Exit Flashrom/AVRDude binmode from terminal?

bz2 · June 2, 2026, 3:39pm

Hello! I would like to use my Bus Pirate 5 to automate some development tasks, using multiple binmodes at different times.

I can switch to binmode BPIO2 flatbuffer interface and back out again via the text terminal interface.

However, though I can switch to binmode Legacy Binary Mode for Flashrom and AVRdude (EXPERIMENTAL), the only way I can switch back out again is by pressing the Bus Pirate’s physical button (or by unplugging/replugging the USB cable).

Would it be possible to permit switching out of the Flashrom/AVRdude binmode from the terminal, without requiring the user to physically interact with the Bus Pirate?

ian · June 2, 2026, 4:33pm

I think it should be possible to a new escape command. I’m not familiar with that code, but I will have a look and see if I can add an easy exit point.

Dreg · August 17, 2026, 1:19am

Done! @bz2 Sorry it took me so long to get this done. I’m active on the forum again now. It’s included in this PR with a ton of improvements. Once @ian merges it, you’ll be able to use it.

github.com/DangerousPrototypes/BusPirate5-firmware

Legacy binary mode: fix flashrom and avrdude support (#307)

main ← legacy-binmode-flashrom

opened 01:18AM - 17 Aug 26 UTC

therealdreg

+1649 -175

This makes the legacy binary mode (flashrom / AVRdude) reliable enough that I t…hink it can drop the EXPERIMENTAL and BETA labels. The previous version had a few bugs that could show up as corrupted bytes or a stalled host, so reads were not always repeatable. Those are fixed, and a few smaller things were tidied along the way: * SPI speeds now match what flashrom documents, so asking for 1M really gives 1MHz. * Errors are reported to the host instead of leaving it waiting. * The mode recovers on its own if the host disappears mid command. * Pin voltage readings pause while the mode runs. On older Bus Pirate 5 hardware the analog mux that measures them briefly touches the SPI pins, which was enough to corrupt transfers. Newer boards are not affected. This is mentioned on screen so it does not look like a fault. * You can now exit with 'q' on the terminal, as well as with the button. ### Tested on hardware Everything below used random data with verified entropy, written and read back with a byte for byte comparison. | Client | Device | Board | Result | |----------|-------------------|-------------|--------| | flashrom | Winbond W25Q64 | BP6 | 56MB read over 7 SPI speeds, identical every time | | flashrom | Atmel AT25DF021A | BP6, BP5r8 | Full erase, write and verify, read back at 7 speeds | | avrdude | ATmega328P | BP6, BP5r8 | Flash and EEPROM | | avrdude | ATmega32U4 | BP6 | Flash | | avrdude | ATmega2560 | BP6, BP5r8 | Flash and EEPROM, exercises the >64KB address path | ### Test tool `tools/legacy4third_usbtest.py` is a host side test for this mode. Every reply in the protocol has a length fixed by the request, so a lost or duplicated byte shows up right away. It replays the real flashrom and avrdude sequences, sweeps the opcode space, and checks recovery from garbage input and from a host that goes away. Combined runs across both boards: around 287,000 commands and 125MB of payload with no framing errors, plus 50 rounds of random garbage that it recovered from every time. ### Why this should be steadier Most of the earlier trouble came from the USB side being driven from both cores at once. It now stays on one core and talks to the protocol through the queues the firmware already had for this, which removes a whole class of intermittent failures. Error paths also keep the reply length intact, so a problem surfaces as a clean error rather than a host that sits and waits.

bz2 · August 17, 2026, 2:20am

Sweet, thanks a bunch!!

ian · August 31, 2026, 2:44pm

Merged, and in the process tested and verified that the auto review is working again.