Re: [PATCH 2/2] x86/trampoline: Rationalise the constants to describe the size

[Jan Beulich <jbeulich@suse.com>]

On 13.11.2024 12:52, Andrew Cooper wrote:
> On 13/11/2024 10:23 am, Jan Beulich wrote:
>> On 13.11.2024 10:30, Andrew Cooper wrote:
>>> --- a/xen/arch/x86/include/asm/config.h
>>> +++ b/xen/arch/x86/include/asm/config.h
>>> @@ -51,8 +51,9 @@
>>>  
>>>  #define IST_SHSTK_SIZE 1024
>>>  
>>> -#define TRAMPOLINE_STACK_SPACE  PAGE_SIZE
>>> -#define TRAMPOLINE_SPACE        (KB(64) - TRAMPOLINE_STACK_SPACE)
>>> +/* See asm/trampoline.h */
>>> +#define TRAMPOLINE_SIZE         KB(64)
>>> +#define TRAMPOLINE_HEAP_END     (TRAMPOLINE_SIZE - PAGE_SIZE)
>> Is there actually a reason these can't move to trampoline.h?
> 
> I considered that, and ruled it out, but I can't remember why exactly.
> 
> Right now, trampoline.h is C-only, but it can gain some __ASSEMBLY__
> easily enough.
> 
> 
> The two later constants, WAKEUP_STACK_MIN and MBI_SPACE_MIN are used
> only in linker assertions, and of dubious value.  In particular, the
> size of the VESA information in the heap is not accounted for in the
> MBI_SPACE_MIN check.
> 
> I have an idea to remove all of this boot metadata shuffling when the
> boot_info work is a bit better done, which is why I left the constants
> alone.

Fair enough then.

Jan