Re: 2.6.24-rc4-mm1

["Kok, Auke" <auke-jan.h.kok@intel.com>]

Andrew Morton wrote:
> On Tue, 11 Dec 2007 13:26:58 -0800
> "Kok, Auke" <auke-jan.h.kok@intel.com> wrote:
> 
>> Andrew Morton wrote:
>>> On Tue, 11 Dec 2007 08:13:52 -0800 "Martin Bligh" <mbligh@google.com> wrote:
>>>
>>>>> - Lots of device IDs have been removed from the e1000 driver and moved
>>>>> over
>>>>>  to e1000e.  So if your e1000 stops working, you forgot to set
>>>>> CONFIG_E1000E.
>>>>>
>>>>>
>>>> Wouldn't it make sense to just default this to on if E1000 was on, rather
>>>> than screwing
>>>> everybody for no good reason (plus breaking all the automated testing, etc
>>>> etc)?
>>>> Much though I love random refactoring, it is fairly painful to just keep
>>>> changing the
>>>> names of things.
>>> (cc netdev and Auke)
>>>
>>> Yes, that would be very sensible.  CONFIG_E1000E should default to whatever
>>> CONFIG_E1000 was set to.
>> which is "y" for x86 and friends, ppc, arm and ia64 through 'defconfig'. the
>> Kconfig files do not have defaults in them.
> 
> I wouldn't be looking at defconfig files - I don't think many people use
> them.  Most people use their previous config, via oldconfig.
> 
> So what we want here is to give them E1000E if they had previously been
> using E1000.  I don't know how one would do this in Kconfig.

ditto. I doubt that "SELECT E1000E" would be a good idea here (maybe not even
work), and I can't think of anything else.

Auke