Re: [patch 4/4] increase traffic on linux-kernel

[Andrew Morton]

Jeff Garzik wrote:
> 
> Andrew Morton wrote:
> > --- 2.5.38/include/linux/kernel.h~might_sleep Wed Sep 25 20:15:27 2002
> > +++ 2.5.38-akpm/include/linux/kernel.h        Wed Sep 25 20:15:27 2002
> > @@ -40,6 +40,13 @@
> >
> >  struct completion;
> >
> > +#ifdef CONFIG_DEBUG_KERNEL
> > +void __might_sleep(char *file, int line);
> > +#define might_sleep() __might_sleep(__FILE__, __LINE__)
> > +#else
> > +#define might_sleep() do {} while(0)
> > +#endif
> 
> I disagree with this -- CONFIG_DEBUG_KERNEL should not enable any machinery.

I did this because adding a function call to down() and friends
has significant cost.

> Magic Sysrq should be enable-able without affecting any other parts of
> the kernel.

A separate config option then?