On Fri, 13 Dec 2019 at 14:44, Paul Wouters <[email protected]> wrote: > > On Fri, 13 Dec 2019, Andrew Cagney wrote: > > > First a fun fact: > > > > whack_log_fd is valid IFF pluto is executing the whack event handler. > > > > Consequently any code logging to whack outside of the whack event > > handler needs to use a reference saved in either state or pending. > > > > When processing a raw message, whack_log_fd is never valid and a > > matching state has yet to be found, so trying to log to whack using: > > > > loglog(RC_LOG_SERIOUS, > > "message %s contained an unknown critical > > payload type (%s)", > > role, enum_show(&ikev2_payload_names, np)); > > > > is pointless. This means, as best I can tell, I can reduce calls like this > > to: > > > > plog_md(md, > > "message %s contained an unknown critical > > payload type (%s)", > > role, enum_show(&ikev2_payload_names, np)); > > I guess the idea is that some code can be called with an administrator > waiting at the whack socket, but sometimes the same code is timer > triggered without an administrator.
For some, yes. The above is only triggered when a message arrives. I tend to suspect the more likely scenario is that no one is really sure if/when whack is attached (cur_state, for instance, comes and goes) so people were hedging their bets. _______________________________________________ Swan-dev mailing list [email protected] https://lists.libreswan.org/mailman/listinfo/swan-dev
