Merge branch 'main' into xegpu_changes

Author: charithaintc

.ci/metrics/metrics.py

@@ -24,6 +24,7 @@ class JobMetrics:
     status: int
     created_at_ns: int
     workflow_id: int
+    workflow_name: str
 
 
 @dataclass
@@ -43,40 +44,60 @@ def get_sampled_workflow_metrics(github_repo: github.Repository):
       Returns a list of GaugeMetric objects, containing the relevant metrics about
       the workflow
     """
+    queued_job_counts = {}
+    running_job_counts = {}
 
     # Other states are available (pending, waiting, etc), but the meaning
     # is not documented (See #70540).
     # "queued" seems to be the info we want.
-    queued_workflow_count = len(
-        [
-            x
-            for x in github_repo.get_workflow_runs(status="queued")
-            if x.name in WORKFLOWS_TO_TRACK
-        ]
-    )
-    running_workflow_count = len(
-        [
-            x
-            for x in github_repo.get_workflow_runs(status="in_progress")
-            if x.name in WORKFLOWS_TO_TRACK
-        ]
-    )
+    for queued_workflow in github_repo.get_workflow_runs(status="queued"):
+        if queued_workflow.name not in WORKFLOWS_TO_TRACK:
+            continue
+        for queued_workflow_job in queued_workflow.jobs():
+            job_name = queued_workflow_job.name
+            # Workflows marked as queued can potentially only have some jobs
+            # queued, so make sure to also count jobs currently in progress.
+            if queued_workflow_job.status == "queued":
+                if job_name not in queued_job_counts:
+                    queued_job_counts[job_name] = 1
+                else:
+                    queued_job_counts[job_name] += 1
+            elif queued_workflow_job.status == "in_progress":
+                if job_name not in running_job_counts:
+                    running_job_counts[job_name] = 1
+                else:
+                    running_job_counts[job_name] += 1
+
+    for running_workflow in github_repo.get_workflow_runs(status="in_progress"):
+        if running_workflow.name not in WORKFLOWS_TO_TRACK:
+            continue
+        for running_workflow_job in running_workflow.jobs():
+            job_name = running_workflow_job.name
+            if running_workflow_job.status != "in_progress":
+                continue
+
+            if job_name not in running_job_counts:
+                running_job_counts[job_name] = 1
+            else:
+                running_job_counts[job_name] += 1
 
     workflow_metrics = []
-    workflow_metrics.append(
-        GaugeMetric(
-            "workflow_queue_size",
-            queued_workflow_count,
-            time.time_ns(),
+    for queued_job in queued_job_counts:
+        workflow_metrics.append(
+            GaugeMetric(
+                f"workflow_queue_size_{queued_job}",
+                queued_job_counts[queued_job],
+                time.time_ns(),
+            )
         )
-    )
-    workflow_metrics.append(
-        GaugeMetric(
-            "running_workflow_count",
-            running_workflow_count,
-            time.time_ns(),
+    for running_job in running_job_counts:
+        workflow_metrics.append(
+            GaugeMetric(
+                f"running_workflow_count_{running_job}",
+                running_job_counts[running_job],
+                time.time_ns(),
+            )
         )
-    )
     # Always send a hearbeat metric so we can monitor is this container is still able to log to Grafana.
     workflow_metrics.append(
         GaugeMetric("metrics_container_heartbeat", 1, time.time_ns())
@@ -157,7 +178,7 @@ def get_per_workflow_metrics(
                 # longer in a testing state and we can directly assert the workflow
                 # result.
                 for step in workflow_job.steps:
-                    if step.conclusion != "success":
+                    if step.conclusion != "success" and step.conclusion != "skipped":
                         job_result = 0
                         break
 
@@ -179,6 +200,7 @@ def get_per_workflow_metrics(
                     job_result,
                     created_at_ns,
                     workflow_run.id,
+                    workflow_run.name,
                 )
             )
 
@@ -235,8 +257,6 @@ def upload_metrics(workflow_metrics, metrics_userid, api_key):
 def main():
     # Authenticate with Github
     auth = Auth.Token(os.environ["GITHUB_TOKEN"])
-    github_object = Github(auth=auth)
-    github_repo = github_object.get_repo("llvm/llvm-project")
 
     grafana_api_key = os.environ["GRAFANA_API_KEY"]
     grafana_metrics_userid = os.environ["GRAFANA_METRICS_USERID"]
@@ -248,20 +268,19 @@ def main():
     # Enter the main loop. Every five minutes we wake up and dump metrics for
     # the relevant jobs.
     while True:
+        github_object = Github(auth=auth)
+        github_repo = github_object.get_repo("llvm/llvm-project")
+
         current_metrics = get_per_workflow_metrics(github_repo, workflows_to_track)
         current_metrics += get_sampled_workflow_metrics(github_repo)
-        # Always send a hearbeat metric so we can monitor is this container is still able to log to Grafana.
-        current_metrics.append(
-            GaugeMetric("metrics_container_heartbeat", 1, time.time_ns())
-        )
 
         upload_metrics(current_metrics, grafana_metrics_userid, grafana_api_key)
         print(f"Uploaded {len(current_metrics)} metrics", file=sys.stderr)
 
         for workflow_metric in reversed(current_metrics):
             if isinstance(workflow_metric, JobMetrics):
                 workflows_to_track[
-                    workflow_metric.job_name
+                    workflow_metric.workflow_name
                 ] = workflow_metric.workflow_id
 
         time.sleep(SCRAPE_INTERVAL_SECONDS)

