Dara Hazeghi wrote:
Hello,
I've been using c-reduce quite extensively to reduce testcases for C compiler
crash bugs, and it's been invaluable. However I also have a substantial
collection of what I believe to be wrong code bugs that I'm trying to reduce.
I've tried using c-reduce for those as well, and I'm running into difficulties
detecting undefined/unspecified behavior during the reduction process. Based
on the c-reduce paper, I gather you found kcc and frama-c to be sufficient for
that task. I've just given kcc a whirl and it seemed to be prohibitively slow
for non-trivial testcases. I was wondering if there was a special way you used
kcc to avoid this slowness, or what other tool(s) I might use to keep the
reduction process both sound and reasonably fast. Many thanks,
Have you tried Frama-C? it is reasonably fast and well-maintained. Based
on my experience in reducing test cases generated by Csmith, it's pretty
good.
- Yang
