[U-Boot] [PATCH v2] Kconfig: cmd: Make networking command dependent on NET

[Heinrich Schuchardt]

On 03/01/2018 12:14 AM, Tom Rini wrote:
> On Thu, Mar 01, 2018 at 12:05:52AM +0100, Heinrich Schuchardt wrote:
>> On 02/28/2018 02:09 AM, Tom Rini wrote:
>>> On Mon, Feb 26, 2018 at 04:01:02PM +0100, Michal Simek wrote:
>>>
>>>> Enable networking command only when NET is enabled.
>>>> And remove selecting NET for CMD_NET
>>>>
>>>> Signed-off-by: Michal Simek <michal.simek at xilinx.com>
>>>> ---
>>>>
>>>> Changes in v2:
>>>> - Check several boards by hand.
>>>>
>>>> There is a huge impact on defconfigs because of select NET.
>>>> But this change makes sense to do but it just needs to be syncup
>>>> properly. Do we have a tool for this kind of change?
>>>
>>> So, I've applied this patch, with a good bit of modification.  What I
>>> wasn't clear about before, sorry, was that we need to make NET default y
>>> in here too.  However, we also have some decent areas of the code that
>>> use "CONFIG_CMD_NET" when it really means "CONFIG_NET", at least
>>> conceptually.  But in order to make everything work as-is today, and
>>> leave these fixes to a later point in time (as they are fixes and should
>>> happen) we change some areas today that reference CONFIG_NET to
>>> reference CONFIG_CMD_NET.  Once net/ gets cleaned up, we can use
>>> CONFIG_NET without CONFIG_CMD_NET in more area.  It's also not quite
>>> 100% size-neutral as the topic_miami* boards were playing some games
>>> that can't quite be done as they were before, but I believe the end
>>> result is they can now more easily and thoroughly disable the networking
>>> stuff that intended to be removed.
>>>
>>>
>>
>> Hello Tom,
>>
>> in spite of you comments above I do not understand why you changed
>> cmd/bootefi.c to depend on CONFIG_CMD_NET instead of CONFIG_NET.
> 
> Because the underlying generic network functionality that efi_loader
> uses is actually gated under CONFIG_CMD_NET and not CONFIG_NET.  This
> should be corrected, in the long term.
> 
>> This was not part of Michal's patch.
>>
>> I would prefer if changes would be sent to the list for review *before*
>> being applied.
>>
>> As the patch that you applied is not Michal's patch your authorship should
>> be documented in the git log.
> 
> Ah, I forgot to add my S-o-B?  Oops, that was unintentional.

It is not that your Signed-of-by is missing. But your signature follows
directly Michal's. So it looks as if you applied the patch signed by
Michal unchanged.

Best regards

Heinrich

> 
>> I cannot see any reason why network support should be disabled in bootefi if
>> there are no network commands available.
> 
> In theory and concept, I agree.  In current implementation, that is not
> the case and I would welcome further changes that make the code reflect
> the general intention.
>