Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 22 Mar 2010 15:40:13 +0200
From:      "Prokofyev S.P." <proks@skylinetele.com>
To:        pyunyh@gmail.com
Cc:        jfv@FreeBSD.org, freebsd-net@freebsd.org
Subject:   Re: Please pay attention to fix bug kern/141285
Message-ID:  <4BA7733D.1020602@skylinetele.com>
In-Reply-To: <20100319174709.GQ9373@michelle.cdnetworks.com>
References:  <4BA38CEC.9060205@skylinetele.com> <20100319174450.GP9373@michelle.cdnetworks.com> <20100319174709.GQ9373@michelle.cdnetworks.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 19.03.2010 19:47, Pyun YongHyeon wrote:
> On Fri, Mar 19, 2010 at 10:44:50AM -0700, Pyun YongHyeon wrote:
>    
>> On Fri, Mar 19, 2010 at 04:40:44PM +0200, Prokofyev S.P. wrote:
>>      
>>> Hi ALL !
>>>
>>> Please pay attention to fix bug kern/141285(kern/141843) !
>>>
>>>        
>> igb(4) also has a similar issue but it seems igb(4) does not even
>> advertise IFCAP_VLAN_HWFILTER capability. igb(4) may have to remove
>> VLAN event handler or should implement IFCAP_VLAN_HWFILTER to
>> support VLAN hardware filtering.
>>
>> I have a patch for the hardware VLAN filtering of em(4). But it
>> wouldn't address the issue reported in the PR. The root cause of
>> issue was em(4) wants to reset controller whenever new VLAN is
>> registered/unregistered. I'm not sure this is requirement of
>> hardware. If this is requirement of hardware there is no way to
>> avoid the controller reset unless you disable vlanhwfilter feature.
>>
>> #ifconfig em0 -vlanhwfilter
>>
>> em(4) in HEAD disabled VLAN hardware filtering by default so if you
>> use that version you wouldn't encounter the issue again. Attached
>> patch is small diff for VLAN hardware filtering which tries to
>> avoid unnecessary controller reset and added missing lock. If
>> hardware allows dynamic changing of VLAN filtering table we could
>> completely bypass the controller reset. Jack may know the details.
>>      
> Oops, posted old patch. Here is new one.
>    
Thank you Pyun.
I have applied your patch (rebuild/reinstall kernel) and have lost 
access to test server after reboot.

ping Test Server from Server A
Server A (10.25.223.4) -> Test Server(10.25.223.2)

tcpdump -enp -i em0 (via console on Test Server):
14:39:44.752754 00:1b:fc:af:a1:b4 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q 
(0x8100), length 64: vlan 77, p 0, ethertype ARP, Request who-has 
10.25.223.2 tell 10.25.223.4, length 46
14:39:44.752765 00:30:48:96:cc:56 > 00:1b:fc:af:a1:b4, ethertype 802.1Q 
(0x8100), length 46: vlan 77, p 0, ethertype ARP, Reply 10.25.223.2 
is-at 00:30:48:96:cc:56, length 28
.........
but do not see reply on Server A.

I "see" the TestServer after ifconfig em0 down/up. The problem with bug 
kern/141285 disappear.

If  write ifconfig_em0="up -vlanhwtag" in /etc/rc.conf, then after 
reboot I have network access to TestServer, but this is not very nice 
decision of a problem.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BA7733D.1020602>