Conversation
…ing-uartlogger2 Enable turbo loader on HalloWing M4 and UART Logger II
…eenable Turn the native .mpy loader back on for all nRF52 boards
Make `MICROPY_PY_DOUBLE_TYPECODE` cover every `'d'` typecode path
Store ROM qstr strings in a blob with 16-bit offsets
- board_init() no longer deletes code.py or creates main.py: the Elioblocs editor now writes code.py. - The default boot.py explains why the drive is read-only and how to change it. - GPIO15-16 are free pins (no 32 kHz crystal), and the NeoPixel count anticipates the next board revision.
The artifact API token `ACTIONS_RUNTIME_TOKEN` is only given to action steps, not `run:` steps, so calling the script with `node` found no records at all. `ci_download_sizes.mjs` is now a module that an `actions/github-script` step imports and calls; the install of `@actions/artifact` stays a shell step. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…main Translations update from Hosted Weblate
The free-flash filter defaulted to "under 1 KiB", which shows an empty table when no board is that tight, as on main today. Default to "all"; the colour bands still mark the tight boards. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…path bitmaptools.blit() fast path
Add ELIO Eliobot S3 board
Report firmware sizes from CI
…ogen Auto-update the Zephyr autogen files
`cpbuild._create_semaphore()` only recognized the `fifo:` jobserver of GNU make 4.4+ and otherwise compiled one file at a time. The ubuntu-24.04 CI runners have make 4.3, which passes the jobserver as file descriptors that are closed for the `west build` recipe, so every CI zephyr-cp build compiled its ~400 CircuitPython objects serially, for each of the 17 languages. Fall back to one compile per available CPU instead of one. A 395-file rebuild with no jobserver went from 44 s to 8 s on a 20-core machine. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
An unchanged rebuild recompiled all ~400 CircuitPython objects, and so did each of the 16 extra language builds in CI, at about 90 s each. Two generated files were rewritten on every build with a new modification time, and everything that depends on them followed: - `zephyr2cp.py` walked the device tree with a `set` of node objects, whose order depends on memory addresses, so the generated `board.c` came out in a different order every run. Walk with a list, write `board.c` only when it changes, and use a `range` instead of a `set` for GPIO pin numbers. - `collect_defs()` wrote each `.collected` file with several `cat` commands, so the later appends made the file look newer than the first command's record. Concatenate in Python and write only when the content changes. - Pass `check_hash` for `makeqstrdata.py`, `makemoduledefs.py` and `make_root_pointers.py`, so an unchanged header keeps its modification time. An unchanged rebuild now compiles nothing, and a language switch compiles `translate.c` and the two per-language files. A clean build round-tripped through de_DE and back to en_US is byte-identical. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
… every event The four zephyr-cp `native_*` boards build a host executable for testing. The other 16 languages were already skipped for them on pull requests; skip them on main and merge_group runs too, instead of building and uploading 17 simulator executables per board. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…DWARE_KEYS Per tannewt's review on PR adafruit#11319: this extension point exists solely for hardwarekey's board.EFUSE_KEY* entries, so name it for that specific purpose instead of presenting it as a generic board-globals extension mechanism. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Per tannewt's review on PR adafruit#11319: replace the hand-rolled cp_enum_obj_t type/print function/switch statement with the standard MAKE_ENUM_VALUE / MAKE_ENUM_MAP / MAKE_PRINTER / MAKE_ENUM_TYPE macros (see shared-bindings/wifi/Packet.c), and cp_enum_find() in place of the hand-rolled hardwarekey_purpose_to_obj() lookup. This moves Purpose's members under the class, matching every other enum in shared-bindings: hardwarekey.HMAC_UP / hardwarekey.UNUSED become hardwarekey.Purpose.HMAC_UP / hardwarekey.Purpose.UNUSED. Docstrings referencing them are updated to match. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Every zephyr-cp board job saved a new ~150 MB ccache entry under a key shared by all boards, and restored whichever entry was saved last. Measured over all 30 boards in one run: 5 boards hit by luck, 25 had one hit each, and the hit boards built no faster. The entries were filling the repository's 10 GB Actions cache and evicting the Zephyr SDK and other ports' caches. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…el-fallback zephyr-cp: stop recompiling every file on every build, and compile in parallel on make 4.3
…planning hardwarekey: Add board-exposed keys usable via hmac.new()
Updated by "Update PO files to match POT (msgmerge)" hook in Weblate. Translation: CircuitPython/main Translate-URL: https://hosted.weblate.org/projects/circuitpython/main/
Port-level never_reset tracking and bulk reset helpers are removed. Objects own their hardware and free it via GC finalizers (__del__ -> deinit).
