From: Ding Tianhong <dingtianhong@huawei.com>
To: Vlad Yasevich <vyasevich@gmail.com>, <vyasevic@redhat.com>,
<kaber@trash.net>, <davem@davemloft.net>, <edumazet@google.com>,
<makita.toshiaki@lab.ntt.co.jp>
Cc: <netdev@vger.kernel.org>, John Fastabend <john.r.fastabend@intel.com>
Subject: Re: [PATCH net-next v2 RESEND 1/4] macvlan: support mac address changes when fwd_priv is enable
Date: Sat, 14 Jun 2014 13:01:05 +0800 [thread overview]
Message-ID: <539BD711.30501@huawei.com> (raw)
In-Reply-To: <539AFCF1.6020402@gmail.com>
On 2014/6/13 21:30, Vlad Yasevich wrote:
> On 06/12/2014 11:10 PM, Ding Tianhong wrote:
>> On 2014/6/12 22:24, Vlad Yasevich wrote:
>>> On 06/11/2014 09:45 PM, Ding Tianhong wrote:
>>>> On 2014/6/10 0:51, Vlad Yasevich wrote:
>>>>> [.. CC John Fastabend <john.r.fastabend@intel.com> ]
>>>>>
>>>>> On 06/07/2014 03:45 AM, Ding Tianhong wrote:
>>>>>> If lowerdev supports L2 forwarding offload, the macvlan's hw address
>>>>>> will be set to the rar of the lowerdev and no need to set uc list,
>>>>>> and when the macvlan's hw address changes, the macvlan should remove
>>>>>> the old fwd and rebuild a new fwd for the macvlan.
>>>>>>
>>>>>> Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
>>>>>> ---
>>>>>> drivers/net/macvlan.c | 21 +++++++++++++++++++++
>>>>>> 1 file changed, 21 insertions(+)
>>>>>>
>>>>>> diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
>>>>>> index 453d55a..67485ab 100644
>>>>>> --- a/drivers/net/macvlan.c
>>>>>> +++ b/drivers/net/macvlan.c
>>>>>> @@ -523,6 +523,27 @@ static int macvlan_sync_address(struct net_device *dev, unsigned char *addr)
>>>>>> if (macvlan_addr_busy(vlan->port, addr))
>>>>>> return -EBUSY;
>>>>>>
>>>>>> + if (vlan->fwd_priv) {
>>>>>> + lowerdev->netdev_ops->ndo_dfwd_del_station(lowerdev,
>>>>>> + vlan->fwd_priv);
>>>>>> + vlan->fwd_priv = NULL;
>>>>>> + ether_addr_copy(dev->perm_addr, dev->dev_addr);
>>>>>> + ether_addr_copy(dev->dev_addr, addr);
>>>>>> + vlan->fwd_priv =
>>>>>> + lowerdev->netdev_ops->ndo_dfwd_add_station(lowerdev,
>>>>>> + dev);
>>>>>> + /* If we get a NULL pointer back, or if we get an error
>>>>>> + * then we should restore the old mac address and fwd.
>>>>>> + */
>>>>>> + if (IS_ERR_OR_NULL(vlan->fwd_priv)) {
>>>>>> + ether_addr_copy(dev->dev_addr, dev->perm_addr);
>>>>>> + vlan->fwd_priv =
>>>>>> + lowerdev->netdev_ops->ndo_dfwd_add_station(lowerdev,
>>>>>> + dev);
>>>>>> + return -EINVAL;
>>>>>> + }
>>>>>> + return 0;
>>>>>> + }
>>>>>
>>>>> Calling del_stations add_station causes all sorts of VMDQ/ring
>>>>> operations to happen... Not sure if we can do that while we have a live
>>>>> macvlan/macvtap that is capable of transmitting data.
>>>>>
>>>>> Wouldn't it be sufficient to have a call to call to update with mac
>>>>> filter with the appropriate VMDQ.
>>>>>
>>>>> John, I defer to you here. The above looks really heavy-weight and
>>>>> I am not sure if its correct.
>>>>>
>>>>> Thanks
>>>>> -vlad
>>>>
>>>> I have used two 82599 nic to test this problem, I modify the ixgbe drivers and set the L2 forwarding offload default,
>>>> without this patch, when I change the macvlan mac address, the kernel oops, and if I apply this patch, the ixgbe could
>>>> work well with the maclvan, so I don't think this solution has problem.
>>>
>>> Have you tried changing the address while at the same time
>>> transmitting data through the macvlan?
>>>
>>> -vlad
>>>
>>
>> Yes, I tried, the lowerdev will be closed when calling ixgbe_fwd_del, then the transmitting was broken and about 1 second later,
>> the transmitting restored and start to work again.
>>
>> 64 bytes from 192.168.10.8: icmp_seq=14 ttl=64 time=0.109 ms
>> 64 bytes from 192.168.10.8: icmp_seq=15 ttl=64 time=0.112 ms
>> ping: sendmsg: Network is down
>> 64 bytes from 192.168.10.8: icmp_seq=17 ttl=64 time=1000 ms
>> 64 bytes from 192.168.10.8: icmp_seq=18 ttl=64 time=0.147 ms
>>
>
> Right. This is what I figured would happen and doesn't happen
> with any other device when changing the mac address.
>
> If this behavior is OK with other folks, I'll shut up, but IMO
> we shouldn't just correctly update the mac filter and leave the
> VMDQs alone.
>
> -vlad
>
Yes, according to your important suggestion, we could further analysis this problem,
and can you explain more about how to fix it better, just like the VMDQ, I am not very
clear about this, thanks for any feedback.
Regards
Ding
>> Ding
>>
>>>>
>>>> Thanks
>>>> Ding
>>>>
>>>>>
>>>>>> if (!vlan->port->passthru) {
>>>>>> err = dev_uc_add(lowerdev, addr);
>>>>>> if (err)
>>>>>>
>>>>>
>>>>>
>>>>> .
>>>>>
>>>>
>>>>
>>>
>>>
>>> .
>>>
>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe netdev" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>
>
>
> .
>
next prev parent reply other threads:[~2014-06-14 5:03 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-06-07 7:45 [PATCH net-next v2 RESEND 0/4] macvlan: fix some problem if mac address changes Ding Tianhong
2014-06-07 7:45 ` [PATCH net-next v2 RESEND 1/4] macvlan: support mac address changes when fwd_priv is enable Ding Tianhong
2014-06-09 16:51 ` Vlad Yasevich
2014-06-12 1:45 ` Ding Tianhong
2014-06-12 14:24 ` Vlad Yasevich
2014-06-13 3:10 ` Ding Tianhong
2014-06-13 13:30 ` Vlad Yasevich
2014-06-14 5:01 ` Ding Tianhong [this message]
2014-06-07 7:45 ` [PATCH net-next v2 RESEND 2/4] net: dev: revert the mac address when notifier failed Ding Tianhong
2014-06-07 7:45 ` [PATCH net-next v2 RESEND 3/4] macvlan: don't set the same mac address for non-passthru mode Ding Tianhong
2014-06-07 7:45 ` [PATCH net-next v2 RESEND 4/4] net: dev: don't set the same mac address for netdev Ding Tianhong
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=539BD711.30501@huawei.com \
--to=dingtianhong@huawei.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=john.r.fastabend@intel.com \
--cc=kaber@trash.net \
--cc=makita.toshiaki@lab.ntt.co.jp \
--cc=netdev@vger.kernel.org \
--cc=vyasevic@redhat.com \
--cc=vyasevich@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).