Re: Ultrasparc III+ kernel panic

BERTRAND Joël a écrit :
> BERTRAND Joël a écrit :
>> Julian Coleman a écrit :
>>> Hi,
>>>
>>> Is it possible to swap the graphics cards in these two?
>>>
>>>> First server (stable) :
>>>> - 2 * USIII/750
>>>> - XVR-500 (no screen, no keyboard)
>>>>
>>>> Second server (unstable) :
>>>> - 2 * USIII+/900
>>>> - Creator-3D (text console)
>>>
>>> The reason that I ask is that I'm seeing a number of "SIR Reset"'s
>>> happening
>>> on my U60, and it's a lot worse with 2 CPU's installed, or with a
>>> Creator-3D
>>> as the console.  With 1 CPU, I haven't seen one, and with serial
>>> console,
>>> they happen less frequently.  I haven't been able to match the resets to
>>> system load - they are more likely when the machine is busy (e.g.
>>> running
>>> /etc/daily), but can also happen when it's idle.  On the other hand,
>>> with 2
>>> CPU's and a serial console, it managed 10 days of continuous pkgsrc
>>> building
>>> before resetting.
>>>
>>> I'm not sure if this is related (different hardware) or not.
>>> However, as
>>> the problem is worse on the U60 with the C-3D as console, there might be
>>> something related to UPA.
>>>
>>> For comparison, I have an SB2000 with:
>>>
>>>    501-6230 system board
>>>    2 * 501-6485 1200MHz US III Cu
>>>    501-4788 Creator-3D (console)
>>>    375-3181 XVR-100
>>>    2 * Fujitsu MAW3300FC in RAID 1
>>>
>>> as my desktop, and this is stable.
>>
>>      Julian,
>>
>>      I have made some tests. First server (XVR500, 2*US-III/750) remains
>> in same configuration and is stable.
>>
>>      I don't understand why I cannot put a XVR500 in second one. System
>> starts but screen was unusable (display takes a bad resolution and was
>> scrambled). In a first time, I thought my second XVR500 was dying. But I
>> have a third Blade 2000 in the same place and this XVR500 runs fine in
>> third Blade 2000... Why ? I don't know. And all PCI slot run fine in
>> server where XVR500 doesn't work...
>>
>>      Thus, I have swapped both servers. I have tested third one with
>> Creator 3D (new creator3D, new memory, one new CPU, one retired from
>> second server...) and it hangs like second one. The last night, I have
>> installed XVR500 and removed Creator 3D). Now, server is stable enough
>> to build NetBSD from sources and pkgsrc :
>>
>> load averages:  7.36,  7.20,  7.02;   up 0+11:33:42        11:54:04
>> 145 processes: 5 runnable, 137 sleeping, 1 zombie, 2 on CPU
>> CPU states: 79.4% user, 0.0% nice, 20.6% system, 0.0% int, 0.0% idle
>> Memory: 1078M Act, 543M In, 9192K Wired, 71M Exec, 1264M File, 42M Free
>> Swap: 8050M Total, 143M Used, 7907M Free
>>
>>      I suspect a bug somewhere in UPA support as you said.

	Bad news. Last night, this server panics twice when idle with :

Feb 23 07:58:27 legendre /netbsd: cpu0: data fault: pc=f000934c 
rpc=103b435e0 addr=1ffee8000
Feb 23 07:58:27 legendre /netbsd: Skipping crash dump on recursive panic
Feb 23 07:58:27 legendre /netbsd: panic: kernel fault
Feb 23 07:58:27 legendre /netbsd: cpu0: Begin traceback...
Feb 23 07:58:27 legendre /netbsd: cpu0: End traceback...
Feb 23 07:58:27 legendre /netbsd: cpu1: shutting down
Feb 23 07:58:27 legendre /netbsd: cpu0: rebooting

	Next saturday, I will try to remove all but quad-hme PCI adapters.

	Regards,

	JKB