kdb
[Top] [All Lists]

Re: [PATCH] Compiling v2.0-2.4.17-i386-1 with CONFIG_KDB=n

To: Tommi Virtanen <tv@xxxxxxxxxx>
Subject: Re: [PATCH] Compiling v2.0-2.4.17-i386-1 with CONFIG_KDB=n
From: Keith Owens <kaos@xxxxxxx>
Date: Sun, 30 Dec 2001 19:45:09 +1100
Cc: kdb@xxxxxxxxxxx
In-reply-to: Your message of "30 Dec 2001 10:35:53 +0200." <874rm9jeae.fsf@xxxxxxxxxxxxx>
Sender: owner-kdb@xxxxxxxxxxx
On 30 Dec 2001 10:35:53 +0200, 
Tommi Virtanen <tv@xxxxxxxxxx> wrote:
>Keith Owens <kaos@xxxxxxx> writes:
>
>> Other way around.  Calls to kdb from non-kdb code need to be wrapped in
>> #ifdef CONFIG_KDB, I missed one.
>
>       Then why do you, in include/linux/kdb.h, do
>
>#ifdef CONFIG_KDB
>extern int kdb(....);
>#else
>#define kdb(....) (0)
>#endif
>
>        then?

Old code.

>       I understand my suggestion is the approach Linus likes..

The kdb patch is included with other software such as xfs.  xfs runs on
more systems than kdb does and the kdb bits fail to compile on some
systems where xfs will run on its own.  I need to ensure that no
vestige of kdb is visible unless CONFIG_KDB=y.

Also a lot of the kdb patch is not just a function call, it is complete
replacements for existing code, you cannot do that without #ifdef.  I
prefer a consistent coding style, since I have to use #ifdef in some
places, I use it everywhere.


<Prev in Thread] Current Thread [Next in Thread>