Date: Wed, 03 Mar 2010 23:42:59 +0900 (JST) From: Kazumaro Aoki <ka@flu.if0.org> To: freebsd-firewire@freebsd.org Subject: Re: reset with fwcontrol -R Message-ID: <20100303.234259.193731492.ka@flu.if0.org> In-Reply-To: <1267546296.2264.0.camel@localhost.localdomain> References: <20100207.220956.59557668.ka@flu.if0.org> <20100303.001031.193727057.ka@flu.if0.org> <1267546296.2264.0.camel@localhost.localdomain>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
From: Sean Bruno <seanbru@yahoo-inc.com>
Subject: Re: reset with fwcontrol -R
Date: Tue, 02 Mar 2010 08:11:36 -0800
>> > In recent month, the system is in good condition, but today morning,
>> > system crashed and made a crash dump.
>> (snip)
>> > # kgdb /boot/kernel/kernel vmcore.1
>> (snip)
>> > Fatal trap 12: page fault while in kernel mode
>> > current process = 42 (fw0_taskq)
>> > panic: page fault
>> (snip)
>> > (kgdb) bt
>> (snip)
>> > #7 0xc05f6d87 in fwohci_task_dma (arg=0xc5662000, pending=1)
>> > at /home/usr/src/sys/dev/firewire/fwohci.c:2206
>> (snip)
>> I noticed lines 2197-2199 are
>> #if 0
>> dump_db(sc, dmach);
>> #endif
>
>> Are the information generated by the commented out the lines useful to
>> analyze the problem?
>
> I've never used that code, but I think it is more debugging. If it is
> useful for debugging this issue, I can make it a sysctl for debugging.
I found that the following sequence cause a panic with high
probability:
1. fwcontrol -R foo.dv
2. terminate fwcontrol using ctrl-c
3. start fwcontrol -R bar.dv before finishing fwcontrol -R foo.dv
The used fwcontrol command is imported from FreeBSD6R, and back-trace
of kgdb command always says line 2206 cause a panic.
I attached the log file generated by line 2198.
Can you fix the problem?
# I write send-pr when it seems to take long time in this mailing-list.
When I used fwcontrol in the original distribution (FreeBSD 7.2R/i386)
with "fwcontrol -M dv -R bar.dv", I got the error message:
fwcontrol: main: Error opening firewire controller #0 /dev/fw0.0 in recv_data
: Device busy
and the system does not panic.
It certainly prohibits a panic, but it also seems to prevent to show
the firewire status that is generated fwcontrol command without any
option.
I also met a panic with another sequence caused by kldunload of
firewire.ko module. However, I usually do not kldunload, and I do not
write the details. If you are interested in the panic, I can write
the back-trace information.
Best regards,
Kazumaro
[-- Attachment #2 --]
j@},[bfV.LBIILK)$b- c)]u"Hù rq PSS1ӏeS mڴ7b 7gӔu`c6#Fw\?xZ"\%o͡ї_ l۴oխHW`K"l?N. k*}p-<Ud)6_$r4ߣ|+՝qf']E?q
EiJSҔW(iR6)MiJSҔ4)MiJSҔ4)MiJSҔ4)MiJSҔ4Cڔ SgS&Ҕ4)M 3QΥ JSҔJ7z:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100303.234259.193731492.ka>