[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5e7270a1-8de6-1563-4e42-df37da161b98@gmail.com>
Date: Fri, 9 Aug 2019 10:14:03 -0600
From: David Ahern <dsahern@...il.com>
To: Roopa Prabhu <roopa@...ulusnetworks.com>,
Jiri Pirko <jiri@...nulli.us>
Cc: netdev <netdev@...r.kernel.org>,
David Miller <davem@...emloft.net>,
Jakub Kicinski <jakub.kicinski@...ronome.com>,
Stephen Hemminger <sthemmin@...rosoft.com>, dcbw@...hat.com,
Michal Kubecek <mkubecek@...e.cz>,
Andrew Lunn <andrew@...n.ch>, parav@...lanox.com,
Saeed Mahameed <saeedm@...lanox.com>,
mlxsw <mlxsw@...lanox.com>
Subject: Re: [patch net-next rfc 3/7] net: rtnetlink: add commands to add and
delete alternative ifnames
On 8/9/19 9:40 AM, Roopa Prabhu wrote:
>>>> diff --git a/include/uapi/linux/rtnetlink.h b/include/uapi/linux/rtnetlink.h
>>>> index ce2a623abb75..b36cfd83eb76 100644
>>>> --- a/include/uapi/linux/rtnetlink.h
>>>> +++ b/include/uapi/linux/rtnetlink.h
>>>> @@ -164,6 +164,13 @@ enum {
>>>> RTM_GETNEXTHOP,
>>>> #define RTM_GETNEXTHOP RTM_GETNEXTHOP
>>>>
>>>> + RTM_NEWALTIFNAME = 108,
>>>> +#define RTM_NEWALTIFNAME RTM_NEWALTIFNAME
>>>> + RTM_DELALTIFNAME,
>>>> +#define RTM_DELALTIFNAME RTM_DELALTIFNAME
>>>> + RTM_GETALTIFNAME,
>>>> +#define RTM_GETALTIFNAME RTM_GETALTIFNAME
>>>> +
>>>
>>> I might have missed the prior discussion, why do we need new commands
>>> ?. can't this simply be part of RTM_*LINK and we use RTM_SETLINK to
>>> set alternate names ?
>>
>> How? This is to add/remove. How do you suggest to to add/remove by
>> setlink?
>
> to that point, I am also not sure why we have a new API For multiple
> names. I mean why support more than two names (existing old name and
> a new name to remove the length limitation) ?
>
> Your patch series addresses a very important problem (we run into this
> limitation all the time and its hard to explain it to network
> operators) and
> its already unfortunate that we have to have more than one name
> because we cannot resize the existing one.
>
> The best we can do for simpler transition/management from user-space
> is to keep the api simple..
> ie keep it close to the management of existing link attributes. Hence
> the question.
>
> I assumed this would be like alias. A single new field that can be
> referenced in lieu of the old one.
>
> Your series is very useful to many of us...but when i think about
> changing our network manager to accommodate this, I am worried about
> how many apps will have to change.
> I agree they have to change regardless but now they will have to
> listen to yet another notification and msg format for names ?
>
> (apologies for joining the thread late and if i missed prior discussion on this)
I agree with Roopa. I do not understand why new RTM commands are needed.
The existing IFLA + ifinfomsg struct give more than enough ways to id
the device for adding / deleting an alternate name.
Powered by blists - more mailing lists