[pull] main from llvm:main

.ci/generate-buildkite-pipeline-premerge

@@ -128,9 +128,8 @@ if [[ "${windows_projects}" != "" ]]; then
         limit: 2
   timeout_in_minutes: 150
   env:
-    CC: 'cl'
-    CXX: 'cl'
-    LD: 'link'
+    MAX_PARALLEL_COMPILE_JOBS: '16'
+    MAX_PARALLEL_LINK_JOBS: '4'
   commands:
   - 'C:\\BuildTools\\Common7\\Tools\\VsDevCmd.bat -arch=amd64 -host_arch=amd64'
   - 'bash .ci/monolithic-windows.sh "$(echo ${windows_projects} | tr ' ' ';')" "$(echo ${windows_check_targets})"'

.ci/metrics/metrics.py

@@ -26,7 +26,67 @@ class JobMetrics:
     workflow_id: int
 
 
-def get_metrics(github_repo: github.Repository, workflows_to_track: dict[str, int]):
+@dataclass
+class GaugeMetric:
+    name: str
+    value: int
+    time_ns: int
+
+
+def get_sampled_workflow_metrics(github_repo: github.Repository):
+    """Gets global statistics about the Github workflow queue
+
+    Args:
+      github_repo: A github repo object to use to query the relevant information.
+
+    Returns:
+      Returns a list of GaugeMetric objects, containing the relevant metrics about
+      the workflow
+    """
+
+    # 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
+        ]
+    )
+
+    workflow_metrics = []
+    workflow_metrics.append(
+        GaugeMetric(
+            "workflow_queue_size",
+            queued_workflow_count,
+            time.time_ns(),
+        )
+    )
+    workflow_metrics.append(
+        GaugeMetric(
+            "running_workflow_count",
+            running_workflow_count,
+            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())
+    )
+    return workflow_metrics
+
+
+def get_per_workflow_metrics(
+    github_repo: github.Repository, workflows_to_track: dict[str, int]
+):
     """Gets the metrics for specified Github workflows.
 
     This function takes in a list of workflows to track, and optionally the
@@ -43,14 +103,14 @@ def get_metrics(github_repo: github.Repository, workflows_to_track: dict[str, in
       Returns a list of JobMetrics objects, containing the relevant metrics about
       the workflow.
     """
-    workflow_runs = iter(github_repo.get_workflow_runs())
-
     workflow_metrics = []
 
     workflows_to_include = set(workflows_to_track.keys())
 
-    while len(workflows_to_include) > 0:
-        workflow_run = next(workflow_runs)
+    for workflow_run in iter(github_repo.get_workflow_runs()):
+        if len(workflows_to_include) == 0:
+            break
+
         if workflow_run.status != "completed":
             continue
 
@@ -70,34 +130,6 @@ def get_metrics(github_repo: github.Repository, workflows_to_track: dict[str, in
         workflow_jobs = workflow_run.jobs()
         if workflow_jobs.totalCount == 0:
             continue
-        if workflow_jobs.totalCount > 1:
-            raise ValueError(
-                f"Encountered an unexpected number of jobs: {workflow_jobs.totalCount}"
-            )
-
-        created_at = workflow_jobs[0].created_at
-        started_at = workflow_jobs[0].started_at
-        completed_at = workflow_jobs[0].completed_at
-
-        job_result = int(workflow_jobs[0].conclusion == "success")
-        if job_result:
-            # We still might want to mark the job as a failure if one of the steps
-            # failed. This is required due to use setting continue-on-error in
-            # the premerge pipeline to prevent sending emails while we are
-            # testing the infrastructure.
-            # TODO(boomanaiden154): Remove this once the premerge pipeline is no
-            # longer in a testing state and we can directly assert the workflow
-            # result.
-            for step in workflow_jobs[0].steps:
-                if step.conclusion != "success":
-                    job_result = 0
-                    break
-
-        queue_time = started_at - created_at
-        run_time = completed_at - started_at
-
-        if run_time.seconds == 0:
-            continue
 
         if (
             workflows_to_track[workflow_run.name] is None
@@ -110,20 +142,45 @@ def get_metrics(github_repo: github.Repository, workflows_to_track: dict[str, in
         ):
             break
 
-        # The timestamp associated with the event is expected by Grafana to be
-        # in nanoseconds.
-        created_at_ns = int(created_at.timestamp()) * 10**9
-
-        workflow_metrics.append(
-            JobMetrics(
-                workflow_run.name,
-                queue_time.seconds,
-                run_time.seconds,
-                job_result,
-                created_at_ns,
-                workflow_run.id,
+        for workflow_job in workflow_jobs:
+            created_at = workflow_job.created_at
+            started_at = workflow_job.started_at
+            completed_at = workflow_job.completed_at
+
+            job_result = int(workflow_job.conclusion == "success")
+            if job_result:
+                # We still might want to mark the job as a failure if one of the steps
+                # failed. This is required due to use setting continue-on-error in
+                # the premerge pipeline to prevent sending emails while we are
+                # testing the infrastructure.
+                # TODO(boomanaiden154): Remove this once the premerge pipeline is no
+                # longer in a testing state and we can directly assert the workflow
+                # result.
+                for step in workflow_job.steps:
+                    if step.conclusion != "success":
+                        job_result = 0
+                        break
+
+            queue_time = started_at - created_at
+            run_time = completed_at - started_at
+
+            if run_time.seconds == 0:
+                continue
+
+            # The timestamp associated with the event is expected by Grafana to be
+            # in nanoseconds.
+            created_at_ns = int(created_at.timestamp()) * 10**9
+
+            workflow_metrics.append(
+                JobMetrics(
+                    workflow_run.name + "-" + workflow_job.name,
+                    queue_time.seconds,
+                    run_time.seconds,
+                    job_result,
+                    created_at_ns,
+                    workflow_run.id,
+                )
             )
-        )
 
     return workflow_metrics
 
@@ -139,12 +196,27 @@ def upload_metrics(workflow_metrics, metrics_userid, api_key):
       metrics_userid: The userid to use for the upload.
       api_key: The API key to use for the upload.
     """
+
+    if len(workflow_metrics) == 0:
+        print("No metrics found to upload.", file=sys.stderr)
+        return
+
     metrics_batch = []
     for workflow_metric in workflow_metrics:
-        workflow_formatted_name = workflow_metric.job_name.lower().replace(" ", "_")
-        metrics_batch.append(
-            f"{workflow_formatted_name} queue_time={workflow_metric.queue_time},run_time={workflow_metric.run_time},status={workflow_metric.status} {workflow_metric.created_at_ns}"
-        )
+        if isinstance(workflow_metric, GaugeMetric):
+            name = workflow_metric.name.lower().replace(" ", "_")
+            metrics_batch.append(
+                f"{name} value={workflow_metric.value} {workflow_metric.time_ns}"
+            )
+        elif isinstance(workflow_metric, JobMetrics):
+            name = workflow_metric.job_name.lower().replace(" ", "_")
+            metrics_batch.append(
+                f"{name} queue_time={workflow_metric.queue_time},run_time={workflow_metric.run_time},status={workflow_metric.status} {workflow_metric.created_at_ns}"
+            )
+        else:
+            raise ValueError(
+                f"Unsupported object type {type(workflow_metric)}: {str(workflow_metric)}"
+            )
 
     request_data = "\n".join(metrics_batch)
     response = requests.post(
@@ -176,16 +248,21 @@ def main():
     # Enter the main loop. Every five minutes we wake up and dump metrics for
     # the relevant jobs.
     while True:
-        current_metrics = get_metrics(github_repo, workflows_to_track)
-        if len(current_metrics) == 0:
-            print("No metrics found to upload.", file=sys.stderr)
-            continue
+        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):
-            workflows_to_track[workflow_metric.job_name] = workflow_metric.workflow_id
+            if isinstance(workflow_metric, JobMetrics):
+                workflows_to_track[
+                    workflow_metric.job_name
+                ] = workflow_metric.workflow_id
 
         time.sleep(SCRAPE_INTERVAL_SECONDS)
 

.ci/monolithic-windows.sh

@@ -50,6 +50,10 @@ echo "--- cmake"
 pip install -q -r "${MONOREPO_ROOT}"/mlir/python/requirements.txt
 pip install -q -r "${MONOREPO_ROOT}"/.ci/requirements.txt
 
