[PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[Matthew Garrett]

EFI doesn't provide any guarantees that any given address will be free,
so the bootloader must have the freedom to position the kernel
appropriately. Make CONFIG_EFI select CONFIG_RELOCATABLE in order to
ensure that this constraint is satisfiable.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
---
 arch/x86/Kconfig |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index da34972..580cd31 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -1454,6 +1454,7 @@ config ARCH_USES_PG_UNCACHED
 config EFI
 	bool "EFI runtime service support"
 	depends on ACPI
+	select RELOCATABLE
 	---help---
 	  This enables the kernel to use EFI runtime services that are
 	  available (such as the EFI variable services).
-- 
1.7.5.2

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[Peter Jones]

On 06/09/2011 10:32 AM, Matthew Garrett wrote:
> EFI doesn't provide any guarantees that any given address will be free,
> so the bootloader must have the freedom to position the kernel
> appropriately. Make CONFIG_EFI select CONFIG_RELOCATABLE in order to
> ensure that this constraint is satisfiable.
>
> Signed-off-by: Matthew Garrett<mjg@redhat.com>

Yeah, this is a requirement by the UEFI spec.

Signed-off-by: Peter Jones <pjones@redhat.com>

> ---
>   arch/x86/Kconfig |    1 +
>   1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index da34972..580cd31 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -1454,6 +1454,7 @@ config ARCH_USES_PG_UNCACHED
>   config EFI
>   	bool "EFI runtime service support"
>   	depends on ACPI
> +	select RELOCATABLE
>   	---help---
>   	  This enables the kernel to use EFI runtime services that are
>   	  available (such as the EFI variable services).


-- 
         Peter

All parts should go together without forcing. You must remember that
the parts you are  reassembling were disassembled by you.  Therefore,
if you can't get them together again, there must be a reason. By all
means, do not use a hammer.
		-- IBM maintenance manual, 1925

01234567890123456789012345678901234567890123456789012345678901234567890123456789

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[H. Peter Anvin]

On 06/09/2011 07:32 AM, Matthew Garrett wrote:
> EFI doesn't provide any guarantees that any given address will be free,
> so the bootloader must have the freedom to position the kernel
> appropriately. Make CONFIG_EFI select CONFIG_RELOCATABLE in order to
> ensure that this constraint is satisfiable.
> 
> Signed-off-by: Matthew Garrett <mjg@redhat.com>
> ---
>  arch/x86/Kconfig |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index da34972..580cd31 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -1454,6 +1454,7 @@ config ARCH_USES_PG_UNCACHED
>  config EFI
>  	bool "EFI runtime service support"
>  	depends on ACPI
> +	select RELOCATABLE
>  	---help---
>  	  This enables the kernel to use EFI runtime services that are
>  	  available (such as the EFI variable services).

Can we make this a depends on instead of select?

	-hpa

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[Peter Jones]

On 06/09/2011 10:37 AM, H. Peter Anvin wrote:
> On 06/09/2011 07:32 AM, Matthew Garrett wrote:
>> EFI doesn't provide any guarantees that any given address will be free,
>> so the bootloader must have the freedom to position the kernel
>> appropriately. Make CONFIG_EFI select CONFIG_RELOCATABLE in order to
>> ensure that this constraint is satisfiable.
>>
>> Signed-off-by: Matthew Garrett<mjg@redhat.com>
>> ---
>>   arch/x86/Kconfig |    1 +
>>   1 files changed, 1 insertions(+), 0 deletions(-)
>>
>> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
>> index da34972..580cd31 100644
>> --- a/arch/x86/Kconfig
>> +++ b/arch/x86/Kconfig
>> @@ -1454,6 +1454,7 @@ config ARCH_USES_PG_UNCACHED
>>   config EFI
>>   	bool "EFI runtime service support"
>>   	depends on ACPI
>> +	select RELOCATABLE
>>   	---help---
>>   	  This enables the kernel to use EFI runtime services that are
>>   	  available (such as the EFI variable services).
>
> Can we make this a depends on instead of select?

Seems weird to me - when you're configuring the kernel, you're more likely
to know that you want EFI than to remember that that requires relocation.

-- 
         Peter

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[Matthew Garrett]

On Thu, Jun 09, 2011 at 07:37:31AM -0700, H. Peter Anvin wrote:
> On 06/09/2011 07:32 AM, Matthew Garrett wrote:
> >  config EFI
> >  	bool "EFI runtime service support"
> >  	depends on ACPI
> > +	select RELOCATABLE
> >  	---help---
> >  	  This enables the kernel to use EFI runtime services that are
> >  	  available (such as the EFI variable services).
> 
> Can we make this a depends on instead of select?

CONFIG_RELOCATABLE has no dependencies so select should be safe, and 
depends means CONFIG_EFI vanishes until you turn on something that seems 
entirely unrelated. I've no problem with it being depends, but this 
seems friendlier.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[H. Peter Anvin]

On 06/09/2011 07:41 AM, Peter Jones wrote:
>>
>> Can we make this a depends on instead of select?
> 
> Seems weird to me - when you're configuring the kernel, you're more likely
> to know that you want EFI than to remember that that requires relocation.
> 

Yes, but "select" is handled very poorly by Kconfig.

	-hpa

Re: [PATCH] x86: CONFIG_EFI should select CONFIG_RELOCATABLE

[Matthew Garrett]

On Thu, Jun 09, 2011 at 08:23:48AM -0700, H. Peter Anvin wrote:
> On 06/09/2011 07:41 AM, Peter Jones wrote:
> >>
> >> Can we make this a depends on instead of select?
> > 
> > Seems weird to me - when you're configuring the kernel, you're more likely
> > to know that you want EFI than to remember that that requires relocation.
> > 
> 
> Yes, but "select" is handled very poorly by Kconfig.

I think it's safe in this case.

-- 
Matthew Garrett | mjg59@srcf.ucam.org