>Number:         22980
>Category:       kern
>Synopsis:       Panic: tcp_ouput REXMT
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Sep 27 21:23:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Christian Biere
>Release:        NetBSD 1.6W
>Organization:
>Environment:
System: NetBSD cyclonus 1.6W NetBSD 1.6W (STARSCREAM) #4: Fri Aug 22
00:03:06 CEST 2003
bin@cyclonus:/usr/obj/sys/arch/i386/compile/STARSCREAM i386
Architecture: i386 Machine: i386

>Description:

After an reboot I found the following messages in /var/log/messages:

I'm not sure whether these warnings are really related. I've seen them
from time to time bug ignored them.

Sep 25 01:31:21 cyclonus /netbsd: WARNING: mclpool limit reached;
increase NMBCLUSTERS Sep 25 01:31:21 cyclonus /netbsd: rtk0: unable to
allocate Tx cluster Sep 25 01:31:21 cyclonus /netbsd: rtk0: unable to
allocate Tx cluster Sep 25 01:31:21 cyclonus last message repeated 5
times

[...]

The above messages repeat for hours.

Finally the panic:

Sep 25 12:24:59 cyclonus /netbsd: panic: tcp_output REXMT
Sep 25 12:24:59 cyclonus /netbsd: Begin traceback...
Sep 25 12:24:59 cyclonus /netbsd:
tcp_setpersist(c0c2f738,6481bbd9,c0100b4c,0,4) at
netbsd:tcp_setpersist+0x2f Sep 25 12:24:59 cyclonus /netbsd:
tcp_timer_persist(c0c2f738,c02f4e64,c016cddc,c0295b60,c0785ba0) at
netbsd:tcp_timer_persist+0xc7 Sep 25 12:24:59 cyclonus /netbsd:
softclock(0,c02f4e58,c021a9c4,c02f4e64,0) at netbsd:softclock+0x1f6 Sep
25 12:24:59 cyclonus /netbsd:
softintr_dispatch(0,d36c0010,d36c0030,c02f0010,c01e0010) at
netbsd:softintr_dispatch+0xb9 Sep 25 12:24:59 cyclonus /netbsd:
Xsoftclock() at netbsd:Xsoftclock+0x25 Sep 25 12:24:59 cyclonus /netbsd:
--- interrupt --- Sep 25 12:24:59 cyclonus /netbsd:
mpidle(c02b4d60,0,c02f4f40,c01d1750,0) at netbsd:mpidle Sep 25 12:24:59
cyclonus /netbsd: ltsleep(c02b4ba0,4,c0263b46,0,0) at
netbsd:ltsleep+0x20f Sep 25 12:24:59 cyclonus /netbsd:
uvm_scheduler(c02f0010,2f0000,2f9000,0,0) at netbsd:uvm_scheduler+0x75
Sep 25 12:24:59 cyclonus /netbsd: main(0,0,0,0,0) at netbsd:main+0x69a
Sep 25 12:24:59 cyclonus /netbsd: End traceback... Sep 25 12:24:59
cyclonus /netbsd: syncing disks... 5 3 1

>How-To-Repeat:

There were probably a few hundred incoming and outgoing (TCP)
connections but nothing really special. The system is connected through
a 128/768 ADSL modem per PPPoE.

>Fix:

None provided. It's a little quaint this panic was only seen on Tru64
until now as it seems (see google).
>Release-Note:
>Audit-Trail:
>Unformatted: