Author: afedorov Date: Thu Dec 17 18:15:07 2020 New Revision: 368727 URL: https://svnweb.freebsd.org/changeset/base/368727
Log: [ng_socket] Don't take the SOCKBUF_LOCK() twice in the RX data path. This is just a minor optimization, but it's sensitive. This gives an improvement of 30-50 kpps. Reviewed by: kp, markj, glebius, lutz_donnerhacke.de Approved by: vmaffione (mentor) Sponsored by: vstack.com Differential Revision: https://reviews.freebsd.org/D27382 Modified: head/sys/netgraph/ng_socket.c Modified: head/sys/netgraph/ng_socket.c ============================================================================== --- head/sys/netgraph/ng_socket.c Thu Dec 17 17:21:12 2020 (r368726) +++ head/sys/netgraph/ng_socket.c Thu Dec 17 18:15:07 2020 (r368727) @@ -987,6 +987,8 @@ ngs_rcvmsg(node_p node, item_p item, hook_p lasthook) m_freem(m); return (ENOBUFS); } + + /* sorwakeup_locked () releases the lock internally. */ sorwakeup_locked(so); return (error); @@ -1025,12 +1027,17 @@ ngs_rcvdata(hook_p hook, item_p item) addr->sg_data[addrlen] = '\0'; /* Try to tell the socket which hook it came in on. */ - if (sbappendaddr(&so->so_rcv, (struct sockaddr *)addr, m, NULL) == 0) { + SOCKBUF_LOCK(&so->so_rcv); + if (sbappendaddr_locked(&so->so_rcv, (struct sockaddr *)addr, m, + NULL) == 0) { + SOCKBUF_UNLOCK(&so->so_rcv); m_freem(m); TRAP_ERROR; return (ENOBUFS); } - sorwakeup(so); + + /* sorwakeup_locked () releases the lock internally. */ + sorwakeup_locked(so); return (0); } _______________________________________________ svn-src-all@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"
