Skip to content

process-node: the bench's own wiring profile, loaded by name - #14

Merged
wardviaene merged 1 commit into
mainfrom
process-node-wiring-profile
Nov 18, 2026
Merged

wardviaene merged 1 commit into
mainfrom
process-node-wiring-profile

Conversation

@wardviaene

Copy link
Copy Markdown
Member

Why

The process-node bench is not wired like the pod's default pins (UART RX LA5, I2C SDA LA1 / SCL LA2): its console is on LA3/LA4, I2C SCL on LA1 and SDA on LA2, and LA5 is the alarm. The tests hard-coded their channels and ran on whatever profile was the pod's default.

What

  • process-node-stm32/wiring/process-node-stm32.json: the README pin table as a schema 2 profile (UART, I2C, SWD with NRST and target/stm32f4x.cfg, signals ALARM, STROBE (bidir), SENSOR_UART_RX/TX, SENSOR_SPI_SCK/MOSI/MISO/CS, 3.3 V, internal 5 V). The SPI sensor pins are signals, not spi_* roles, because here the pod is the SPI device (it drives MISO), so the role directions would be the wrong way round.
  • tests/conftest.py: a bench_profile fixture loads it by name (bp.load_wiring("process-node-stm32")); wiring reads every LA channel from it. When the pod lacks it, the run stops with the save command. A run that names its own wiring (--benchpod-wiring, --benchpod-wiring-profile, BENCHPOD_WIRING, BENCHPOD_WIRING_PROFILE) keeps that one. The pod's default profile is neither used nor changed.
  • README: a "Wiring profile" section on storing it (benchpod wiring save ... --name process-node-stm32) and loading it (wiring_profile= in the SDK and in MCP connect).

Tests (benchpod-30a9f1, fw 3.10.11 / gw 58, SDK from embeddedci-com/embeddedci-python#113)

Profile stored on the pod as process-node-stm32; full suite with BENCH_NEW_WIRING=1 BENCH_LA6_RESISTOR=1 and the main node build flashed through the profile's SWD settings: 178 passed, 2 skipped, 2 failed.

  • test_sensors_faults.py::test_uart_drop passed when run again (with the rest of the faults and UART sensor files).
  • test_timing.py::test_alarm_follows_its_reading fails with TLS handshake with 192.168.1.150:8443: Connection reset by peer when it opens its second connection for the parallel capture. It fails the same way with the conftest from main, so it is not caused by this change (it looks like the pod refuses a second LAN TLS connection).

🤖 Generated with Claude Code

wiring/process-node-stm32.json is the bench's full pin table as a wiring profile (schema 2): UART
RX LA3 / TX LA4, I2C SCL LA1 / SDA LA2, SWD LA11/LA12 with NRST, the named signals ALARM, STROBE,
SENSOR_UART_* and SENSOR_SPI_*, 3.3 V, internal 5 V. The bench is not wired like the pod's default
pins (UART RX LA5, I2C SDA LA1 / SCL LA2), which tests used to fall back on.

The conftest loads the profile by name (bp.load_wiring) and reads every LA channel from it; it stops
with the save command when the pod does not have it. A run that names its own wiring keeps it. The
README says how to store it on the pod.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@wardviaene
wardviaene merged commit bdd4443 into main Nov 18, 2026
@wardviaene
wardviaene deleted the process-node-wiring-profile branch November 18, 2026 20:50
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant