A re-run attempt merged into an empty base: another PR's HIL job ran
between attempt 1 and the retry and rewrote the shared hil_report.json,
so the run-stamp guard (correctly) refused the foreign base but the
full-fleet results were lost - the retry report contained only the
re-run cells.
Give each (run id, job) its own report dir instead:
- attempts of the same run share a dir, so the retry always finds its
own sidecar and .failed spec intact
- interleaved runs of other PRs/jobs write elsewhere and cannot clobber
- the run-stamp mechanism (.failed.run file) becomes redundant and is
removed
- stale per-run dirs are pruned after 2 weeks
ceedling 1.1.0 (released 2026-07-17) fails this project's mock
preprocessing ('Failed to read _build/test/preprocess/.../raw/*.h for
comment stripping'); reproduced locally with an isolated 1.1.0 install
while 1.0.1 passes all 61 tests on the same tree. Unpin once fixed
upstream.
Full-fleet profiling (HIL_PROFILE=1 instrumentation, included) showed each
uPD720201 controller's serialized usbtest battery chain dominates wall time,
and a board whose marginal device port bounces during concurrent batteries
can wedge or kill the controller ("xHCI host not responding to stop endpoint
command"). Every such death traced to mimxrt1015's port (its old "kills the
uPD720201" reputation) - it is removed from the config until recabled;
mimxrt1064's enum-retry stalls were a loose device cable (re-seated).
nrf54lm20dk moves to boards-skip until its failing J-Link probe is replugged.
With the hardware fixed both cards run width-4 batteries plus full flash
churn clean, so scheduling stays simple: two symmetric knobs, flashes and
batteries budgeted per controller.
- schedule_boards(): dispatch boards round-robin across host controllers from
a persisted hint cache (~/.cache/tinyusb-hil/ctrl_cache.json), learned and
merge-on-write refreshed each run (concurrent HIL jobs keep each other's
entries). Only the cached PCI address is consumed - dispatch order and
first-flash budgeting, never battery serialization (batteries resolve live
or fail closed to an all-slot permit).
- HIL_FLASH_PARALLEL (8) and HIL_USBTEST_PARALLEL (4) are budgeted per
controller via lock slots assigned on first sight.
- re-runs: a failed run writes <report dir>/<config>.failed with the exact
re-run spec (--accumulate -b <failed board> -bt <board>:<its failed
tests>) instead of the inverted --skip-board list of everything that
passed; --skip-board is gone, --flasher/--exclude-flasher scope a config
across CI jobs by flasher type (no board names hardcoded in workflows),
and -a/--accumulate merges a re-run into the existing report. The spec is
stamped with GITHUB_RUN_ID and cleared on fresh runs, so a retry can never
consume a spec left behind by a different run's dead or skipped attempt.
- CI: esp-idf firmware builds move out of hil-build into hil-build-esp, and
the esptool-flashed boards run in their own hil-tinyusb-esp job, so the
main hil-tinyusb run starts as soon as the fast toolchains finish instead
of waiting on the slow esp-idf build (an esp toolchain flake previously
skipped the whole rig run). Artifacts are namespaced per toolchain so the
esp job downloads only esp-idf binaries.
- HIL_PROFILE=1: timestamped log lines, per-flash durations, permit-wait
logging, uid->controller map dump for analysis.
- hil_report: per-variant test duration as a dedicated trailing column,
recorded only by full runs.
Validated on the ci rig (fixed seeds 20260716/777, full fleet at 8/4):
738s/780s walls with only known-flake failures and no controller deaths,
vs 1134-1211s serialized-battery baseline.
