started getting repeated "bge0: PHY read timed out" messages

Michael K. Smith - Adhost mksmith at
Thu Mar 15 06:48:24 UTC 2007

Hello Chad:

-----Original Message-----
From: owner-freebsd-questions at
[mailto:owner-freebsd-questions at] On Behalf Of Chad Leigh --
Shire.Net LLC
Sent: Wednesday, March 14, 2007 11:19 PM
To: User Questions
Subject: Re: started getting repeated "bge0: PHY read timed out"

On Mar 6, 2007, at 9:55 PM, Chad Leigh -- Shire.Net LLC wrote:

> On Mar 6, 2007, at 5:08 PM, Michael K. Smith - Adhost wrote:
>> Have you looked at the output of 'netstat -i' to see if there are
>> interface errors?  Also, have you looked at the switch-side interface
>> for errors, buffer problems, etc. (if that's possible)?  Finally,  
>> have
>> you swapped ports/cables on the switch?
>> Regards,
>> Mike
> Hi Mike
> Right now it seems to be behaving.  netstat -i shows no issues and   
> I have not had the problem recur again once I forced it from auto  
> negotiation which ended up at 1000tx full duplex to now be 100tx at  
> full duplex.  I don't know if it is related to the problem or not  
> or is a coincidence that once I did that the problem has not  
> arisen.  It happened about 3 times today before forcing the  
> connection to be 100tx.

OK, this problem has not resurfaced since I hardwired the settings on  
the port to 100tx and full duplex.

I have another system with the same board that got the same problem  
last Fall when I upgraded from 5.4 to 6.0 or 6.1.  The problem this  
time was on 6.2.  I have another one of these boards running on a  
system with 5.4 without issue at gigabit speed.  It appears that  
something happened in the bge driver from 5.x to 6.x at gigabit  
speeds, at least with some of the supported chipsets.



Have you considered hard-setting the speed/duplex to 1000/Full instead
of 100/Full?  There may be some issues in the autonegotiation happening
between switch and server.  We used to see some of this early on in
inter-vendor GigE connections; perhaps the switch vendor and the FreeBSD
devels are reading the standards differently.


More information about the freebsd-questions mailing list