Hello all
What does it mean if the last 9 items in dmesg from a 2.2.19 kernel say
"recvmsg bug: copied 870A11AD seq 0"? Same sequence number each time.
A kernel (known) bug? Some bogus TCP segments?
Somehow this has something (dunno how much) to do with INN and pullnews.
This morning I had 43 inews's stuck in tcp_recvmsg at the schedule() just
before found_ok_skb and news.daily was stuck. telnet to localhost port 119
didn't produce the banner as it normally does.
Weird thing too is that inews would have been connected to 127.0.0.1, out
of harm's way.
Unfortunately I very much doubt I can reproduce this - it's the first time
I've seen it (INN has been running since last year, but not continuously).
linux/net/ipv4/tcp.c <tcp_recvmsg>:
skb = skb_peek(&sk->receive_queue);
do {
if (!skb)
break;
/* Now that we have two receive queues this
* shouldn't happen.
*/
if (before(*seq, TCP_SKB_CB(skb)->seq)) {
printk(KERN_INFO "recvmsg bug: copied %X seq %X\n",
*seq, TCP_SKB_CB(skb)->seq);
break;
}
[I'd greatly appreciate Cc'ed replies.]
--
[email protected] is probably better to bookmark than any
employer-specific email address I may have appearing in the headers.
Vanity page: http://www.tsct.co.za/~berndj/ Hmm, appears to be down.
On Mon, 14 Oct 2002, Bernd Jendrissek wrote:
> Hello all
>
> What does it mean if the last 9 items in dmesg from a 2.2.19 kernel say
> "recvmsg bug: copied 870A11AD seq 0"? Same sequence number each time.
>
> A kernel (known) bug? Some bogus TCP segments?
It's been reported a few times, but I'm not sure if it's been resolved.
Can you please post the version of the compiler which compiled the kernel,
what kind of hardware you're running on, and if it's a stock kernel or has
vendor/other patches applied.
- James
--
James Morris
<[email protected]>
On Mon, Oct 14, 2002 at 11:16:05PM +1000, James Morris wrote:
> On Mon, 14 Oct 2002, Bernd Jendrissek wrote:
>
> > Hello all
> >
> > What does it mean if the last 9 items in dmesg from a 2.2.19 kernel say
> > "recvmsg bug: copied 870A11AD seq 0"? Same sequence number each time.
> >
> > A kernel (known) bug? Some bogus TCP segments?
>
> It's been reported a few times, but I'm not sure if it's been resolved.
>
> Can you please post the version of the compiler which compiled the kernel,
> what kind of hardware you're running on, and if it's a stock kernel or has
> vendor/other patches applied.
$ cat /proc/version
Linux version 2.2.19 ([email protected]) (gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)) #2 Wed Aug 1 13:51:15 SAST 2001
$ lspci
00:00.0 Host bridge: Intel Corporation 440BX/ZX - 82443BX/ZX Host bridge (rev 03)
00:01.0 PCI bridge: Intel Corporation 440BX/ZX - 82443BX/ZX AGP bridge (rev 03)
00:07.0 ISA bridge: Intel Corporation 82371AB PIIX4 ISA (rev 02)
00:07.1 IDE interface: Intel Corporation 82371AB PIIX4 IDE (rev 01)
00:07.2 USB Controller: Intel Corporation 82371AB PIIX4 USB (rev 01)
00:07.3 Bridge: Intel Corporation 82371AB PIIX4 ACPI (rev 02)
00:08.0 VGA compatible controller: S3 Inc. ViRGE/DX or /GX (rev 01)
$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 6
model name : Celeron (Mendocino)
stepping : 0
cpu MHz : 334.094
cache size : 128 KB
fdiv_bug : no
hlt_bug : no
sep_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr
bogomips : 666.82
Network card, hmm, some ne2k type ISA card. The phrase "8390" also springs
to mind.
$ less /var/log/dmesg
...
ne.c:v1.10 9/23/94 Donald Becker ([email protected])
NE*000 ethercard probe at 0x300: 00 00 e8 2f b9 ca
eth0: NE2000 found at 0x300, using IRQ 5.
...
Stock kernel. No patches. Unfortunately I've lost .config
BTW ipv6 module is loaded, for extra coolness points. Not that I use it! :)
I can provide a dissassembly of tcp_recvmsg; please quote "c0162daf" - that
is where the inews's were all sleeping.
Hmm, curiouser and curiouser. I haven't booted the (this) machine since
77 days ago; and now dmesg shows another couple of messages:
recvmsg bug: copied 870A11AD seq 0
recvmsg bug: copied 870A11AD seq 0
tcp_close: socket already locked!
Same "copied" sequence number! Pity the kernel doesn't timestamp its
messages. Why doesn't syslog catch these messages into /var/log/messages?
I'll leave the machine running (unless it gets *very* weird) so we can peek
its memory. Is it too late to copy /proc/kcore now?
Hah! I bet this has *something* to do with the whole spiel:
$ netstat -nt |grep human-brain
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address Foreign Address State
tcp 6 0 127.0.0.1:3287 127.0.0.1:119 CLOSE
When I su to root and add netstat's -p option, this socket doesn't have an
owner process. And I use rtin, not some pthread-monster, so it's not that.
[I'd greatly appreciate Cc'ed replies]
--
[email protected] is probably better to bookmark than any
employer-specific email address I may have appearing in the headers.
Vanity page: http://www.tsct.co.za/~berndj/