Pins 37-48 were rotated as a block in 0.2.0 (BOOT0/GPIOs on the wrong pads); fixed in 0.2.1 from ST open pin data
Versions up to 0.2.0 had 12 pin numbers wrong (pins 37-48 rotated as a block). Fixed in 0.2.1.
In the KiCad symbol and the Fusion .lbr of 0.2.0, PA15 was on pad 48 instead of 37, PD0-PD3 on 44-47 instead of 38-41, and PB3-PB9 on 37-43 instead of 42-48. The footprint was correct, because pads go by number, so a board laid out from 0.2.0 has those signals on the wrong physical pins.
- Found by: stm32-designer, while building a board, then confirmed against ST's machine-readable pinout, STM32_open_pin_data
mcu/STM32G071C(6-8-B)Tx.xml. - Cause: the 0.2.0 pinout came from the older single-model extractor (ds2sf), which attached a contiguous run of names to the wrong numbers. Every pin name existed and every pad was used, so the existing checks could not see a permutation.
- Fix (0.2.1): each affected pin keeps its name, description and symbol graphic and takes the number ST gives it. This covers
page.json, the README pin table, the KiCad symbol, the regenerated .lbr (adom-lbr validateOK) and the regenerated viewers. The live page now matches ST on every signal pin. - The brief no longer says "Datasheet-verified pinout"; it says what was actually checked.
- If you used 0.2.0: re-import the symbol/library from 0.2.1 and re-check any schematic or board that used pins 37-48.
The same audit ran on every STM32 page on the wiki. This page and one other were the only ones with wrong pins. New STM32 pages are now checked against ST's data before they can publish.