Pool/config:
- record real uids (ra8m1_ek), enable usbtest for espressif s3/p4, then
park ra6m5_ek and ra8m1_ek in boards-skip (ra6m5's usbtest/MSC traffic
can kill the uPD720201 host on its ROM firmware; ra8m1 USBHS bring-up
pending); max32666/nrf54lm20 stay enabled - their MosChip flakiness
never wedges
- re-enable device/usbtest on HS boards (mimxrt1064, ch32v307) now that
uPD720201 firmware 2.0.2.6 fixes the command-ring death; mimxrt1015
stays skipped - its HS battery killed the controller on both ROM and
2.0.2.6 firmware (board-specific); match the moved host-test bundles
(f723 <-> rt1064); skip never-passing tests on the new
nrf5340dk/nrf54lm20dk boards and the detached pico host bundle, each
documented with a comment
Host-controller quirk gating in usbtest.py (auto-skip, self-heals on a
healthy xHCI):
- MosChip MCS9990 EHCI: case 25 (int-OUT never scheduled, FRINDEX bug)
and case 11 (unlinked reads complete short/EREMOTEIO)
- Renesas uPD720201 xHCI: firmware-gated. The card must run firmware
>= 2.0.2.6 (RAM-uploaded - it reverts to ROM on every power cycle):
on older firmware the command ring dies under unlink stress (a
Configure Endpoint command stops completing; the hub worker deadlocks
holding the device lock; only a host power cycle recovers; three
boards reproduced it). usbtest.py reads the FW version register (PCI
config 0x6c) and refuses to run at all on older firmware - hil_test
surfaces that as a failed test with the reason. On current firmware
the full 30-case battery runs (validated FS+HS: metro_m4, f723,
f723-DMA all 30/30).
Scheduling (hil_test.py):
- Shuffle each (board, variant)'s test order with a seeded RNG
(HIL_SHUFFLE_SEED to replay) so usbtest batteries and flash churn spread
across the timeline instead of convoying on one controller.
- Per-controller usbtest + flash semaphores: HIL_USBTEST_PARALLEL
(default 4) concurrent usbtest batteries and HIL_FLASH_PARALLEL
(default 8) concurrent flashes per host controller. Profiled on
uPD720201 firmware 2.0.2.6 across 8/1..12/8: wall time falls
22.2/14.3/12.5/10.8 min at usbtest width 1/2/3/4 and plateaus there;
zero controller errors everywhere; first battery case failures
(leaf-hub bandwidth stretch) appear at 12/8, and flash width 12 only
amplifies flasher-hub contention flakes - so 8/4 is the optimum. A
separate battery-window flash throttle was profiled and dropped.
- Give every example a unique hardcoded USB PID (0x4001-0x4022, usbtest
keeps 0x4010) instead of the PID_MAP interface bitmap: different
examples now always re-enumerate back-to-back, even on boards whose
CPU reset does not drop D+ (WCH CH58x), so the EXAMPLE_PID table and
same-PID adjacency reordering in hil_test.py are gone; only the
variant-boundary same-example repeat needs a swap.
- Report matrix: stable columns with the metric-bearing tests pinned
first (usbtest, cdc_msc_throughput, msc_file_explorer[_freertos]),
the rest alphabetical.
Fail fast:
- enum wait budget 8 s on the first attempt, 4 s on retries; dfu waits
are deadline-based so dfu-util's own runtime counts against the
budget.
A device-absent failure now costs ~3-5x a passing test (20-30 s)
instead of 10-30x (47-150 s).
- CI runs hil_test with --retry 1 and no in-run second pass: a broken
fixture fails the job fast instead of holding the self-hosted runner
for hours and blocking other PRs' HIL jobs. hil_test still writes the
.skip sidecar, so a manual re-run attempt only retests what failed.
Review fixes (multi-agent adversarial review of this commit):
- tinyusb_win_usbser.inf: the PID rework moved five CDC examples onto
even PIDs the INF's odd-only DeviceList never matched (legacy-Windows
usbser binding) - appended 0x4006/4008/400a/4020/4022 to both lists.
- usbtest example: USBTEST_TIER is now overridable and the descriptors
and pumps are tier-conditional, so a board whose DCD cannot serve a
tier lowers it instead of skipping the whole example - RA2A1 (RUSB2
with no isochronous pipe) builds at tier 3 via its BOARD_ define; the
host battery follows the tier advertised in bcdDevice. Tier-4 output
verified byte-identical after the refactor.
- dynamic_configuration's second config derived USB_PID + 11 = 0x4018,
colliding with net_lwip_webserver - now USB_PID + 0x0100, outside the
per-example space. tools/check_example_pids.py (pre-commit hook)
enforces PID uniqueness incl. derived and literal idProduct values.
- usbtest.py firmware gate: matched by device ID (uPD720201/720202,
both use the 0x6c FW register), and an unreadable version (setpci
missing/denied) now refuses with its own message instead of
masquerading as "firmware 0x00000000"; noted the gate is necessary
but not sufficient (board-specific kills stay per-board skips).
