ot_otbn: Connect Recoverable alert to OTBN's error bits.
#170
Conversation
|
@AlexJones0 Do we have OT test coverage for this behaviour? |
|
@luismarques I think we might? The OTBN smoketest started failing after the recent alert catching changes were merged to OpenTitan Edit: yes, looks like there are two OT tests that are now passing with this change: |
…bits The Fatal alert was connected for the different fatal faults, but the recoverable alert for recoverable errors such as bad instruction address errors, invalid sideloaded key errors, illegal instructions, etc. was not connected. This commit simply adds the logic for this alert. Also take this opportunity to add a trace on OTBN alerts, and to modify the implementation so that ALERT_TEST and recoverable alerts are transient (i.e. immediately cleared backed to 0 after being sent once), whereas fatal alerts are sticky (also fatal, as they can't be cleared). Signed-off-by: Alex Jones <alex.jones@lowrisc.org>
Signed-off-by: Alex Jones <alex.jones@lowrisc.org>
AlexJones0
force-pushed
the
otbn_recov_alerts
branch
from
ef8e719 to
84f9476
Compare
|
@rivos-eblot is this ok to merge for you? |
Sorry I missed your comment. Can you take care of the alert_test transient thing, for ex. |
The alert_test transience should have already been added by the last force push - after replying to your comment I added it 😃 |
Ooops, I missed it. Ok then. |
4 participants
The fatal alert was connected for the different fatal faults, but the recoverable alert for recoverable errors (such as bad instruction address errors, invalid sideloaded key errors, illegal instructions, etc.) was not connected. This commit simply adds the logic for this alert.