Re: Fresh install of BETA2: console error about a LOR...

From: Julien Gabel <jpeg_at_thilelli.net>
Date: Mon, 15 Aug 2005 23:09:33 +0200 (CEST)
>>>> During the configuration phase, just after installing 6.0-BETA2 on a
>>>> notebook, i got the following message on the console:
>>>>
>>>> lock order reversal
>>>>  1st 0xc097b240 UMA lock (UMA lock) _at_ /usr/src/sys/vm/uma_core.c:1494
>>>>  2nd 0xc1060144 system map (system map) _at_
>>>> /usr/src/sys/vm/vm_kern.c:295

>>> looks pretty much like
>>> 	http://sources.zabbadoz.net/freebsd/lor.html#110

>> As reported by Thierry Herbelot, the #109 appears on a UP and SMP
>> system, but the #110 seems to be SMP only (at least is only reproduced
>> on a HTT system with SMP kernel).
>>
>> When i got the second LOR, it was followed by a panic and a crash
>> (three
>> times on four).  It was during 'make buildworld' each time, with this
>> message:
>>
>> Memory modified after free 0xc2af2500(256) val=deadc0df _at_ 0xc2af2508
>> panic: receive: m == 0 so->so_rcv.sb_cc == 4294967295
>> cpuid = 0
>> KDB: enter: panic
>> [thread pid 3908 tid 100152 ]
>> Stopped at      kdb_enter+0x30: leave
>> db>
>>
>> I didn't have a crash dump yet, since it was during a post-installation
>> phase.  I am currently running an UP kernel to avoid problem, but can
>> switch to a SMP one if somebody is interested.

> in case you get it a gain please type
> db> where
> so we'll know the code path where this happens. That might help or not.

Ok.  Recompile a kernel with SMP support, then rerun the buildworld test.
The system crashed in the middle of the run, "as expected".  Here it is:

panic: receive: m == 0 so->so_rcv.sb_cc == 4294967295
[...]
db> where
db> Tracing pid 38892 tid 100125
kdb_enter
panic
soreceive
fifo_read_f
dofileread
kern_read_v
read
syscall
 Xint0x80_syscall
-- syscall

Note: pid 38892 is a 'make' process.

Because i haven't a serial console on the notebook, i can't have the
complete message (without typing it by hand).  I noted the beginning
of each line; if all the content of each line is needed i will try to
get a cable to get a serial console (tomorrow maybe) to be able to
copy and paste the result here.

I get a crash dump this time, but after uncompressing the dump under
/var/crash, i get the following error:
# cd /usr/obj/usr/src/sys/BOBOCHE
# kgdb kernel.debug /var/crash/vmcore.3
kgdb: cannot read IdlePTD

Because i am not a developer, i don't really know what do next.

Here is the content of info.3:
# cat info.3
Dump header from device /dev/ad0s1b
  Architecture: i386
  Architecture Version: 33554432
  Dump Length: 1071185920B (1021 MB)
  Blocksize: 512
  Dumptime: Mon Aug 15 22:09:54 2005
  Hostname: boboche.thilelli.net
  Magic: FreeBSD Kernel Dump
  Version String: FreeBSD 6.0-BETA2 #0: Mon Aug 15 21:04:09 CEST 2005
    root_at_boboche.thilelli.net:/usr/obj/usr/src/sys/BOBOCHE
  Panic String: receive: m == 0 so->so_rcv.sb_cc == 4294967295
  Dump Parity: 3704850192
  Bounds: 3
  Dump Status: good

What to do/try next?

--
-jpeg.
Received on Mon Aug 15 2005 - 19:09:41 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:41 UTC