Re: [PATCH] KGDB: refactor kconfig menu

[Jan Kiszka]

Jason Wessel wrote:
> Jan Kiszka wrote:
>> This is an attempt to improve the so far, well, unfortunate Kconfig menu
>> organization of KGDB. Most notably, it pushes all sub-entries into their
>> own menuconfig, removes the (IMHO) unneeded "Method for KGDB
>> communication" choice, and ensures everything is indented properly. This
>> should keep the original semantic while reducing the number of knobs the
>> user is confronted with.
>>
>>   
> 
> While it is probably reasonable to put the kgdb options into a sub
> entry,  you cannot get rid of the "Method for KGDB communication"
> choice, unless you re-implement it another way.
> 
> It is designed to implement the following:
> 1) You can chose to build 1 and only 1 KGDB I/O module into the kernel
> 2) Or you select modules only and can you can build as many additional
>     KGDB I/O modules as you like.
> 3) Additional I/O modules can be built and you can load one at run time
>     if you want to use another I/O method vs #1

I see no reason why these cases cannot be enforced later on (when there
are >1 I/O modules) via Kconfig depends-on rules. But maybe I'm
overseeing something, please correct me.

> 
> KGDB's Kconfig options are designed to allow you to have alternate ways
> to select static or run time configuration of the I/O connection to the
> KGDB core, IE: over serial, over ethernet, over USB or what ever module
> you would like to implement.

Well, I'm looking at this from the user perspective now, and as a user I
don't understand why I shouldn't be able to configure my setup via the
tristates of the involved I/O drivers (+ some underlying dependencies).

Jan

Attachment: signature.asc
Description: OpenPGP digital signature