+export CC=cl
+export CXX=cl
+export LD=link
+
 # The CMAKE_*_LINKER_FLAGS to disable the manifest come from research
 # on fixing a build reliability issue on the build server, please
 # see https://github.com/llvm/llvm-project/pull/82393 and
@@ -72,8 +76,8 @@ cmake -S "${MONOREPO_ROOT}"/llvm -B "${BUILD_DIR}" \
       -D CMAKE_EXE_LINKER_FLAGS="/MANIFEST:NO" \
       -D CMAKE_MODULE_LINKER_FLAGS="/MANIFEST:NO" \
       -D CMAKE_SHARED_LINKER_FLAGS="/MANIFEST:NO" \
-      -D LLVM_PARALLEL_COMPILE_JOBS=16 \
-      -D LLVM_PARALLEL_LINK_JOBS=4
+      -D LLVM_PARALLEL_COMPILE_JOBS=${MAX_PARALLEL_COMPILE_JOBS} \
+      -D LLVM_PARALLEL_LINK_JOBS=${MAX_PARALLEL_LINK_JOBS}
 
 echo "--- ninja"
 # Targets are not escaped as they are passed as separate arguments.

.github/new-prs-labeler.yml

@@ -730,6 +730,9 @@ llvm:regalloc:
 lldb:
   - lldb/**
 
+lldb-dap:
+  - lldb/tools/lldb-dap/**
+
 backend:AMDGPU:
   - '**/*amdgpu*'
   - '**/*AMDGPU*'

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

@@ -20,41 +20,56 @@ on:
 jobs:
   build-ci-container:
     if: github.repository_owner == 'llvm'
-    runs-on: depot-ubuntu-22.04-16
-    outputs:
-      container-name: ${{ steps.vars.outputs.container-name }}
-      container-name-tag: ${{ steps.vars.outputs.container-name-tag }}
-      container-filename: ${{ steps.vars.outputs.container-filename }}
+    runs-on: ${{ matrix.runs-on }}
+    strategy:
+      matrix:
+        include:
+          # The arch names should match the names used on dockerhub.
+          # See https://github.com/docker-library/official-images#architectures-other-than-amd64
+          - arch: amd64
+            runs-on: depot-ubuntu-22.04-16
+          - arch: arm64v8
+            runs-on: depot-ubuntu-22.04-arm-16
     steps:
       - name: Checkout LLVM
         uses: actions/checkout@v4
         with:
           sparse-checkout: .github/workflows/containers/github-action-ci/
+      # podman is not installed by default on the ARM64 images.
+      - name: Install Podman
+        if: runner.arch == 'ARM64'
+        run: |
+          sudo apt-get install podman
       - name: Write Variables
         id: vars
         run: |
-          tag=`date +%s`
-          container_name="ghcr.io/$GITHUB_REPOSITORY_OWNER/ci-ubuntu-22.04"
+          tag=$(git rev-parse --short=12 HEAD)
+          container_name="ghcr.io/$GITHUB_REPOSITORY_OWNER/${{ matrix.arch }}/ci-ubuntu-22.04"
           echo "container-name=$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" >> $GITHUB_OUTPUT
+          echo "container-agent-filename=$(echo $container_name-agent:$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-filename }}
+          podman save ${{ steps.vars.outputs.container-name-agent-tag }} > ${{ steps.vars.outputs.container-agent-filename }}
 
       - name: Upload container image
         uses: actions/upload-artifact@v4
         with:
-          name: container
-          path: ${{ steps.vars.outputs.container-filename }}
+          name: container-${{ matrix.arch }}
+          path: "*.tar"
           retention-days: 14
 
       - name: Test Container
@@ -76,13 +91,29 @@ jobs:
     steps:
       - name: Download container
         uses: actions/download-artifact@v4
-        with:
-          name: container
 
       - name: Push Container
         run: |
-          podman load -i ${{ needs.build-ci-container.outputs.container-filename }}
-          podman tag ${{ needs.build-ci-container.outputs.container-name-tag }} ${{ needs.build-ci-container.outputs.container-name }}:latest
+          function push_container {
+            image_name=$1
+            latest_name=$(echo $image_name | sed 's/:[a-f0-9]\+$/:latest/g')
+            podman tag $image_name $latest_name
+            echo "Pushing $image_name ..."
+            podman push $image_name
+            echo "Pushing $latest_name ..."
+            podman push $latest_name
+          }
+
           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
+          for f in $(find . -iname *.tar); do
+            image_name=$(podman load -q -i $f | sed 's/Loaded image: //g')
+            push_container $image_name
+
+            if echo $image_name | grep '/amd64/'; then
+              # For amd64, create an alias with the arch component removed.
+              # This matches the convention used on dockerhub.
+              default_image_name=$(echo $(dirname $(dirname $image_name))/$(basename $image_name))
+              podman tag $image_name $default_image_name
+              push_container $default_image_name
+            fi
+          done