Re: [PATCH 0/2] Enable ZONE_DEVICE on POWER

On 05/08/2016 11:07 PM, Aneesh Kumar K.V wrote:

oliver [off-list ref] writes:

quoted

quoted

Hi,

I've been working on kernel support for a persistent memory (nvdimm)
device and the kernel driver infrastructure requires ZONE_DEVICE for
DAX support.. I've had it enabled in my tree for some time (without
altmap support) without any real issues.

IIUC a DAX usage of pmem doesn't really need ZONE_DEVICE. It is only
for struct page backing to those pmem area we need ZONE_DEVICE.

quoted

quoted

I wasn't planning on upstreaming any of my changes until 4.8 at the
earliest so I am ok with carrying these patches myself. However, there
has been some interest in using ZONE_DEVICE for other things on ppc
(wasn't that you?) and given that ZONE_DEVICE is gated behind
CONFIG_EXPERT I can't see there being any kind of negative impact on
end users by merging it now. At the very least it lets the rest of the
kernel development community know that changes affecting zones should
also be tested on powerpc.

A partially done patch like that will miss quiet a lot details. For
example if I look at the x86 changes related to altmap
(4b94ffdc4163bae1e ("x86, mm: introduce vmem_altmap to augment
vmemmap_populate")) i see them handling pagetable free and memory
hotplug. This patch doesn't do any of those. From the commit message it is
also not clear how we intent to use those zone device memory on ppc64.
If we say they will not get hotplugged out or they will never be part
of page table then those changes I mentioned above are really not
needed. But the patch is missing a lot of those details.

Right. The current patch just enables ZONE_DEVICE with vmem_altmap support
(so that struct pages can be allocated in the device range instead of system
RAM) where any driver can own the rest of the PFNs for its use. Right now
these PFNs will not make into process page table. I can update the commit
message with these details if you like.