Port-level never_reset tracking and bulk reset helpers are removed. Objects own their hardware and free it via GC finalizers (__del__ -> deinit).
Pin the dependency's one-shot GPTimer correction and version-gated GDMA setup. MatrixPortal S3 hardware now displays correct one- through four-bit patterns, including RGB ramps and pause/resume at three and four bits. Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Keep the ABC and ESP32 refresh implementation while using the dependency's minimal diff against its IDF6 base. Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Update the Protomatter dependency for PR adafruit#92 targeting master with IDF6 support and reliable ESP32 timer scheduling. The dependency source tree matches the previously validated d51abd7 tree exactly. Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Help define board.DISPLAY
Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
stm: switch to finalizer cleanup (CIRCUITPY_BULK_RESET = 0)
Add analogio support to Zephyr
Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Reject invalid height, width, bit depth, tile count, and pin arguments before registering an RGBMatrix object in the static display-bus array. This avoids leaking a slot and leaving an incomplete object for reset handling when those checks raise an exception. Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Expose RowAddressMode.BINARY and RowAddressMode.ABC and validate the constructor argument with the standard CircuitPython enum helpers. Keep binary addressing as the default and update the API documentation. Co-authored-by: Limor Fried <ladyada@users.noreply.github.com>
Addresses review feedback: don't reference the old never-reset implementation in comments, just explain how it works now.
nordic: switch to finalizer cleanup (CIRCUITPY_BULK_RESET = 0)
…xrt10xx mimxrt10xx: switch to finalizer cleanup (CIRCUITPY_BULK_RESET = 0)
Updated the boot button definition for safe mode.
Boot button seems to be 28 (tested with a GPIO monitor script)
…_1_47 Waveshare esp32 c5 lcd 1 47
rgbmatrix: add serial ABC row-addressing mode
…x-argument-validation Validate RGBMatrix arguments before reserving a display bus
Implement RTC on Zephyr
Two failure modes that cost real debugging time, written down so the next board port does not rediscover them: - Replacing the partitions node without re-adding "ranges;" mis-links the whole image on any SoC whose flash base is not 0x0, and on RP2040 also silently drops the second-stage bootloader. Fixed on the RP2040 and STM32WBA overlays in ffd62e1; documented here as a property to keep, and as a check that check_partitions.py now performs. - Undersizing or misaligning the settings partition makes NVS refuse to mount, so bt_enable() fails before opening the HCI driver and the application sees an unexplained OSError from "import _bleio". Also notes the counterpart parity rule that constrains where the space can come from. Refers to the boards/<vendor>/<board>/ layout. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
tyeth
force-pushed
the
zephyr-cp-overlay-pitfalls-doc
branch
from
October 3, 2026 13:39
74e414d to
d68c306
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
14 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Rework of the README half of #13, against the current
boards/<vendor>/<board>/layout. Documentation only.Two flash-layout mistakes in board overlays that each cost real debugging time and fail far from the cause:
ranges;on a rebuiltpartitionsnode. Now historical —ffd62e1c59restored it on the RP2040 and STM32WBA overlays (zephyr-cp: RP2040 boards are unbootable — overlays drop "ranges" from the partitions node #6, zephyr-cp: stm32wba65i_dk1 overlay has the same missing "ranges" defect as the RP2040 boards #7) — so this is framed as why the property matters and do not drop it when copying an overlay, not as a live bug. Also points atcptools/check_partitions.py, which now flags a partition resolving outside its flash device.nvs_mount()refuses fewer than two erase sectors (-EINVAL), and a partition too small to hold one gives-EDOM; both surface as a bareOSErrorfromimport _bleio.main's RP2040/RP2350storageis one 4K sector, which is fine without BT and one short with it — the fix on zephyr-cp: enable BLE on the Raspberry Pi Pico 2 W #4/zephyr-cp: enable BLE on the Raspberry Pi Pico W #14 grows it into the code partition. Also records thecounterpartparity rule (nvm/circuitpymust matchports/raspberrypi), which is what constrains where that space can come from.pre-commit run --files ports/zephyr-cp/README.mdpasses.🤖 Generated with Claude Code