[PATCH 5/6] arm64: Implement pmem API support

[Dan Williams]

On Fri, Aug 4, 2017 at 10:43 AM, Robin Murphy <robin.murphy at arm.com> wrote:
> On 04/08/17 16:25, Catalin Marinas wrote:
>> Two minor comments below.
>>
>> On Tue, Jul 25, 2017 at 11:55:42AM +0100, Robin Murphy wrote:
>>> --- a/arch/arm64/Kconfig
>>> +++ b/arch/arm64/Kconfig
>>> @@ -960,6 +960,17 @@ config ARM64_UAO
>>>        regular load/store instructions if the cpu does not implement the
>>>        feature.
>>>
>>> +config ARM64_PMEM
>>> +    bool "Enable support for persistent memory"
>>> +    select ARCH_HAS_PMEM_API
>>> +    help
>>> +      Say Y to enable support for the persistent memory API based on the
>>> +      ARMv8.2 DCPoP feature.
>>> +
>>> +      The feature is detected at runtime, and the kernel will use DC CVAC
>>> +      operations if DC CVAP is not supported (following the behaviour of
>>> +      DC CVAP itself if the system does not define a point of persistence).
>>
>> Any reason not to have this default y?
>
> Mostly because it's untested, and not actually useful without some way
> of describing persistent memory regions to the kernel (I'm currently
> trying to make sense of what exactly ARCH_HAS_MMIO_FLUSH is supposed to
> mean in order to enable ACPI NFIT support).

This is related to block-aperture support described by the NFIT where
a sliding-memory-mapped window can be programmed to access different
ranges of the NVDIMM. Before the window is programmed to a new
DIMM-address we need to flush any dirty data through the current
window setting to media. See the call to mmio_flush_range() in
acpi_nfit_blk_single_io(). I think it's ok to omit ARCH_HAS_MMIO_FLUSH
support, and add a configuration option to compile out the
block-aperture support.