Re: [PATCH v2 05/41] acpi : add helper function for mapping memory

[Parth Dixit <parth.dixit@linaro.org>]

+shannon

On 24 May 2015 at 13:01, Julien Grall <julien.grall@citrix.com> wrote:
> Hi Parth,
>
> On 24/05/2015 07:40, Parth Dixit wrote:
>>
>>     On 17/05/15 21:03, Parth Dixit wrote:
>>      > diff --git a/xen/arch/arm/Makefile b/xen/arch/arm/Makefile
>>      > index 935999e..096e9ef 100644
>>      > --- a/xen/arch/arm/Makefile
>>      > +++ b/xen/arch/arm/Makefile
>>      > @@ -2,6 +2,7 @@ subdir-$(arm32) += arm32
>>      >  subdir-$(arm64) += arm64
>>      >  subdir-y += platforms
>>      >  subdir-$(arm64) += efi
>>      > +subdir-$(HAS_ACPI) += acpi
>>      >
>>      >  obj-$(EARLY_PRINTK) += early_printk.o
>>      >  obj-y += cpu.o
>>      > diff --git a/xen/arch/arm/acpi/Makefile
>> b/xen/arch/arm/acpi/Makefile
>>      > new file mode 100644
>>      > index 0000000..b5be22d
>>      > --- /dev/null
>>      > +++ b/xen/arch/arm/acpi/Makefile
>>      > @@ -0,0 +1 @@
>>      > +obj-y += lib.o
>>      > diff --git a/xen/arch/arm/acpi/lib.c b/xen/arch/arm/acpi/lib.c
>>      > new file mode 100644
>>      > index 0000000..650beed
>>      > --- /dev/null
>>      > +++ b/xen/arch/arm/acpi/lib.c
>>      > @@ -0,0 +1,8 @@
>>      > +#include <xen/acpi.h>
>>      > +#include <asm/mm.h>
>>      > +
>>      > +void __iomem *
>>      > +acpi_os_map_iomem(acpi_physical_address phys, acpi_size size)
>>      > +{
>>      > +    return __va(phys);
>>      > +}
>>
>>     I would have prefer two distinct patch: one for the refactoring of
>>     acpi_os_map_memory and the other for implementing the ARM part
>>     explaining why only using __va.
>>
>>     __va should only be used when the memory is direct-mapped to Xen (i.e
>>     accessible directly). On ARM64, this only the case for the RAM. Can
>> you
>>     confirm that ACPI will always reside to the RAM?
>>
>>     I already asked the same question on the previous version but got no
>>     answer from you...
>>
>> I did not found any document which says it will always reside in RAM or
>> otherwise..
>
>
> If so, you have use vmap or ioremap_cache as suggested by Jan.
>
> Regards,
>
> --
> Julien Grall