Re: [HMM 12/15] mm/migrate: new memory migration helper for use with device memory v4

From	Jerome Glisse <jglisse@redhat.com>
Newsgroups	linux.kernel
Subject	Re: [HMM 12/15] mm/migrate: new memory migration helper for use with device memory v4
Date	2017-06-07 16:10 +0200
Message-ID	<tPJOP-7Fm-43@gated-at.bofh.it> (permalink)
References	<tKIgF-5fw-3@gated-at.bofh.it> <tKIgH-5fw-51@gated-at.bofh.it> <tN37k-8e3-9@gated-at.bofh.it> <tNGV4-ib-13@gated-at.bofh.it> <tPF8u-4Fk-15@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

> On Fri, Jun 2, 2017 at 8:35 AM, Jerome Glisse <jglisse@redhat.com> wrote:
> > On Wed, May 31, 2017 at 01:59:54PM +1000, Balbir Singh wrote:
> >> On Wed, 24 May 2017 13:20:21 -0400
> >> Jérôme Glisse <jglisse@redhat.com> wrote:
> >>
> >> > This patch add a new memory migration helpers, which migrate memory
> >> > backing a range of virtual address of a process to different memory
> >> > (which can be allocated through special allocator). It differs from
> >> > numa migration by working on a range of virtual address and thus by
> >> > doing migration in chunk that can be large enough to use DMA engine
> >> > or special copy offloading engine.
> >> >
> >> > Expected users are any one with heterogeneous memory where different
> >> > memory have different characteristics (latency, bandwidth, ...). As
> >> > an example IBM platform with CAPI bus can make use of this feature
> >> > to migrate between regular memory and CAPI device memory. New CPU
> >> > architecture with a pool of high performance memory not manage as
> >> > cache but presented as regular memory (while being faster and with
> >> > lower latency than DDR) will also be prime user of this patch.
> >> >
> >> > Migration to private device memory will be useful for device that
> >> > have large pool of such like GPU, NVidia plans to use HMM for that.
> >> >
> >>
> >> It is helpful, for HMM-CDM however we would like to avoid the downsides
> >> of MIGRATE_SYNC_NOCOPY
> >
> > What are the downside you are referring too ?
> 
> IIUC, MIGRATE_SYNC_NO_COPY is for anonymous memory only.

It can migrate anything, file back page too. It just forbid that latter
case if it is ZONE_DEVICE HMM. I should have time now to finish the CDM
patchset and i will post, previous patches already enabled file back
page migration for HMM-CDM.

The NOCOPY is for no CPUCOPY, i couldn't think of a better name.

Cheers,
Jérôme

Back to linux.kernel | Previous | Next — Previous in thread | Find similar | Unroll thread

Thread

Re: [HMM 12/15] mm/migrate: new memory migration helper for use  with device memory v4 Balbir Singh <bsingharora@gmail.com> - 2017-05-31 06:10 +0200
  Re: [HMM 12/15] mm/migrate: new memory migration helper for use with  device memory v4 Jerome Glisse <jglisse@redhat.com> - 2017-06-02 00:40 +0200
    Re: [HMM 12/15] mm/migrate: new memory migration helper for use with  device memory v4 Balbir Singh <bsingharora@gmail.com> - 2017-06-07 11:10 +0200
      Re: [HMM 12/15] mm/migrate: new memory migration helper for use  with device memory v4 Jerome Glisse <jglisse@redhat.com> - 2017-06-07 16:10 +0200

csiph-web