Re: mark irqbalance not-recommended

[Adrian Bunk <bunk@kernel.org>]

On Tue, Oct 23, 2007 at 01:57:09PM +0200, Pavel Machek wrote:
> 
> Mark CONFIG_IRQBALANCE as not recommended any more. It eats too much
> power, and better solutions are available in userspace.

What about additionally a printk() during init and a 
feature-removal-schedule.txt entry scheduling removal
one year from now?

> Signed-off-by: Pavel Machek <pavel@suse.cz>
> 
> ---
> commit 2665f7971b5b97830b028308553a78a95c131988
> tree e1782cceebfb5900373469c0043d113de88f861f
> parent f5085f6ea1c5771708c44a4a42ee217e760ff4a0
> author Pavel <pavel@amd.ucw.cz> Tue, 23 Oct 2007 13:56:15 +0200
> committer Pavel <pavel@amd.ucw.cz> Tue, 23 Oct 2007 13:56:15 +0200
> 
>  arch/i386/Kconfig |    9 ++++++---
>  1 files changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/i386/Kconfig b/arch/i386/Kconfig
> index eba2d88..10f3c38 100644
> --- a/arch/i386/Kconfig
> +++ b/arch/i386/Kconfig
> @@ -779,10 +779,13 @@ config EFI
>  config IRQBALANCE
>   	bool "Enable kernel irq balancing"

If you want people to notice it's deprecated you should add a
" (DEPRECATED)" to the kconfig prompt.

>  	depends on SMP && X86_IO_APIC
> -	default y
> +	default n

Please drop the "default n", it has no effect.

>  	help
> - 	  The default yes will allow the kernel to do irq load balancing.
> -	  Saying no will keep the kernel from doing irq load balancing.
> + 	  Saying yes will allow the kernel to do irq load balancing;
> +	  but it generates too many CPU wakeups, eating power, and
> +	  is better done in userspace.
> +
> +	  Say N.
>  
>  # turning this on wastes a bunch of space.
>  # Summit needs it only when NUMA is on

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed