2017-05-24 21:25 GMT+09:00 Julian Anastasov <ja@xxxxxx>:
> The IPVS part from patch looks good but can be extended
> to also remove rcu_read_lock and rcu_read_unlock from:
> 1. all app_conn_bind methods because ip_vs_bind_app() is called
> always under RCU lock from ip_vs_try_bind_dest() and ip_vs_conn_new().
> I.e. from sctp_app_conn_bind, tcp_app_conn_bind and udp_app_conn_bind.
> 2. ip_vs_xmit.c, all locks
> IMHO, using comments instead of locks is not needed, it is
> enough that the commit message explains why RCU locks are removed
I agree that. so I'll remove comments in v2 patch.
> More details:
> In IPVS we have the following contexts:
> - packet RX/TX: does not need locks because packets come from hooks
> - sync msg RX: backup server uses RCU locks while registering new conns
> - ip_vs_ctl.c: configuration get/set, RCU locks needed
> - xt_ipvs.c: It is a NF match
This comments are very helpful for me.
> As result, rcu_read_lock and rcu_read_unlock can be removed from:
> - ip_vs_core.c: all
> - ip_vs_ctl.c:
> - from ip_vs_has_real_service
> - all other places need the RCU locking
> - ip_vs_ftp.c: all
> - ip_vs_proto_sctp.c: all
> - ip_vs_proto_tcp.c: all
> - ip_vs_proto_udp.c: all
> - ip_vs_xmit.c: all (contains only packet processing)
> Locks should remain in:
> - ip_vs_conn.c
> - ip_vs_pe.c
> - ip_vs_sync.c
> Julian Anastasov <ja@xxxxxx>
Thank you so much for helpful review!
I'll make a v2 patch included your suggestion.
Thanks a lot!
To unsubscribe from this list: send the line "unsubscribe lvs-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html