Debugging BigFork -> Random failues


(Sudarshan Venkataraman) #1

Hey ,

My partner and I are done with ASST3.2. But the thing is we dont have a consistent 3.2 to start with 3.3 , 3.2 randomly/very-rarely/sometimes-frequently doesn’t give us full score and most of the times it it because “BIGFORK” fails(maybe resulting in vm-stability being skipped) . After finding one unsuccessful run , I am still confused as to which part to debug. I am guessing this is a synchronization issue (but where do I start looking to correct this ?).

Any input is highly welcomed. Tips on debugging big-fork are highly encouraged.

What does “2 failures” indicate ?


(Siddharth Rajiv Shah) #2

We faced an issue with our big-fork as well. It was a locking problem in our as_destroy. We used GDB and manually iterated multiple steps of big-fork. We than also manually assumed various scenarios of multiple processes forking and went through each line of as_copy on paper. This proved really helpful for us.

Hope this helps!


(Sudarshan Venkataraman) #3

Thanks for the input . Your response brings up my next question ,
my as-destroy ,as_copy , as_create arent synchronized. Could this randomness be an effect of that ?


(Siddharth Rajiv Shah) #4

Ours as_destroy function is not synchronized. The part where we free pages in memory, we did not add any synchronization there which created issues for us.


(Sudarshan Venkataraman) #5

Will look into it , thanks.