[ofw] assert on ipoib_port.c @ 3629

Smith, Stan stan.smith at intel.com
Wed Nov 12 09:06:09 PST 2008


Tzachi Dar wrote:
> I'll try your suggestion, and let you know if it works or not.
>
> In any case, if I was not clear enough, this happens when the machine
> is going up and not down, so I don't think that halt was called there.

Can you provide more environmental details?
So far I have rebooted 35 WinOF RC5 installed systems (win2k3/8/XP/Vista[x86,x64,ia64]) and not witnessed this problem? Yes - some systems were running a debug ipoib.sys.
How do you reproduce it?
What switches are in your fabric?

Thanks,

Stan.


>
> Thanks
> Tzachi
>
>> -----Original Message-----
>> From: Alex Estrin [mailto:alex.estrin at qlogic.com]
>> Sent: Tuesday, November 11, 2008 11:03 PM
>> To: Tzachi Dar; ofw at lists.openfabrics.org
>> Subject: RE: [ofw] assert on ipoib_port.c @ 3629
>>
>> Tzachi,
>>
>>>
>>> While rebooting a machine with our latest code I have received the
>>> assert above:
>>
>> I think network can issue DHCPDISCOVER when adapter indicates
>> connect status change: Please see ipoib_port.c @ lines
>> 4573-4579.(rev. 1748). Reporting
>> NDIS_STATUS_MEDIA_DISCONNECT/CONNECT to the Netowrk while port is
>> destroying probably a bad idea.
>> WDK doc chapter "Indicating Connection Status" says: "...
>> Additional requirements for indicating media connection
>> status changes for miniport drivers are as follows: ...
>> Halting
>> A miniport driver must not indicate any media connection
>> status changes when NDIS calls MiniportHalt. "
>>
>> I wonder if network will try to send that bogus request on
>> reboot with lines 4573-4579 removed.
>>
>> Thanks,
>> Alex.
>>
> _______________________________________________
> ofw mailing list
> ofw at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw




More information about the ofw mailing list