padFuncsBefore after the above alignment logic but before
+    // symbol addresses are decided.
+    if (!BC.HasRelocations) {
+      BC.errs() << "BOLT-ERROR: -pad-before-funcs is not supported in "
+                << "non-relocation mode\n";
+      exit(1);
+    }
+
+    // Preserve Function.getMinAlign().
+    if (!isAligned(Function.getMinAlign(), Padding)) {
+      BC.errs() << "BOLT-ERROR: user-requested " << Padding
+                << " padding bytes before function " << Function
+                << " is not a multiple of the minimum function alignment ("
+                << Function.getMinAlign().value() << ").\n";
+      exit(1);
+    }
+
+    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: padding before function " << Function
+                      << " with " << Padding << " bytes\n");
+
+    // Since the padding is not executed, it can be null bytes.
+    Streamer.emitFill(Padding, 0);
+  }
+
   MCContext &Context = Streamer.getContext();
   const MCAsmInfo *MAI = Context.getAsmInfo();
 
@@ -373,7 +411,7 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
   emitFunctionBody(Function, FF, /*EmitCodeOnly=*/false);
 
   // Emit padding if requested.
-  if (size_t Padding = opts::padFunction(Function)) {
+  if (size_t Padding = opts::padFunctionAfter(Function)) {
     LLVM_DEBUG(dbgs() << "BOLT-DEBUG: padding function " << Function << " with "
                       << Padding << " bytes\n");
     Streamer.emitFill(Padding, MAI->getTextAlignFillValue());

bolt/lib/Core/CMakeLists.txt

@@ -35,6 +35,7 @@ add_llvm_library(LLVMBOLTCore
   ParallelUtilities.cpp
   Relocation.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
   LINK_LIBS
   ${LLVM_PTHREAD_LIB}

bolt/lib/Passes/CMakeLists.txt

@@ -46,6 +46,7 @@ add_llvm_library(LLVMBOLTPasses
   VeneerElimination.cpp
   RetpolineInsertion.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   LINK_LIBS

bolt/lib/Passes/ReorderFunctions.cpp

@@ -28,7 +28,8 @@ extern cl::OptionCategory BoltOptCategory;
 extern cl::opt<unsigned> Verbosity;
 extern cl::opt<uint32_t> RandomSeed;
 
-extern size_t padFunction(const bolt::BinaryFunction &Function);
+extern size_t padFunctionBefore(const bolt::BinaryFunction &Function);
+extern size_t padFunctionAfter(const bolt::BinaryFunction &Function);
 
 extern cl::opt<bolt::ReorderFunctions::ReorderType> ReorderFunctions;
 cl::opt<bolt::ReorderFunctions::ReorderType> ReorderFunctions(
@@ -304,8 +305,10 @@ Error ReorderFunctions::runOnFunctions(BinaryContext &BC) {
                           return false;
                         if (B->isIgnored())
                           return true;
-                        const size_t PadA = opts::padFunction(*A);
-                        const size_t PadB = opts::padFunction(*B);
+                        const size_t PadA = opts::padFunctionBefore(*A) +
+                                            opts::padFunctionAfter(*A);
+                        const size_t PadB = opts::padFunctionBefore(*B) +
+                                            opts::padFunctionAfter(*B);
                         if (!PadA || !PadB) {
                           if (PadA)
                             return true;

bolt/lib/Profile/CMakeLists.txt

@@ -7,6 +7,7 @@ add_llvm_library(LLVMBOLTProfile
   YAMLProfileReader.cpp
   YAMLProfileWriter.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   LINK_COMPONENTS

bolt/lib/Rewrite/CMakeLists.txt

@@ -25,6 +25,7 @@ add_llvm_library(LLVMBOLTRewrite
   RewriteInstance.cpp
   SDTRewriter.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   LINK_LIBS

bolt/lib/RuntimeLibs/CMakeLists.txt

@@ -11,6 +11,7 @@ add_llvm_library(LLVMBOLTRuntimeLibs
   HugifyRuntimeLibrary.cpp
   InstrumentationRuntimeLibrary.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
   )
 

bolt/lib/Target/AArch64/CMakeLists.txt

@@ -19,6 +19,7 @@ endif()
 add_llvm_library(LLVMBOLTTargetAArch64
   AArch64MCPlusBuilder.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   DEPENDS

bolt/lib/Target/RISCV/CMakeLists.txt

@@ -20,6 +20,7 @@ endif()
 add_llvm_library(LLVMBOLTTargetRISCV
   RISCVMCPlusBuilder.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   DEPENDS

bolt/lib/Target/X86/CMakeLists.txt

@@ -21,6 +21,7 @@ add_llvm_library(LLVMBOLTTargetX86
   X86MCPlusBuilder.cpp
   X86MCSymbolizer.cpp
 
