Re: kobject.c changes in -mm

[Frederik Deweerdt <deweerdt@free.fr>]

On Mon, Jan 08, 2007 at 12:31:56PM -0800, Greg KH wrote:
> On Mon, Jan 08, 2007 at 07:25:07PM +0000, Frederik Deweerdt wrote:
> > On Mon, Jan 08, 2007 at 11:09:42AM -0800, Greg KH wrote:
> > > On Mon, Jan 08, 2007 at 01:37:47PM +0000, Christoph Hellwig wrote:
> > > > --- linux-2.6.20-rc3/lib/kobject.c      2007-01-01 23:04:49.000000000 -0800
> > > > +++ devel/lib/kobject.c 2007-01-04 21:13:21.000000000 -0800
> > > > @@ -15,6 +15,8 @@
> > > >  #include <linux/module.h>
> > > >  #include <linux/stat.h>
> > > >  #include <linux/slab.h>
> > > > +#include <linux/kallsyms.h>
> > > > +#include <asm-generic/sections.h>
> > > > 
> > > > +#ifdef CONFIG_X86_32
> > > > +static int ptr_in_range(void *ptr, void *start, void *end)
> > > > +{
> > > > +       /*
> > > > +        * This should hopefully get rid of causing warnings
> > > > +        * if the architecture did not set one of the section
> > > > +        * variables up.
> > > > +        */ 
> > > > +       if (start >= end)
> > > > +               return 0;
> > > > +
> > > > +       if ((ptr >= start) && (ptr < end))
> > > > +               return 1;
> > > > +       return 0;
> > > > +}      
> > > > 
> > > > 
> > > > Can anyone explain WTF is going on here?  Including asm-generic headers
> > > > in core code definitly is not okay.  As are random CONFIG_X86_32 ifdefs
> > > > in said code.
> > > 
> > > It's a hack for debugging.  See the full patch at:
> > > 	http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/driver/warn-when-statically-allocated-kobjects-are-used.patch
> > > 
> > > It is never going to go to mainline, due to the arch-specific hacks as
> > > you have noted.  But is good to have for debugging and getting error
> > > reports from users of -mm.
> > > 
> > Could a CONFIG_{MM,HACK}  option be added for this kind of hacks? It could
> > help clarify what the aim of the code is.
> 
> How would that help here?  I don't think we want to #ifdef all patches
> in the -mm tree that are of this type, that would be a bit nasty.
I see how this would be messy, but this could help advertising the fact
that the patch is not going to mainline, if only because mainline wouldn't
have that CONFIG_HACK thing.
Another alternative would a mm-only- prefix to the name of the patches,
a simple grep in broken-out would the be enough...

Regards,
Frederik
> thanks,
> 
> greg k-h
>