Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
10/10/11 22:35
Read: times


 
#184168 - Problem With This
Responding to: ???'s previous message
The problem with instruction set tests is how do you test something using the very thing that you are intending to test??

What happens of some of the first instructions out of reset at the ones that fail?

What if an instruction failure ends up altering the execution flow of the test program itself causing it to crash?

If one of the instructions executed in the code logic that tries to verify the results of the IUT (instruction under test) happens to be one that fails can you ever trust the test results?

At best you can say that if an Instruction Set Test happens to execute smoothly to its end then you can say - Everything appears OK. On the other side of things any failure may be indicate something has gone wrong. You cannot be sure of what caused it but you know there was a failure.

If you do intend to write an instruction set test make sure to enable the watch Dog timer so that there is some chance of recovering the MCU if it goes out into the weeds during the test.

Michael Karas


List of 8 messages in thread
TopicAuthorDate
MCS51 instruction set validation test            01/01/70 00:00      
   Problem With This            01/01/70 00:00      
      Test bootstrap            01/01/70 00:00      
         I know of none            01/01/70 00:00      
            interesting            01/01/70 00:00      
            Bare minimum test            01/01/70 00:00      
               be careful what you wish for ...            01/01/70 00:00      
   2 cents            01/01/70 00:00      

Back to Subject List