On Thu, 17 Nov 2022 12:26:13 +0000 "Richard W.M. Jones" <rjones(a)redhat.com> wrote:This is a hack. Ideally there'd be a way to build a "non-production" build of passt which would turn off all the encapsulation features. They are not relevant for fuzzing and simply add overhead.I'm not sure how quantitatively relevant this is, but I was thinking about cases where sandboxing or "security" features cause issues (not necessarily security-relevant ones) that would be discovered by fuzzing. Partially fitting example: https://archives.passt.top/passt-dev/20221115012400.2240327-1-sbrivio@redha… there, perror() in glibc results in a dup() call, with seccomp terminating the proceess, in a way that was totally unexpected to me. Should fuzzing trigger a case like this one, without a seccomp filter loaded, we won't notice. Now, the guest affecting its own availability isn't security relevant, and that's the worst that can happen, but still it would be something to fix. Considering this, I'm actually more inclined to polish your hack (into an 'afl' Makefile target or similar). -- Stefano