Re: [PATCH 2/4] x86/boot: replace __PHYSICAL_START with LOAD_PHYSICAL_ADDR

[Wei Yang]

On Wed, Mar 13, 2024 at 11:29:33AM +0100, Ingo Molnar wrote:
>
>* Wei Yang <richard.weiyang@xxxxxxxxx> wrote:
>
>> Both __PHYSICAL_START and LOAD_PHYSICAL_ADDR are defined to get aligned
>> CONFIG_PHYSICAL_START, so we can replace __PHYSICAL_START with
>> LOAD_PHYSICAL_ADDR. And then remove the definition of __PHYSICAL_START,
>> which is only used to define __START_KERNEL.
>> 
>> Since <asm/boot.h> includes <asm/pgtable_types.h>, which includes
>> <asm/page_types.h>, it is fine to move definition from <asm/boot.h> to
>> <asm/page_types.h>.
>> 
>> Signed-off-by: Wei Yang <richard.weiyang@xxxxxxxxx>
>> ---
>>  arch/x86/include/asm/boot.h       | 5 -----
>>  arch/x86/include/asm/page_types.h | 8 +++++---
>>  2 files changed, 5 insertions(+), 8 deletions(-)
>> 
>> diff --git a/arch/x86/include/asm/boot.h b/arch/x86/include/asm/boot.h
>> index a38cc0afc90a..12cbc57d0128 100644
>> --- a/arch/x86/include/asm/boot.h
>> +++ b/arch/x86/include/asm/boot.h
>> @@ -6,11 +6,6 @@
>>  #include <asm/pgtable_types.h>
>>  #include <uapi/asm/boot.h>
>>  
>> -/* Physical address where kernel should be loaded. */
>> -#define LOAD_PHYSICAL_ADDR ((CONFIG_PHYSICAL_START \
>> -				+ (CONFIG_PHYSICAL_ALIGN - 1)) \
>> -				& ~(CONFIG_PHYSICAL_ALIGN - 1))
>> -
>>  /* Minimum kernel alignment, as a power of two */
>>  #ifdef CONFIG_X86_64
>>  # define MIN_KERNEL_ALIGN_LG2	PMD_SHIFT
>> diff --git a/arch/x86/include/asm/page_types.h b/arch/x86/include/asm/page_types.h
>> index 86bd4311daf8..acc1620fd121 100644
>> --- a/arch/x86/include/asm/page_types.h
>> +++ b/arch/x86/include/asm/page_types.h
>> @@ -31,10 +31,12 @@
>>  
>>  #define VM_DATA_DEFAULT_FLAGS	VM_DATA_FLAGS_TSK_EXEC
>>  
>> -#define __PHYSICAL_START	ALIGN(CONFIG_PHYSICAL_START, \
>> -				      CONFIG_PHYSICAL_ALIGN)
>> +/* Physical address where kernel should be loaded. */
>> +#define LOAD_PHYSICAL_ADDR ((CONFIG_PHYSICAL_START \
>> +				+ (CONFIG_PHYSICAL_ALIGN - 1)) \
>> +				& ~(CONFIG_PHYSICAL_ALIGN - 1))
>
>I agree with this simplification, but the ALIGN() expression is far easier 
>to read, so please keep that one instead of the open-coded version.

Sure, will send v2.

>
>Thanks,
>
>	Ingo

-- 
Wei Yang
Help you, Help me