- hil_test: deadline waits use time.monotonic(); multiprocessing
context pinned to fork (raw semaphores in Pool initargs); flash and
usbtest permits unified into one fail-closed, exception-safe
ctrl_permit (unknown controller takes every slot and logs a warning
instead of silently borrowing slot 0); an all-skipped battery
reports as skip, not "0/0" failure; slow-body polls (mtp, printer,
disk read) go through a shared deadline-based wait_until so their
bodies count against the enum budget; throughput's FS detection
compares serials case-insensitively like every other walk; a missing
MSC read-speed line now fails the host msc_file_explorer test
instead of passing with an empty metric.
Hardening:
- fail fast (15 s) when a driver-registry sysfs write blocks: a wedged
device otherwise turns every subsequent battery into an unkillable
D-state writer and silently hangs the whole run
- usb-recover skill: a VM reboot is not a reliable cure (MosChip hubs
latch up across the PCIe reset); full host power cycle is
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HeF2gZ1M7GWkz6Av4BpKPg
The log shows two gated tools, not one: compound Bash pipelines AND Write — the
review tried to drop check_headings.py (at /tmp, then the workdir, both denied).
Add Write to the allowlist.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Every blocked call in the review log was a compound Bash pipeline; Read/Grep/
Glob/Task already ran un-prompted, so only bare Bash needs allowlisting.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The auto-review job runs /code-review headless, which uses Bash (git diff, gh),
file search, and Task (it fans out sub-agent reviewers). None were allowlisted,
so every such call stalled on a per-tool approval prompt and the review couldn't
gather the diff or spawn reviewers. Add --allowedTools.
Safe here (unlike claude.yml): this job is gated to same-repo PRs and its token
is contents:read, so it cannot push. Bash is broad — scope to git/gh/grep if
preferred.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The BSP family and MCU option were named "ch58x"/"CH58X", but the supported part is
the CH583/CH582 (and the SDK repo is openwch/ch583); CH585 is a separate MCU family,
so the CH58x umbrella was misleading. Rename to the specific family:
- hw/bsp/ch58x -> hw/bsp/ch583 (dir), and the BSP-local files ch58x_it.* ->
ch583_it.*, system_ch58x.* -> system_ch583.* (include guards/refs updated). The
vendor SDK files (CH58x_common.h, CH58x_*.c in hw/mcu/wch/ch583) keep their names.
- OPT_MCU_CH58X -> OPT_MCU_CH583 in tusb_option.h, tusb_mcu.h, and the shared WCH
USBFS driver (ch32_usbfs_reg.h, dcd_ch32_usbfs.c). OPT_MCU_CH582 is kept as an
alias (same value), so either name selects the same code.
- FAMILY_MCUS CH58X -> CH583, CFG_TUSB_MCU=OPT_MCU_CH583, mcu:CH58X -> mcu:CH583 in
the example skip lists, the CI build matrix (ci_set_matrix.py), the get_deps family
tag, and docs/reference/boards.rst.
Board names (ch582m_evt, yd-ch582m) are unchanged. Verified: make + cmake build for
ch582m_evt, and ci.lan HIL (all device examples pass).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pass --effort max to the claude CLI in the auto-review workflow so PR
reviews run at maximum reasoning effort. Switch claude_args to a
multi-line block scalar for readability, keeping --max-turns 50 and
--model claude-opus-4-8 unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* hil: enable nanoch32v203 in CI with fsdev + usbfs variants
nanoch32v203 was parked in boards-skip; move it into the active pool now
that the board is wired to the ci.lan rig. Cover both USB device IPs as
build variants:
- nanoch32v203-fsdev: RHPORT_DEVICE=0 (USBD / stm32 FSDev IP)
- nanoch32v203-usbfs: RHPORT_DEVICE=1 (WCH USBFS IP)
The review action currently runs on the default Sonnet 4.6. On PR #3643
(musb EP0 race) it posted "No issues found" while an Opus pass on the
same diff surfaced substantive questions (ISR-boundary RXRDY lifetime,
regression scope of the DATA-state split). Pin the reviewer to
claude-opus-4-8 for higher-signal reviews; subagents keep their cheaper
default models.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Claude Code Review action runs /code-review:code-review with a hard
--max-turns cap. On large PRs (e.g. #3636 "add stm32c5 support", 29 files
/ +1689), the agent exhausts 20 turns exploring the diff before it can
produce and post its review, so the SDK returns an error and the
claude-review check fails red with:
Reached maximum number of turns (20)
Raise the cap to 50 so port-sized PRs complete and post their review.
