[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [csmith-bugs] csmith hang

Hi Ahmed,

Nice to see you here :)

Although Csmith often runs pretty quickly, there are, buried within it, some algorithms that probably scale cubically or worse with respect to function size, number of pointer variables, and stuff like that. So, as you have noticed, it sometimes runs for a long long time.

My position would be that we should definitely fix real termination errors (actual infinite loops) but that fixing the "very long run-time" problem is probably impractical given the level of resources that we have. Unfortunately, telling these cases apart is hard! And probably not worth it.

On the bright side, we have done some looking into the problem of teaching a random testcase generator how to "escape" from the program it is generating -- while still producing a valid test case -- when some resource budget (time, memory, or whatever) is exceeded. It appears that this is feasible, but we're not going to try it out in Csmith.


On 10/18/11 11:56 PM, Ahmed Charles wrote:
The following seed seems to cause csmith (github version) to hang. If
these hangs aren't interesting, just let me know.

  * Generator: csmith 2.1.0
  * Git version: a5e5fd1
  * Options:   --bitfields --packed-struct --output test4.c
  * Seed:      386941147

#include "csmith.h"

static long __undefined;