Timothy Trippel
Timothy Trippel
We can ignore the kokoro error here just FYI.
After getting up to speed on the above, it seems while the hardware has been altered to fix part of the underlying issue here (the fatal-ness, i.e., "stickiness", of multi-bit...
After further discussion with @cfrantz, the performance hit of option 1 may not be that bad, and is worth measuring.
The idea would be that when copying the code section from flash to the HMAC input FIFO, we would preload a temp value with something that would poison the signature...
It is just pseudocode to illustrate the idea, there are several ways we can take of potential compiler optimizations, `volatile` being one.
> @timothytrippel could you check out the OTTF change? (Final commit) OTTF changes LGTM.
Any luck on tracking this down @pamaury ?
> I started to review all the uses of `trigger_lc_transition` and here are my conclusions (but feel free to double-check): > > * the `bitstream_otp_ctrl_functest` bitstream has execution disabled although...
> I have been working on `manuf_cp_device_info_flash_wr` and I think I found the root cause. This test starts with a bitstream with ROM execution disabled so there is no issue...
I think we can close this now. The last outstanding issue is tracked in #20580. Feel free to reopen again if needed.