Cost scales with tokens, not the cap: a finished review pays the same
whether the ceiling is 25 or 50 — the cap only bites when the agent
would otherwise be force-stopped mid-run.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* test/hil: replace build.flags_on with named variant schema
Boards declare build variants as `variant: [{name, flags}]` instead of
`build.flags_on`. The variant `name` is the build dir (cmake-build-<name>) and
the HIL report row; `flags` is the raw CFLAGS string (-D...=1) injected via
CFLAGS_CLI. No `variant` => a single build named after the board.
- build.py: --build-name <name> (dir) + --cflag=<token> (raw CFLAGS, repeatable,
=form survives the matrix's shell word-splitting); drop -f1/CFLAGS wrapping.
- hil_ci_set_matrix.py: emit one build arg per variant.
- hil_test.py: iterate variants; report row + build dir = variant name.
- hil_ci.sh: copy all cmake-build-<board>* dirs for -b runs.
- get_deps.py: accept (ignore) --build-name/--cflag from matrix args.
- tinyusb.json: migrate all 6 flags_on boards to variant.
* board_test: park CI build with busy spin instead of wfe
The Size Difference Report and HIL comments rendered their titles at h1, which
is oversized inside a PR comment. Use h2 for both titles (with subsections
demoted to h3 to keep the hierarchy), and rename the HIL comment from
"HIL test results" to "Hardware-in-the-loop (HIL) Test Report" for consistency.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
hil-hfp-iar runs hil_test.py on hfp.json built with IAR on its own rig. Upload
its report as the hil-report-hfp-iar artifact and add the job to the hil-report
combine job's needs, so the sticky comment shows a third table for the IAR rig
alongside tinyusb.json and hfp.json (gcc). The combine gate now runs if either
HIL job produced results.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
hil_test.py persists results in a hil_report.json sidecar and regenerates
hil_report.md from it. A full run starts fresh; a re-run (--skip-board / -bt,
i.e. the .skip file) merges into the existing report so already-passed
boards/tests are preserved while only re-run cells update. The report dir is
configurable via HIL_REPORT_DIR.
build.yml: each HIL rig writes the report to a workspace-sibling dir that
survives the per-attempt workspace clean, and uploads it as an artifact. A new
hil-report job merges the rigs' reports into one sticky PR comment (marocchino)
with one table per rig.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* ci: carry metrics baseline forward on no-code-change pushes
The code-metrics job is gated on code_changed and only uploads the
metrics-tinyusb artifact on push, so a workflow/docs-only push to master (e.g.
removing an unrelated workflow) leaves the latest master Build run without a
baseline. PRs download the baseline from the latest master run, so the size
comparison then finds nothing and silently falls back to absolute sizes.
Add a small metrics-carry-forward job that, on a non-code-change push, downloads
the previous metrics-tinyusb artifact and re-publishes it, so the latest run
always carries a usable baseline. Carry-forward runs re-upload too, so the
baseline chains across consecutive no-code pushes (bounded by artifact retention).
This personal automation now lives in the hathach/hathach repo alongside the
other personal project-sync workflows; it has no place in the tinyusb library.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* ci(labeler): auto-apply Port labels from changed driver files
Add path-based labeling so a PR touching a dcd/hcd driver under
src/portable/ gets the matching "Port <ip>" label automatically.
Labels were renamed to add emojis (Adafruit 🌸, Sponsor 💖, Prio 🚩,
Prio Top 🚨); update the hardcoded label names in the labeler script
to match so they attach to the existing labels instead of recreating
plain ones.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The search drops is:issue to include PRs, but the GraphQL selection only
had '... on Issue { id }', so PR nodes returned no id and were skipped.
Add '... on PullRequest { id }'. (Codex/Copilot review finding.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Drop the is:issue qualifier so sponsor pull requests are synced to the
board too (search type ISSUE already returns both). A sponsor's open PR
is exactly the kind of work to prioritize reviewing.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cron (6h) + manual workflow that adds open issues opened by GitHub
sponsors (public and private) and Adafruit org members across the
adafruit org and the maintainer's repos to the private Sponsor Triage
project board, setting Tier and Visibility. Logs counts only to avoid
leaking private sponsor logins. Needs the SPONSOR_TOKEN PAT secret.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- rename priority labels usage to Prio / Prio Top
- label Adafruit members (Adafruit + Sponsor + Prio Top) and public
GitHub sponsors by tier; contributors get Prio
- skip sponsor/Adafruit perks for the maintainer's own issues/PRs
- support discussions via the GraphQL addLabelsToLabelable mutation
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The review workflow posted nothing when a review found no issues:
with use_sticky_comment unset, the only output path was inline
comments, so a clean review surfaced no comment at all on the PR.
Enable use_sticky_comment so a single summary comment is posted/
updated every run, making "no issues found" results visible.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add allowed_bots: 'claude' so that when claude[bot] pushes commits the
workflow skips gracefully instead of erroring.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Codex correctly noted that allowing python3/python (and really cmake/make
too) is arbitrary code + network execution: a command allowlist cannot
contain a prompt-injected or malicious fork PR when this job holds the
OAuth secret + write token, and the review workflow directs fork PRs to
@claude.
The Bash allowlist was beyond the original scope (auto-commit/PR) anyway.
Remove it: Claude edits files and the action commits/opens the PR, and the
resulting commit is verified by the existing CircleCI matrix. Keep
use_commit_signing and --max-turns 30.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Codex flagged that @claude can be summoned on a fork PR (the review
workflow even directs fork PRs here), so the checked-out PR content is
potentially attacker-controlled. Unrestricted Bash in this write-token +
OAuth-secret job let prompt injection steer Claude into arbitrary
shell/network commands.
Scope Bash to the repo's actual verification commands (cmake, ninja,
make, ctest, python/python3, pre-commit, clang-format, codespell, git).
This blocks the injection-to-arbitrary-command path while still letting
Claude build/test before committing. Building fork code itself is already
done by the existing CircleCI, so that surface is unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Configure the @claude summon workflow so it can actually produce a
verified fix when asked in an issue/PR comment:
- use_commit_signing: bot commits show as Verified
- --allowedTools Bash: lets Claude build/test to verify the fix before
committing (default allowlist blocks Bash). Safe because the job `if`
gate restricts this to OWNER/MEMBER/COLLABORATOR.
- --max-turns 30: enough turns to investigate -> fix -> verify
Auto-commit/PR is already built into claude-code-action and the
required write permissions were already present, so no permission
changes are needed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- claude.yml: drop the issues "assigned" trigger; its author_association
gate keys on the issue author, not the assigner, so a maintainer
assigning an outside contributor's issue would be wrongly skipped.
- claude-code-review.yml: issues: read -> write so use_sticky_comment can
create/update its PR comment via the issues API.
- hil SKILL.md: make local/remote command blocks copy-pasteable (drop
[-b BOARD_NAME] notation for concrete examples) and fix timeout
(600000 ms is 10 min; use 1200000 ms for the stated 20 min).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- claude.yml: gate @claude on author_association (OWNER/MEMBER/COLLABORATOR)
so the write-scoped token and OAuth secret are never issued for an
untrusted commenter on this public repo (defense-in-depth).
- claude-code-review.yml: skip fork PRs in the job condition
(head.repo.full_name == github.repository) since forks get no secrets
and would only fail noisily; fix the misleading token comment; pass
additional_permissions: actions: read so actions: read is effective.
- hil SKILL.md: reword hostname guidance, use full test/hil/* paths, and
show an explicit CONFIG= assignment so the local command is runnable.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- claude-code-review.yml: re-enable (drop `if: false`); switch from
pull_request_target to pull_request so fork PRs never receive the
OAuth token (avoids prompt-injection token leak). Auto-review on
open/synchronize/reopen/ready_for_review, skip drafts, sticky comment.
- claude.yml: grant contents/pull-requests/issues write so @claude can
reply and push fixes; @claude is the on-demand path for fork PRs.
