Skip to content

type: worklist status: current summary: Open work on the base and programmer firmware: SWD fault diagnostics for field bases, unbuilt SPI SWD sources, and SWD speed ideas for a future base.

Bases Worklist

Delete an item when it is done.

  • SWD fault diagnostics on field bases. bittag-base-v7 and tag-breakout-base-l432-v1 log SWD faults on a USB serial port (branch base-0x81), which found the 0x81 cause (design/investigations/2026-10-0x81-dbg-sleep.md). Field bases cannot be reflashed with that; a vendor command beside STLINK_TAGBASE_IDENTIFY (F6 99) could return a ring buffer of the last faults (acknowledgement, register or address, retries) for any host tool to dump after a run.
  • tag-breakout-base-l432-v1/src/ll_swd_spi.c is not built. The target compiles ll_swd.c. A note in that directory recorded that ll_swd_spi.c did not work properly on macOS with ChibiOS 21.11. Fix it or remove it, and remove src/ll_swd_spi.c-back beside it.
  • Unused local sources. tag-breakout-base-l432-u375-1v8/src/ll_swd.c and tag-base-c071/src/stlink-local.c and src/ll_swd.c are not in their targets' project.mk.
  • Hybrid SPI SWD for a faster F042 base. The F042 bases still bitbang everything. In the optimized tag-breakout-base-l432-v1/src/ll_swd.c, every shift-out width is a multiple of 8 bits (8, 16, 24, 32) and the odd widths are all shift-in (5, 9, 24), so SPI can carry every shift-out phase. That is the split tag-base-c071 and tag-breakout-base-l432-u375-1v8 already use. It was considered for a future base rather than a retrofit.
  • Hot SWD helpers in RAM. On a future base with a faster MCU and SRAM headroom, consider placing only the hottest bitbang helpers in RAM. The F042 base is not a candidate: its build uses essentially all of its SRAM.