ercoppa
I think the code was written in this was because the symbolic helper takes as input the `ret` value computed by `INDEX_op_setcond_i32`.
Notice that this proposal may help fix #13 (I tested it), #12 (I am not sure), #6 (It can for sure improve the situation).
One possible fix could be to: 1) modify `tcg_gen_qemu_ld_i{32, 64}` replacing `load_size` with `memop` (which encodes the length and the sign). 2) modify `sym_load_guest_i{32, 64}` to propagate `mmop` to `sym_load_guest_internal`...
@vanhauser-thc thank you :) We'll see in the next hours the results. I feel that there could be some crashes: is there any way to get the logs? The parameters...
> there seems to be a bug in fuzzolic which is visible in libpcap: I need to look at the logs and maybe run it locally to be sure (we...
@vanhauser-thc when looking at the logs for `libpcap`, fuzzolic has spent the whole time waiting for new inputs from AFL, which produced "only" three inputs. Fuzzolic Z3 generated three new...
> libpcap only progresses if there is a solver that can solve memcmp(buf + x, MAGIC_TYPE, 4) which normal afl-fuzz cannot. > yes it is. and that first solve that...
@vanhauser-thc I have debugged a bit libpcap and I see two problems in the driver used by FuzzBench: 1) [minor] This [check](https://github.com/the-tcpdump-group/libpcap/blob/fbcc461fbc2bd3b98de401cc04e6a4a10614e99f/testprogs/fuzz/fuzz_both.c#L62) is something that is a bit hard to...