Date: Thu, 29 Dec 2011 16:18:42 -0500 From: John Baldwin <jhb@freebsd.org> To: Pawel Jakub Dawidek <pjd@freebsd.org> Cc: =?utf-8?q?=D0=9A=D0=BE=D0=BD=D1=8C=D0=BA=D0=BE=D0=B2?=, Andre Oppermann <andre@freebsd.org>, freebsd-net@freebsd.org, =?utf-8?q?_=D0=95=D0=B2=D0=B3=D0=B5=D0=BD=D0=B8=D0=B9?= <kes-kes@yandex.ru>, freebsd-current@freebsd.org, Kostik Belousov <kostikbel@gmail.com>, Lawrence Stewart <lstewart@freebsd.org> Subject: Re: 9.0-RC1 panic in tcp_input: negative winow. Message-ID: <201112291618.43170.jhb@freebsd.org> In-Reply-To: <20111229202501.GA1889@garage.freebsd.pl> References: <20111022084931.GD1697@garage.freebsd.pl> <201112291112.59912.jhb@freebsd.org> <20111229202501.GA1889@garage.freebsd.pl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday, December 29, 2011 3:25:02 pm Pawel Jakub Dawidek wrote: > On Thu, Dec 29, 2011 at 11:12:59AM -0500, John Baldwin wrote: > > On Sunday, December 25, 2011 11:01:33 am =D0=9A=D0=BE=D0=BD=D1=8C=D0=BA= =D0=BE=D0=B2 =D0=95=D0=B2=D0=B3=D0=B5=D0=BD=D0=B8=D0=B9 wrote: > > > =D0=97=D0=B4=D1=80=D0=B0=D0=B2=D1=81=D1=82=D0=B2=D1=83=D0=B9=D1=82=D0= =B5, John. > > >=20 > > > =D0=92=D1=8B =D0=BF=D0=B8=D1=81=D0=B0=D0=BB=D0=B8 20 =D0=B4=D0=B5=D0= =BA=D0=B0=D0=B1=D1=80=D1=8F 2011 =D0=B3., 16:52:44: > > >=20 > > > JB> On Saturday, December 17, 2011 6:21:27 pm Pawel Jakub Dawidek wro= te: > > > >> On Mon, Dec 12, 2011 at 11:00:23AM -0500, John Baldwin wrote: > > > >> > An update. I've sent Pawel a testing patch to see if my hypothe= sis is correct > > > >> > (www.freebsd.org/~jhb/patches/tcp_negwin_test.patch). If it is = then I intend > > > >> > to commit www.freebsd.org/~jhb/patches/tcp_negwin2.patch as the = fix. > > > >>=20 > > > >> Unfortunately it paniced today. Take a look at: > > > >>=20 > > > >> http://people.freebsd.org/~pjd/misc/tcp_panic.jpg > > >=20 > > > JB> Ok, the one use case I was worried about is happening regularly b= efore your > > > JB> panic, so that is good. Can you use gdb to figure out which call= to > > > JB> tcp_output() is actually panic'ing? I wonder if it is this case: > > >=20 > > > JB> /* > > > JB> * Return any desired output. > > > JB> */ > > > JB> if (needoutput || (tp->t_flags & TF_ACKNOW)) { > > > JB> (void) tcp_output(tp); > > > JB> /* XXX: Debug */ > > > JB> KASSERT(SEQ_GEQ(tp->rcv_adv, tp->rcv_nxt), > > > JB> ("tcp_input: negative window after ACK")); > > >=20 > > > JB> And if 'needoutput' is true, but TF_ACKNOW is not set, and tcp_ou= tput() decides > > > JB> to not do anything. I've updated tcp_negwin_test.patch to not pa= nic if that call > > > JB> to tcp_output() doesn't actually send a packet. Please re-test. > > >=20 > > >=20 > > > # uname -a > > > FreeBSD meta-up 9.0-PRERELEASE FreeBSD 9.0-PRERELEASE #4: Sat Dec 24 = 13:59:20 EET 2011 @:/usr/obj/usr/src/sys/KES_KERN_v10 i386 > > >=20 > > > rebooting once per day. Now I compile kernel with debug options. > > > Can you advice me which and where I find debug info when it will > > > reboting next time? so I can help to debug problem > >=20 > > Are you using the patch at the URL above (tcp_negwin_test.patch)? If n= ot, > > can you try applying that patch and seeing if you still get any panics? >=20 > I applied 1.5 days ago, so far now panics and no other messages. > I modified the patch a bit to not panic, but print a message when panic > was suppose to happen. This box is too valuable for me to panic it too > often. Because there were no debug messages I understand that the > scenerio didn't happen yet and not that the problem is fixed, right? Yes. It would be best to see the messages logged to be safe. Thanks. =2D-=20 John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201112291618.43170.jhb>