Captain Bob's CI directive (2026-11-17) and the design he ruled point by point (docs/v4.0.0/plans/2026-11-17-emergency-diagnostic-build.md). - v4/system/dump.c, v4/include/v4/dump.h: what a node was doing when it stopped -- the fault's kind, address and count; P A B; the data stack, T first; the return stack, R first, each entry named by the word it falls in (the backtrace; ROM and below-every-word said so); the running line. Bounded dictionary walk; no C library. Ruled: every build, on a stop only (a handled fault keeps its line). - kernel/src/v4/sk_v4.c: the dump before each of the three halts (not started; POST's blocks not let go; a node stopped on a fault) -- Hera and any stopped node. kernel/Makefile links dump.o. - scripts/nightly.sh diagnose [<stage>]: after a failed stage, that configuration built again with EMERGENCY_CONSOLE_ENABLED=1 -- build only, nothing booted (ruled) -- its artifacts kept in logs/<TS>/diagnose/ only from a build that passed, with a README saying how to launch it with --log-level=debug. Always exits 0: evidence, never a result. The protocol runs it after a failed stage and still exits with that stage's code; acceptance.yml, nightly.yml and release.yml run it `if: failure()`, before the logs are kept. Any failed stage triggers it (ruled). - capsules/BLOCK_MAP.md: as the build writes it -- the v4:physics.4th row, missing since28ffed8f(ruled: commit the regenerated map). - Test first: test_dump.c failed with no dump.c (logs/20261117-dump), then 15 checks at both widths and under ASan+UBSan. The diagnose stage: a real EMERGENCY amd64 build; a failing build exits 0 and keeps no artifacts (a defect found and fixed this way); the protocol path keeps the stage's exit. - Also the write-up of the regression at046ce0be, for ruling. Not yet: acceptance on a clean tree (the next run); the kernel's stop wiring is compile-tested only -- no boot reaches a stop on purpose. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017RzZgrf7V7Zk2iczBXDRbm
88 lines
3.9 KiB
YAML
88 lines
3.9 KiB
YAML
# The acceptance protocol, as the gate on every pull request into master (Captain Bob, 2026-11-06:
|
|
# "failures should not be allowed to be merged to master"). It is .claude/CLAUDE.md's protocol,
|
|
# run by the same stages a developer and the nightly run (scripts/nightly.sh), in its order, one
|
|
# QEMU at a time:
|
|
#
|
|
# amd64, aarch64, riscv64 to [zuse@Hera] ok>, POST: PASSED, `3 4 + .` -> 7 ok
|
|
# the PC stick loaded high the same; and the three architectures' dict_hash agree
|
|
# the 32-bit build make -C v4 test-32, 0 failures
|
|
#
|
|
# master's branch protection requires this check to pass before a PR merges. The serial logs and
|
|
# DoE CSVs are kept as the run's artifacts, pass or fail.
|
|
name: Acceptance
|
|
|
|
on:
|
|
pull_request:
|
|
branches: [master]
|
|
types: [opened, synchronize, reopened]
|
|
workflow_dispatch:
|
|
|
|
# No concurrency group (Captain Bob, 2026-11-11, #140: "It should be global"). The one runner
|
|
# (starshipos-vps) has capacity 1, so every job of every workflow waits its turn in Gitea's queue
|
|
# and none is cancelled; a runner-image rebuild still blocks the rest (2026-11-07). A shared
|
|
# concurrency group kept only one waiting run, and each run that joined cancelled it.
|
|
# The capacity is explicit in the runner's config.yaml: runner.capacity: 1 (FABRIC-3, 2026-11-11).
|
|
|
|
jobs:
|
|
acceptance:
|
|
name: Acceptance
|
|
runs-on: ubuntu-latest
|
|
container:
|
|
image: gitea.starshipos-f18.org/rajames/lithosananke-runner:4f675c8b75b2 # ci/runner-image (#120)
|
|
credentials: # the registry refuses anonymous pulls
|
|
username: rajames
|
|
password: ${{ secrets.PACKAGES_TOKEN }}
|
|
timeout-minutes: 120
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
with:
|
|
fetch-depth: 0 # the SBOM's version is git describe
|
|
lfs: false # only the logs and DoE CSVs are in LFS; the build needs neither
|
|
|
|
# The developer protocol's evidence first (#135): the PR's head carries a passing
|
|
# logs/<TS>/PROTOCOL.md for its own tree (scripts/nightly.sh protocol, run by hand), or it is
|
|
# refused here, before the boots. The head, not the merge commit the checkout made.
|
|
- name: The developer protocol's evidence
|
|
run: scripts/check_protocol.sh "${{ github.event.pull_request.head.sha || 'HEAD' }}"
|
|
|
|
# The gate into master refuses an unsigned build (#174; Captain Bob, 2026-11-12: "Yes, reject
|
|
# unsigned builds"). It warned and went on. scripts/test_acceptance_signing.sh runs this step.
|
|
- name: The capsule-signing key
|
|
env:
|
|
KEY: ${{ secrets.CAPSULE_SIGN_KEY }}
|
|
run: | # ci/pki/make-ca.sh; kernel/Makefile signs with SIGN_KEY
|
|
if [ -n "$KEY" ]; then
|
|
f="${RUNNER_TEMP:-/tmp}/capsule-signing.key"
|
|
(umask 077; printf '%s\n' "$KEY" > "$f")
|
|
echo "SIGN_KEY=$f" >> "$GITHUB_ENV"; echo "The capsules will be signed."
|
|
else
|
|
echo "::error::No CAPSULE_SIGN_KEY secret: the gate into master does not accept an unsigned build."
|
|
exit 1
|
|
fi
|
|
|
|
- name: Acceptance amd64
|
|
run: scripts/nightly.sh accept amd64
|
|
- name: Acceptance aarch64
|
|
run: scripts/nightly.sh accept aarch64
|
|
- name: Acceptance riscv64
|
|
run: scripts/nightly.sh accept riscv64
|
|
- name: Acceptance PC stick loaded high
|
|
run: scripts/nightly.sh accept stick-high
|
|
- name: Acceptance 32-bit build
|
|
run: scripts/nightly.sh test-32
|
|
|
|
- name: EMERGENCY diagnostic build (on failure; evidence, not a result)
|
|
if: failure()
|
|
run: scripts/nightly.sh diagnose # build only; the failed step stays the job's failure
|
|
|
|
- name: Keep the logs and DoE CSVs
|
|
if: always()
|
|
uses: actions/upload-artifact@v3
|
|
with:
|
|
name: acceptance-logs-${{ github.run_number }}
|
|
path: |
|
|
logs/
|
|
experiments/bare_metal/runs/
|
|
build/nightly/accept-*.txt
|
|
build/nightly/test-32.txt
|