[PATCH v2] powerpc: fix compile error on 64K pages on 40x, 44x

[Christian Lamparter]

On Saturday, September 30, 2017 10:25:45 AM CEST christophe leroy wrote:
> 
> Le 28/09/2017 à 20:32, Christian Lamparter a écrit :
> > The mmu context on the 40x, 44x does not define pte_frag
> > entry. This causes gcc abort the compilation due to:
> > 
> > setup-common.c: In function ‘setup_arch’:
> > setup-common.c:908: error: ‘mm_context_t’ has no ‘pte_frag’
> > 
> > This patch fixes the issue by adding additional guard
> > conditions, that limit the initialization to just
> > platforms that define the pte_frag variable in the
> > mmu context.
> 
> Wouldn't it be better to amend mm_context_t with the following, just 
> like other platforms, instead of a #ifdef in setup_arch() ?
>
> #ifdef CONFIG_PPC_64K_PAGES
> 	/* for 4K PTE fragment support */
> 	void *pte_frag;
> #endif

This would require to add this dead weight to the mm_context_t of
asm/mmu-40x.h, asm/mmu-44x.h and asm/mmu-8xx.h. 
asm/mmu-book3e.h already has the pte_frag defined.
But it is not used by any mmu-book3e/mmu-32 code as far as I can tell.

> Otherwise, since init_mm is defined in BSS, is setting pte_frag to NULL 
> needed at all ?
> I think removing this line would be better.
Yes, in fact it's guaranteed to be NULL by C99.
This is because init_mm is of external linkage and this makes it of "static 
storage class" according to 6.2.4.3:
"An object whose identifier is declared with external or internal linkage,
or with the storage-class specifier static has static storage duration."

And C99 defines in 6.7.8.10 Initialization that: "
If an object that has static storage duration is not initialized explicitly,
then:
— if it has pointer type, it is initialized to a null pointer;
[...]
"

so yes, init_mm.context.pte_frag = NULL; can be dropped. 

The question is what to do with BOOK3E mmu.h? After all the
pte_frag serves no purpose there.

Regards,
Christian