Re: [PATCH net-next] net: vxlan: do not use vxlan_net before checking event type

[Daniel Borkmann <dborkman@redhat.com>]

On 01/18/2014 06:57 PM, Cong Wang wrote:
> On Sat, Jan 18, 2014 at 9:18 AM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
>> On Fri, 2014-01-17 at 19:50 -0800, Cong Wang wrote:
>>> On Fri, Jan 17, 2014 at 10:32 AM, Daniel Borkmann <dborkman@redhat.com> wrote:
>>>>
>>>> If you want to do cleanups, whatever, I really don't care.
>>>> You had your chance to complain about that when you reviewed
>>>> the initial version ... it has nothing to do with the fix.
>>>
>>> This is not for stable, as long as it doesn't harm the readability
>>> we are free to do any cleanup's.
>>>
>>> If unsure, check Eric's patch for tunnel dst cache.
>>>
>>> BTW, I am the original author of the patch, you just updated
>>> it *trivially* and set yourself as the author. :) I don't mind, but
>>> remember that this may be not appropriate for others. At
>>> very least I didn't and don't do this myself.
>>
>> Hmm... Daniel mentioned in the changelog you wrote the initial patch,
>> and you are credited as the author of the patch, since he kept your
>> "Signed-off-by: ..." as the first one.
>
> Author == 'From: ...', you knew it, right?
>
> But WITHOUT even asking for my permission. I am sure this is
> not how we usually work. At least, why not ask me before doing
> anything? Why not give me a chance to response?
>
>> Quite frankly, keeping vxlan_handle_lowerdev_unregister() was the right
>> choice.
>>
>> Stop thinking that a function needs to be used more than once to have
>> the right to exist. Splitting code in small parts ease readability and
>> code reuse/refactor, this should be obvious to you.
>
> When did I say because that it is only used once? Please, stop guessing
> my mind.

Cong, I'm really tired of discussing this BS with you, and this is my
last mail on this topic.

You said "There is no need to keep vxlan_handle_lowerdev_unregister(),
it is too short." I, however, think keeping vxlan_handle_lowerdev_unregister()
is the right choice as it makes the code more readable, plus you clearly
agreed with the code earlier as you've given your Reviewed-by tag. You
even got your Fixes tag wrong and I do care that an actual fix for a bug
has a bit more in-depth commit message telling what's going on. I think
the message in the commit is equally important as the code itself, you
should know. Maybe, I was just in the wrong timezone, but while waiting
for a v2 and not having endless discussions about vxlan_handle_lowerdev_unregister(),
I do care that this gets fixed asap! Clearly, it seems it was an honest
mistake to do so.