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   DEPENDS

bolt/lib/Utils/CMakeLists.txt

@@ -29,6 +29,8 @@ add_llvm_library(LLVMBOLTUtils
   CommandLineOpts.cpp
   Utils.cpp
   ${version_inc}
+
+  NO_EXPORT
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   LINK_LIBS

bolt/test/AArch64/pad-before-funcs.s

@@ -0,0 +1,48 @@
+# Test checks that --pad-before-funcs is working as expected.
+# It should be able to introduce a configurable offset for the _start symbol.
+# It should reject requests which don't obey the code alignment requirement.
+
+# Tests check inserting padding before _start; and additionally a test where
+# padding is inserted after start. In each case, check that the following
+# symbol ends up in the expected place as well.
+
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -Wl,-q -Wl,--section-start=.text=0x4000
+# RUN: llvm-bolt %t.exe -o %t.bolt.0 --pad-funcs-before=_start:0
+# RUN: llvm-bolt %t.exe -o %t.bolt.4 --pad-funcs-before=_start:4
+# RUN: llvm-bolt %t.exe -o %t.bolt.8 --pad-funcs-before=_start:8
+# RUN: llvm-bolt %t.exe -o %t.bolt.4.4 --pad-funcs-before=_start:4 --pad-funcs=_start:4
+# RUN: llvm-bolt %t.exe -o %t.bolt.4.8 --pad-funcs-before=_start:4 --pad-funcs=_start:8
+
+# RUN: not llvm-bolt %t.exe -o %t.bolt.8 --pad-funcs-before=_start:1 2>&1 | FileCheck --check-prefix=CHECK-BAD-ALIGN %s
+
+# CHECK-BAD-ALIGN: user-requested 1 padding bytes before function _start(*2) is not a multiple of the minimum function alignment (4).
+
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.0 | FileCheck --check-prefix=CHECK-0 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.4 | FileCheck --check-prefix=CHECK-4 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.8 | FileCheck --check-prefix=CHECK-8 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.4.4 | FileCheck --check-prefix=CHECK-4-4 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.4.8 | FileCheck --check-prefix=CHECK-4-8 %s
+
+# Trigger relocation mode in bolt.
+.reloc 0, R_AARCH64_NONE
+
+.section .text
+
+# CHECK-0: 0000000000400000 <_start>
+# CHECK-4: 0000000000400004 <_start>
+# CHECK-4-4: 0000000000400004 <_start>
+# CHECK-8: 0000000000400008 <_start>
+.globl _start
+_start:
+    ret
+
+# CHECK-0: 0000000000400004 <_subsequent>
+# CHECK-4: 0000000000400008 <_subsequent>
+# CHECK-4-4: 000000000040000c <_subsequent>
+# CHECK-4-8: 0000000000400010 <_subsequent>
+# CHECK-8: 000000000040000c <_subsequent>
+.globl _subsequent
+_subsequent:
+    ret

clang-tools-extra/clang-tidy/bugprone/UnhandledSelfAssignmentCheck.cpp

@@ -74,9 +74,11 @@ void UnhandledSelfAssignmentCheck::registerMatchers(MatchFinder *Finder) {
     // Matcher for standard smart pointers.
     const auto SmartPointerType = qualType(hasUnqualifiedDesugaredType(
         recordType(hasDeclaration(classTemplateSpecializationDecl(
-            hasAnyName("::std::shared_ptr", "::std::unique_ptr",
-                       "::std::weak_ptr", "::std::auto_ptr"),
-            templateArgumentCountIs(1))))));
+            anyOf(allOf(hasAnyName("::std::shared_ptr", "::std::weak_ptr",
+                                   "::std::auto_ptr"),
+                        templateArgumentCountIs(1)),
+                  allOf(hasName("::std::unique_ptr"),
+                        templateArgumentCountIs(2))))))));
 
     // We will warn only if the class has a pointer or a C array field which
     // probably causes a problem during self-assignment (e.g. first resetting