Repository navigation
Bug fix: CMSIS-DAP probe could stop responding (#87)
Since v0.2.0 (RTIC 2 migration) the CMSIS-DAP firmwares deferred command processing to a software task via spawn(). When a USB event arrived while a command was still being processed, spawn() failed silently and the request already read from the endpoint was never processed; the probe then stopped answering until a USB bus reset. This was easiest to hit with --features set_clock (which changes command timing) but could occur with any build.
dap_process is now a hardware task on an unused interrupt (PIO1_IRQ_1 on RP2040, SW1_IRQ on RP2350, DAC on SAMD21) pended from the USB interrupt, which never drops a request. Verified on hardware: a back-to-back CMSIS-DAP stress that wedged v0.3.0 + set_clock within ~10 cycles every time now runs 27,000 commands without a stall.
Affected firmwares: Raspberry Pi Pico, Raspberry Pi Pico 2, XIAO RP2040, XIAO M0 (CMSIS-DAP binaries). The GDB server binaries are unchanged.
What's Changed
Thanks to @hender14 for the detailed report and bisection in #87.
Full Changelog: v0.3.0...v0.3.1
What's Changed
- doc: SWEST28 EmbLT 版に LT スライドを更新 (Dabaoボードであそんでみた) by @ciniml in #86
- cmsis-dap: run dap_process as a pended hardware task (fixes #87) by @ciniml in #88
Full Changelog: v0.3.0...v0.3.1