Re: [PATCH] fix ia64 breakage with PHYSDEVOP_pirq_eoi_mfn (was Re: [PATCH 2/2] linux/x86: use shared page indicating the need for an EOI notification)

["Jan Beulich" <jbeulich@novell.com>]

>>> Isaku Yamahata <yamahata@valinux.co.jp> 10.12.08 05:09 >>>
>Like this?

Almost.

>--- a/include/asm-i386/mach-xen/asm/pgtable.h
>+++ b/include/asm-i386/mach-xen/asm/pgtable.h
>@@ -488,6 +488,17 @@ void make_pages_writable(void *va, unsig
> 	(((maddr_t)pte_mfn(*virt_to_ptep(va)) << PAGE_SHIFT)		\
> 	 | ((unsigned long)(va) & (PAGE_SIZE - 1)))
> 
>+#ifdef CONFIG_HIGHPTE
>+#define ptep_to_machine(ptep)						\
>+({									\
>+	(unsigned long)(ptep) >= (unsigned long)high_memory?		\
>+		arbitrary_virt_to_machine(ptep) :			\
>+		virt_to_machine(ptep);					\
>+})

I was intending to make use of kmap_atomic_to_page() here. Due to the
check at the beginning of that function, it would even be possible to get
away without a conditional in the highpte case of the macro:

#define ptep_to_machine(ptep) page_to_phys(kmap_atomic_to_page(ptep))

page_to_phys() and page_to_bus() could be used here interchangeably due
to them being identical on x86.

Jan