Hi,
I am trying to use Daikon and Kvasir on a C program. First I tried to follow the instructions in the manual and pipe the output of Kvasir directly to Daikon. However, it took so long (more than 24 hours) that I had to cancel it.
Do we have an estimate on the time it takes to infer the invariants for the bzip2 code in the c-example when use these commands on a typical core i7 processor with 32GB of RAM:
- kvasir-dtrace --decls-only ./bzip2 bzip2.c
- kvasir-dtrace --no-dyncomp --dtrace-no-decls --dtrace-file=- ./bzip2 bzip2.c | java -cp $DAIKONDIR/daikon.jar daikon.Daikon daikon-output/bzip2.decls -
Then I tried to produce the trace just for on function (another large c program): First, I dumped the ppt-list file and then I commented all ppts except one. However, when I run Kvasir again to produce the traces, there is "nonce" info for every function calls in the trace file (this time I did not pipe the output to understand why it is slow).
I did the same procedure this time for var-list and it, still, took a long time to produce the trace just for one integer variable.
Am I missing anything? I have attached the original and edited var-list and ppt-list file.
Thanks
Ahmad