Skip to content

Conversation

@rswarbrick
Copy link
Contributor

@rswarbrick rswarbrick commented Jan 3, 2026

The use of ASSERT_*_TRIGGER_ALERT means that they were never really doing anything sensible (but were presumably burning lots of energy!) Since we prove correctness of the alert system separately, we can make this all prove in a minute or two using the "_IN" variants.

The tweak I made to prim_fifo_sync when depth=1 stopped the hack that this FPV test previously used from working. Rather than figure out something to black-box, let's just use a stopat and put those assertions into a separate task.

I'd love to say that this now works... Unfortunately, FpvSecCmRBignumOnehotCheck_A sees a counterexample. That's because the assertion is just false: the bignum write-enable failure can be squashed for a cycle by a reported SW error and the stopats in use then make a counterexample possible.

Tidying that up properly will be quite a bit of work, which I'll do in a follow-up PR.

@rswarbrick rswarbrick requested a review from vogelpi January 3, 2026 15:19
@rswarbrick rswarbrick requested a review from a team as a code owner January 3, 2026 15:19
@rswarbrick rswarbrick requested review from eshapira and removed request for a team January 3, 2026 15:19
@rswarbrick rswarbrick added Component:FPV FPV issue: formal testbench, property, etc IP:otbn labels Jan 3, 2026
The use of ASSERT_*_TRIGGER_ALERT means that they were never really
doing anything sensible (but were presumably burning lots of energy!)
Since we prove correctness of the alert system separately, we can make
this all prove in a minute or two using the "_IN" variants.

The tweak I made to prim_fifo_sync when depth=1 stopped the hack that
this FPV test previously used from working. Rather than figure out
something to black-box, let's just use a stopat and put those
assertions into a separate task.

I'd love to say that this now works... Unfortunately,
FpvSecCmRBignumOnehotCheck_A sees a counterexample. That's because the
assertion is just false: the bignum write-enable failure can be
squashed for a cycle by a reported SW error and the stopats in use
then make a counterexample possible.

Tidying that up properly will be quite a bit of work, which I'll do in
a follow-up PR.

Signed-off-by: Rupert Swarbrick <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Component:FPV FPV issue: formal testbench, property, etc IP:otbn

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant