Small patch..
Andre Oppermann
andre at freebsd.org
Thu May 8 22:30:52 UTC 2008
John Baldwin wrote:
> At work all the log(LOG_DEBUG, ...) statements in the TCP code in 7.x are
> spamming the heck out of our dmesg so I am #if 0'ing all of them out and
> while doing so ran into this case. Specifically, it doesn't actually bump
sysctl net.inet.tcp.log_debug=0 is simpler than #if 0 and has the same effect.
In RELENG_7 it was disabled by default.
> the stat counter unless it succeeds in allocating memory to log the debug
> message. All the other places in the syncache and tcp input code always bump
> the stats, so this patch fixes it to do that.
Your fix is correct.
--
Andre
> Index: tcp_syncache.c
> ===================================================================
> RCS file: /usr/cvs/src/sys/netinet/tcp_syncache.c,v
> retrieving revision 1.143
> diff -u -r1.143 tcp_syncache.c
> --- tcp_syncache.c 19 Apr 2008 03:39:17 -0000 1.143
> +++ tcp_syncache.c 8 May 2008 20:22:21 -0000
> @@ -567,10 +567,11 @@
> "connection attempt aborted by remote endpoint\n",
> s, __func__);
> tcpstat.tcps_sc_reset++;
> - } else if ((s = tcp_log_addrs(inc, th, NULL, NULL))) {
> - log(LOG_DEBUG, "%s; %s: RST with invalid SEQ %u != IRS %u "
> - "(+WND %u), segment ignored\n",
> - s, __func__, th->th_seq, sc->sc_irs, sc->sc_wnd);
> + } else {
> + if ((s = tcp_log_addrs(inc, th, NULL, NULL)))
> + log(LOG_DEBUG, "%s; %s: RST with invalid SEQ %u != "
> + "IRS %u (+WND %u), segment ignored\n",
> + s, __func__, th->th_seq, sc->sc_irs, sc->sc_wnd);
> tcpstat.tcps_badrst++;
> }
>
>
More information about the freebsd-net
mailing list