.github/new-prs-labeler.yml

@@ -499,6 +499,7 @@ clang:static analyzer:
   - clang/tools/scan-build/**
   - clang/utils/analyzer/**
   - clang/docs/analyzer/**
+  - clang/test/Analysis/**
 
 pgo:
   - llvm/lib/Transforms/Instrumentation/CGProfile.cpp

.github/workflows/libcxx-build-and-test.yaml

@@ -37,7 +37,7 @@ jobs:
   stage1:
     if: github.repository_owner == 'llvm'
     runs-on: libcxx-self-hosted-linux
-    container: ghcr.io/llvm/libcxx-linux-builder:d8a0709b1090350a7fe3604d8ab78c7d62f10698
+    container: ghcr.io/llvm/libcxx-linux-builder:b319dfef21f6c7b0bc6a356d6b9f41a3b3b98ae9
     continue-on-error: false
     strategy:
       fail-fast: false
@@ -48,8 +48,8 @@ jobs:
           'generic-cxx26',
           'generic-modules'
         ]
-        cc: [  'clang-20' ]
-        cxx: [ 'clang++-20' ]
+        cc: [  'clang-21' ]
+        cxx: [ 'clang++-21' ]
         include:
           - config: 'generic-gcc'
             cc: 'gcc-14'
@@ -75,7 +75,7 @@ jobs:
   stage2:
     if: github.repository_owner == 'llvm'
     runs-on: libcxx-self-hosted-linux
-    container: ghcr.io/llvm/libcxx-linux-builder:d8a0709b1090350a7fe3604d8ab78c7d62f10698
+    container: ghcr.io/llvm/libcxx-linux-builder:b319dfef21f6c7b0bc6a356d6b9f41a3b3b98ae9
     needs: [ stage1 ]
     continue-on-error: false
     strategy:
@@ -88,18 +88,22 @@ jobs:
           'generic-cxx20',
           'generic-cxx23'
         ]
-        cc: [ 'clang-20' ]
-        cxx: [ 'clang++-20' ]
+        cc: [ 'clang-21' ]
+        cxx: [ 'clang++-21' ]
         include:
           - config: 'generic-gcc-cxx11'
             cc: 'gcc-14'
             cxx: 'g++-14'
-          - config: 'generic-cxx23'
-            cc: 'clang-18'
-            cxx: 'clang++-18'
+          - config: 'generic-cxx26'
+            cc: 'clang-20'
+            cxx: 'clang++-20'
           - config: 'generic-cxx26'
             cc: 'clang-19'
             cxx: 'clang++-19'
+          # Release transition
+          - config: 'generic-cxx23'
+            cc: 'clang-18'
+            cxx: 'clang++-18'
     steps:
       - uses: actions/checkout@v4
       - name: ${{ matrix.config }}
@@ -163,14 +167,14 @@ jobs:
         - config: 'generic-msan'
           machine: libcxx-self-hosted-linux
     runs-on: ${{ matrix.machine }}
-    container: ghcr.io/llvm/libcxx-linux-builder:d8a0709b1090350a7fe3604d8ab78c7d62f10698
+    container: ghcr.io/llvm/libcxx-linux-builder:b319dfef21f6c7b0bc6a356d6b9f41a3b3b98ae9
     steps:
       - uses: actions/checkout@v4
       - name: ${{ matrix.config }}
         run: libcxx/utils/ci/run-buildbot ${{ matrix.config }}
         env:
-          CC: clang-20
-          CXX: clang++-20
+          CC: clang-21
+          CXX: clang++-21
       - uses: actions/upload-artifact@26f96dfa697d77e81fd5907df203aa23a56210a8 # v4.3.0
         if: always()
         with:

.github/workflows/premerge.yaml

@@ -14,8 +14,6 @@ on:
       # do this is that it allows us to take advantage of concurrency groups
       # to cancel in progress CI jobs whenever the PR is closed.
       - closed
-    paths:
-      - .github/workflows/premerge.yaml
   push:
     branches:
       - 'main'
@@ -144,7 +142,7 @@ jobs:
           call C:\\BuildTools\\Common7\\Tools\\VsDevCmd.bat -arch=amd64 -host_arch=amd64
           bash .ci/monolithic-windows.sh "${{ steps.vars.outputs.windows-projects }}" "${{ steps.vars.outputs.windows-check-targets }}"
 
-  permerge-check-macos:
+  premerge-check-macos:
     runs-on: macos-14
     if: >-
       github.repository_owner == 'llvm' &&

