Functional watchdog response definitions on TLF35584

Announcements

From sunburn to sun earn – we’ve got the power! Watch our #poweringgreen videos now.

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
nsyed
Level 5
Level 5
5 likes given 100 sign-ins 50 sign-ins
TLF35584 device’s functional watchdog has question-answer challenger implementation. TLF device generates a question and MCU is expected to respond back with correct answers (RESP3 to RESP0). I have following question with respect to functional watchdog on this device

  • Does SBC generates the questions in a random fashion or will it always be from 0x0 to 0xF ?
  • Response from the MCU, how it should be implemented ?
Can it be a simple lookup table based search (key of the search is question from SBC)
On other SBC’s which I worked with, MCU can use a XOR/Shift/arithmetic based operations to generate key based on the question.
Is there any such arithmetic operation which can be applied to this TLF device ?

I prefer MCU to use ALU operation to determine the key, using this approach I can verify the integrity of the ALU.
0 Likes
1 Solution
NeMa_4793301
Level 6
Level 6
10 likes received 10 solutions authored 5 solutions authored
If you really want an ALU involved, you could create a table with the expected answers with an XOR/shift applied.

I consider that overkill, given the extremely high diagnostic coverage of a lockstep core, but you are always free to implement additional safety mechanisms.

View solution in original post

0 Likes
5 Replies
NeMa_4793301
Level 6
Level 6
10 likes received 10 solutions authored 5 solutions authored
If you're using a lockstep core: between the LBIST (latent fault checking at startup) and the continuous Lockstep Comparator, there shouldn't be a need to verify ALU integrity.

The TLF35584 team says the question is random.
0 Likes
nsyed
Level 5
Level 5
5 likes given 100 sign-ins 50 sign-ins
Any thoughts on the below question.

2. Response from the MCU, how it should be implemented ? Can it be a simple lookup table based search (key of the search is question from SBC) or it there any XOR/Shift operation which can be used to generate the answer ?
0 Likes
nsyed
Level 5
Level 5
5 likes given 100 sign-ins 50 sign-ins
UC_wrangler wrote:
If you're using a lockstep core: between the LBIST (latent fault checking at startup) and the continuous Lockstep Comparator, there shouldn't be a need to verify ALU integrity.

The TLF35584 team says the question is random.


Well it depends on the system. From component level it may not be needed, but at System level it is much more safe for an external device (SBC) to do the ALU integrity check, to prevent Single point failures.
MCU could be a single point failure in this case.
0 Likes
nsyed
Level 5
Level 5
5 likes given 100 sign-ins 50 sign-ins
Any thoughts on the below question.

2. Response from the MCU, how it should be implemented ? Can it be a simple lookup table based search (key of the search is question from SBC) or it there any XOR/Shift operation which can be used to generate the answer ?

I referred the Application note and looks like it is simple lookup table based. Just wanted to confirm if there is any better way of implementing the answer
0 Likes
NeMa_4793301
Level 6
Level 6
10 likes received 10 solutions authored 5 solutions authored
If you really want an ALU involved, you could create a table with the expected answers with an XOR/shift applied.

I consider that overkill, given the extremely high diagnostic coverage of a lockstep core, but you are always free to implement additional safety mechanisms.
0 Likes