Skip to content

test(validation): silicon-vs-emulator lockstep for wide-decode fixes#333

Merged
w1ne merged 1 commit into
mainfrom
test/silicon-decoder-validation
Jun 22, 2026
Merged

test(validation): silicon-vs-emulator lockstep for wide-decode fixes#333
w1ne merged 1 commit into
mainfrom
test/silicon-decoder-validation

Conversation

@w1ne

@w1ne w1ne commented Jun 22, 2026

Copy link
Copy Markdown
Owner

A bare Cortex-M4 firmware exercises the LDRB.W/LDRH.W/{S,U}XT{B,H}.W/{S,U}XTA{B,H}.W decode fixes (PRs #328/#329/#331) and writes results to RAM. Validated byte-for-byte on a real NUCLEO-L476RG over SWD; lockstep.yaml is a CI-runnable (no-hardware) regression asserting the labwired stm32l476 model matches the silicon-captured values.

…ecoder fixes

A bare Cortex-M4 firmware runs the exact instructions covered by the cortex_m
wide-decode unit tests (LDRB.W/LDRH.W signedness, {S,U}XT{B,H}.W, and the
extend-and-add {S,U}XTA{B,H}.W) and writes results to a fixed RAM array.

Validated byte-for-byte on a real NUCLEO-L476RG over SWD: all nine results
match the emulator and the unit-test expectations exactly. lockstep.yaml is a
CI-runnable (no hardware) regression that runs the same ELF on the stm32l476
model and asserts the silicon-captured values, so any future decoder change
that diverges from silicon fails. README documents the hardware re-validation
procedure.
@w1ne w1ne merged commit 33fe382 into main Jun 22, 2026
3 of 4 checks passed
@w1ne w1ne deleted the test/silicon-decoder-validation branch June 22, 2026 12:16
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