On Tue, 2010-03-30 at 19:17 -0700, Yinghai Lu wrote:
> 32bit now can use generic __find_lmb_area now. So we can turn arch one off
>
> Signed-off-by: Yinghai Lu <[email protected]>
> ---
> arch/x86/Kconfig | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index 1aa53ae..ce097ab 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -584,7 +584,7 @@ config PARAVIRT_DEBUG
> a paravirt_op is missing when it is called.
>
> config ARCH_LMB_FIND_AREA
> - default y
> + default n
> bool "Use x86 own lmb_find_area()"
> ---help---
> Use lmb_find_area() version instead of generic version, it get free
So can't the whole ARCH_LMB_FIND_AREA go away then? And the generic
version be made non-weak?
cheers
On 04/07/2010 04:38 PM, Michael Ellerman wrote:
> On Tue, 2010-03-30 at 19:17 -0700, Yinghai Lu wrote:
>> 32bit now can use generic __find_lmb_area now. So we can turn arch one off
>>
>> Signed-off-by: Yinghai Lu <[email protected]>
>> ---
>> arch/x86/Kconfig | 2 +-
>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
>> index 1aa53ae..ce097ab 100644
>> --- a/arch/x86/Kconfig
>> +++ b/arch/x86/Kconfig
>> @@ -584,7 +584,7 @@ config PARAVIRT_DEBUG
>> a paravirt_op is missing when it is called.
>>
>> config ARCH_LMB_FIND_AREA
>> - default y
>> + default n
>> bool "Use x86 own lmb_find_area()"
>> ---help---
>> Use lmb_find_area() version instead of generic version, it get free
>
> So can't the whole ARCH_LMB_FIND_AREA go away then? And the generic
> version be made non-weak?
Later after more test, to make sure x86 could endure from high to low.
then we can remove ARCH_LMB_FIND_AREA.
YH