On 23 January 2017 at 14:31, Sascha Steinbiss <[email protected]> wrote: > Hi Arturo, > >>> I've prepared some first code [1] and will be testing it today in a QEMU >>> VM with SSSE3 disabled. If that works I'll ping upstream, OK? >> >> Ok, the patch looks good. Ping me if you need additional testing. > > Yes, please! I would be grateful for another pair of eyes. So far I've > been able to confirm that the patched Suricata builds, starts up and > stays up correctly on systems with and without SSSE3, and also emitting > the appropriate debug messages. If you have got something more > functional on one of these systems, I would be happy to hear about how > it works for you.
I don't have a big suricata deployment here... yet :-) Anyway, it works here! I tested the patch in my basic environment, with/without SSSE3 support in the cpu. Suricata just works in both cases with my basic, default ruleset. You could add: Tested-by: Arturo Borrero Gonzalez <[email protected]> > While I have been careful to keep potential performance issues (e.g. > function call overhead) to a minimum, the default MPM matcher is > obviously now no longer known at compile time. I'd leave it to upstream > to comment on how this might affect performance in practice. > Sure. I don't know the cosebase so I can't help here. > I have slightly adapted my patch and did some rewording on the commit > message, please see my 'hs_valid_platform' branch on GitHub [1]. > > Do you think it would be a good idea to adjust the install-time debconf > disclaimer message in the libhyperscan4 package if there is an > application-specific fall-back in place such as here? > E.g. notify the user that "libhyperscan4 won't work in most applications > unless they explicitly support a fall-back (such as, for instance, > Suricata)"? > That sounds good indeed. @Robert, are you willing to accept these changes? best regards.