.github/workflows/release-binaries.yml

@@ -138,6 +138,11 @@ jobs:
             arches=arm64
           else
             arches=x86_64
+            # Disable Flang builds on macOS x86_64.  The FortranLower library takes
+            # 2-3 hours to build on macOS, much slower than on Linux.
+            # The long build time causes the release build to time out on x86_64,
+            # so we need to disable flang there.
+            target_cmake_flags="$target_cmake_flags -DLLVM_RELEASE_ENABLE_PROJECTS='clang;lld;lldb;clang-tools-extra;bolt;polly;mlir'"
           fi
           target_cmake_flags="$target_cmake_flags -DBOOTSTRAP_BOOTSTRAP_DARWIN_osx_ARCHS=$arches -DBOOTSTRAP_BOOTSTRAP_DARWIN_osx_BUILTIN_ARCHS=$arches"
         fi

bolt/CMakeLists.txt

@@ -202,3 +202,11 @@ endif()
 
 configure_file(${CMAKE_CURRENT_SOURCE_DIR}/include/bolt/RuntimeLibs/RuntimeLibraryVariables.inc.in
                ${CMAKE_CURRENT_BINARY_DIR}/include/bolt/RuntimeLibs/RuntimeLibraryVariables.inc @ONLY)
+
+set(BOLT_ENUM_TARGETS "")
+foreach(t ${BOLT_TARGETS_TO_BUILD})
+  set(BOLT_ENUM_TARGETS "${BOLT_ENUM_TARGETS}BOLT_TARGET(${t})\n")
+endforeach(t)
+
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/include/bolt/Core/TargetConfig.def.in
+               ${CMAKE_CURRENT_BINARY_DIR}/include/bolt/Core/TargetConfig.def @ONLY)

bolt/include/bolt/Core/TargetConfig.def.in

@@ -0,0 +1,23 @@
+//===-- TargetConfig.def.in - Information about available targets ---------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file is configured by the build system to define the available bolt
+// targets.
+//
+// The variant of this file not ending with .in has been autogenerated by the
+// LLVM build. Do not edit!
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef BOLT_TARGET
+#  error Please define the macro BOLT_TARGET(TargetName)
+#endif
+
+@BOLT_ENUM_TARGETS@
+
+#undef BOLT_TARGET

bolt/lib/Core/BinaryFunction.cpp

@@ -498,6 +498,11 @@ void BinaryFunction::print(raw_ostream &OS, std::string Annotation) {
     if (!IslandOffset)
       return;
 
+    // Print label if it exists at this offset.
+    if (const BinaryData *BD =
+            BC.getBinaryDataAtAddress(getAddress() + *IslandOffset))
+      OS << BD->getName() << ":\n";
+
     const size_t IslandSize = getSizeOfDataInCodeAt(*IslandOffset);
     BC.printData(OS, BC.extractData(getAddress() + *IslandOffset, IslandSize),
                  *IslandOffset);
@@ -1066,7 +1071,7 @@ size_t BinaryFunction::getSizeOfDataInCodeAt(uint64_t Offset) const {
   auto Iter = Islands->CodeOffsets.upper_bound(Offset);
   if (Iter != Islands->CodeOffsets.end())
     return *Iter - Offset;
-  return getSize() - Offset;
+  return getMaxSize() - Offset;
 }
 
 std::optional<uint64_t>

bolt/lib/RuntimeLibs/InstrumentationRuntimeLibrary.cpp

@@ -219,7 +219,7 @@ void InstrumentationRuntimeLibrary::link(
   }
   outs() << "BOLT-INFO: output linked against instrumentation runtime "
             "library, lib entry point is 0x"
-         << Twine::utohexstr(RuntimeFiniAddress) << "\n";
+         << Twine::utohexstr(RuntimeStartAddress) << "\n";
   outs() << "BOLT-INFO: clear procedure is 0x"
          << Twine::utohexstr(
                 Linker.lookupSymbol("__bolt_instr_clear_counters").value_or(0))

bolt/test/AArch64/data-in-code.s

@@ -7,7 +7,8 @@
 ## Check disassembly of BOLT input.
 # RUN: llvm-objdump %t.exe -d | FileCheck %s
 
-# RUN: llvm-bolt %t.exe -o %t.bolt --print-disasm | FileCheck %s
+# RUN: llvm-bolt %t.exe -o %t.bolt --print-disasm \
+# RUN:   | FileCheck %s --check-prefixes CHECK,CHECK-BOLT-ONLY
 
 .text
 .balign 4
@@ -16,16 +17,21 @@
 .type _start, %function
 _start:
   mov x0, #0x0
+  ldr x1, .L1
   .word 0x4f82e010
   ret
+.size _start, .-_start
+.L1:
   .byte 0x0, 0xff, 0x42
 # CHECK-LABEL: _start
 # CHECK:        mov x0, #0x0
+# CHECK-NEXT:   ldr x1
+# CHECK-BOLT-ONLY-SAME: ISLANDat[[ADDR:]]
 # CHECK-NEXT:   .word 0x4f82e010
 # CHECK-NEXT:   ret
+# CHECK-BOLT-ONLY-NEXT: ISLANDat[[ADDR]]
 # CHECK-NEXT:   .short 0xff00
 # CHECK-NEXT:   .byte 0x42
-.size _start, .-_start
 
 ## Force relocation mode.
   .reloc 0, R_AARCH64_NONE