> > Don't we already have that with sysevents (ddi_log_sysevent(9F))?
>
> Ah cute :).
>
> What about non-device-driver code that doesn't have a dev_info though?
You mean like log_sysevent()? And what DDI-compliant code are you
thinking of that wouldn't have a dev_info_t?
> IIUC it seems to be a way for drivers to get mostly-opaque stuff to
> driver-specific userland code, rather than a generic kernel->userspace
> event reporting channel.
Actually, it's used throughout Solaris, especially in userland.
> Think 'PF_ROUTE on steroids' type of thing, or essentially PF_ROUTE but
> with wider, more future-proof type and attribute fields (and maybe a way
> for listeners to 'subscribe' to certain classes of events) - extensible
> enough to use for reporting all the other classes of events which the
> NWAM lists (tunnel states, dladm related events, wifi events ("New AP,
> ESSID: ....., <etc>").
Sysevents already seems to have all that functionality.
--
meem
