On 10/03/2017 23:41, Brijesh Singh wrote:
>> Maybe there's a reason this fires:
>>
>> WARNING: modpost: Found 2 section mismatch(es).
>> To see full details build your kernel with:
>> 'make CONFIG_DEBUG_SECTION_MISMATCH=y'
>>
>> WARNING: vmlinux.o(.text+0x48edc): Section mismatch in reference from
>> the function __change_page_attr() to the function
>> .init.text:memblock_alloc()
>> The function __change_page_attr() references
>> the function __init memblock_alloc().
>> This is often because __change_page_attr lacks a __init
>> annotation or the annotation of memblock_alloc is wrong.
>>
>> WARNING: vmlinux.o(.text+0x491d1): Section mismatch in reference from
>> the function __change_page_attr() to the function
>> .meminit.text:memblock_free()
>> The function __change_page_attr() references
>> the function __meminit memblock_free().
>> This is often because __change_page_attr lacks a __meminit
>> annotation or the annotation of memblock_free is wrong.
>> 
>> But maybe Paolo might have an even better idea...
> 
> I am sure he will have better idea :)

Not sure if it's better or worse, but an alternative idea is to turn
__change_page_attr and __change_page_attr_set_clr inside out, so that:
1) the alloc_pages/__free_page happens in __change_page_attr_set_clr;
2) __change_page_attr_set_clr overall does not beocome more complex.

Then you can introduce __early_change_page_attr_set_clr and/or
early_kernel_map_pages_in_pgd, for use in your next patches.  They use
the memblock allocator instead of alloc/free_page

The attached patch is compile-tested only and almost certainly has some
thinko in it.  But it even skims a few lines from the code so the idea
might have some merit.

Paolo