Re: [PATCH 4/4] arch, mm: make kernel_page_present() always available

[Mike Rapoport]

On Mon, Oct 26, 2020 at 12:54:01AM +0000, Edgecombe, Rick P wrote:
> On Sun, 2020-10-25 at 12:15 +0200, Mike Rapoport wrote:
> > index 7f248fc45317..16f878c26667 100644
> > --- a/arch/x86/mm/pat/set_memory.c
> > +++ b/arch/x86/mm/pat/set_memory.c
> > @@ -2228,7 +2228,6 @@ void __kernel_map_pages(struct page *page, int
> > numpages, int enable)
> >  }
> >  #endif /* CONFIG_DEBUG_PAGEALLOC */
> >  
> > -#ifdef CONFIG_HIBERNATION
> >  bool kernel_page_present(struct page *page)
> >  {
> >         unsigned int level;
> > @@ -2240,7 +2239,6 @@ bool kernel_page_present(struct page *page)
> >         pte = lookup_address((unsigned long)page_address(page),
> > &level);
> >         return (pte_val(*pte) & _PAGE_PRESENT);
> >  }
> > -#endif /* CONFIG_HIBERNATION */
> 
> This is only used by hibernate today right? Makes sense that it should
> return a correct answer if someone starts to use it without looking too
> closely at the header. But could we just remove the default static
> inline return true implementation and let the linker fail if someone
> starts to use it outside hibernate? Then we could leave it compiled out
> until then.

Hmm, I'm not sure I follow you here.
We'd need some stub for architectures that have
ARCH_HIBERNATION_POSSIBLE and do not have
CONFIG_ARCH_HAS_SET_DIRECT_MAP.

I don't see how the kernel would compile for ppc or sparc with
hibernation enabled if I remove the default implementation.

> Also it looks like riscv does not have ARCH_HIBERNATION_POSSIBLE so the
> new function added here couldn't be used yet. You could also just let
> the linker catch it if riscv ever enables hibernate?

-- 
Sincerely yours,
Mike.