cvs commit: src/sys/kern kern_descrip.c uipc_socket.c uipc_syscalls.c uipc_usrreq.c src/sys/net raw_cb.c raw_usrreq.c src/sys/netatm atm_socket.c src/sys/netatalk ddp_pcb.c src/sys/netgraph ng_ksocket.c src/sys/netgraph/bluetooth/socket ...

[Alfred Perlstein]

* Robert Watson <rwatson at freebsd.org> [040614 12:24] wrote:
> 
> We chose to maintain the existing naming scheme for sockets present in the
> code since it's origins in BSD, and consistent with other BSD platforms. 
> Otherwise, I generally agree :-).  Given the volume of other changes going
> in here, I was reluctant to introduce non-functional changes in order to
> ease merging.  Once we have the basic version of locking in place, we will
> have the opportunity to revisit this (and a great many other things). 
> 
> One thing I should point out, though, is that the reference counting in
> sockets isn't a simple reference count, since in addition to so_count,
> there's also a flag indicating whether a file descriptor reference is
> present, and an un-counted reference from the pcb to the socket, which is
> also considered "real".  The various
> sofree()/sotryfree()/sorele()/soref()/..  APIs reflect this complexity,
> and hence some inconsistency with a more simple API.

Thank you, I am familiar with SS_NOFDREF. :(

-- 
- Alfred Perlstein
- Research Engineering Development Inc.
- email: bright at mu.org cell: 408-480-4684