Re: [PATCH] RCU torture-testing kernel module

[Ingo Oeser <ioe-lkml@rameria.de>]

[-- Attachment #1: Type: text/plain, Size: 1284 bytes --]

Hi Paul,

On Sunday 23 October 2005 16:36, Paul E. McKenney wrote:
> On Sun, Oct 23, 2005 at 09:22:18AM +0200, Ingo Oeser wrote:
> > On Sunday 23 October 2005 01:12, Paul E. McKenney wrote:
> > > +config RCU_TORTURE_TEST
> > > +	tristate "torture tests for RCU"
> > > +	default n
> > 
> > Please put this into lib/Kconfig.debug and make it dependent on
> > DEBUG_KERNEL there, which illustrates its actual purpose much better.
> 
> If I make it depend on DEBUG_KERNEL, I get other extraneous debug code
> as well, for example, in verify_mm_writelocked(). 

Oh, this seems to be either not intended or misguided.

DEBUG_KERNEL should do nothing more than showing the debugging
options. 

E.g. I don't expect to enable any additional code in an 
unrelated file, if I enable Magic-SysRQ on an embedded, unattended device
to be able to analyze potential problems via serial console.

@Andrew: Would you accept a patch to fix that?

My idea to put the option into lib/Kconfig.debug was just to ease 
maintainence, once it is dependend on DEBUG_KERNEL.

On the other hand it makes sense to add another option TEST_KERNEL
and put some existing kernel internal test suites under that.

@Andrew: Would you accept a patch for this?


Regards

Ingo Oeser


[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]