From: Wei Xu <weixugc@google.com>
To: "ying.huang@intel.com" <ying.huang@intel.com>
Cc: Alistair Popple <apopple@nvidia.com>,
Yang Shi <shy828301@gmail.com>,
Aneesh Kumar K V <aneesh.kumar@linux.ibm.com>,
Jagdish Gediya <jvgediya@linux.ibm.com>,
Dave Hansen <dave.hansen@linux.intel.com>,
Dan Williams <dan.j.williams@intel.com>,
Davidlohr Bueso <dave@stgolabs.net>,
Linux MM <linux-mm@kvack.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Andrew Morton <akpm@linux-foundation.org>,
Baolin Wang <baolin.wang@linux.alibaba.com>,
Greg Thelen <gthelen@google.com>, MichalHocko <mhocko@kernel.org>,
Brice Goglin <brice.goglin@gmail.com>,
Feng Tang <feng.tang@intel.com>
Subject: Re: [PATCH v2 0/5] mm: demotion: Introduce new node state N_DEMOTION_TARGETS
Date: Thu, 28 Apr 2022 19:58:56 -0700 [thread overview]
Message-ID: <CAAPL-u8cdpi6r3wM6UN9qT22-xyd6J0-a_8Kgmdc7mT1smAq-A@mail.gmail.com> (raw)
In-Reply-To: <f6282d421ee66b56ae889d408ca5bbe843bd9494.camel@intel.com>
On Thu, Apr 28, 2022 at 7:21 PM ying.huang@intel.com
<ying.huang@intel.com> wrote:
>
> On Fri, 2022-04-29 at 11:27 +1000, Alistair Popple wrote:
> > On Friday, 29 April 2022 3:14:29 AM AEST Yang Shi wrote:
> > > On Wed, Apr 27, 2022 at 9:11 PM Wei Xu <weixugc@google.com> wrote:
> > > >
> > > > On Wed, Apr 27, 2022 at 5:56 PM ying.huang@intel.com
> > > > <ying.huang@intel.com> wrote:
> > > > >
> > > > > On Wed, 2022-04-27 at 11:27 -0700, Wei Xu wrote:
> > > > > > On Tue, Apr 26, 2022 at 10:06 PM Aneesh Kumar K V
> > > > > > <aneesh.kumar@linux.ibm.com> wrote:
> > > > > > >
> > > > > > > On 4/25/22 10:26 PM, Wei Xu wrote:
> > > > > > > > On Sat, Apr 23, 2022 at 8:02 PM ying.huang@intel.com
> > > > > > > > <ying.huang@intel.com> wrote:
> > > > > > > > >
> > > > > > >
> > > > > > > ....
> > > > > > >
> > > > > > > > > 2. For machines with PMEM installed in only 1 of 2 sockets, for example,
> > > > > > > > >
> > > > > > > > > Node 0 & 2 are cpu + dram nodes and node 1 are slow
> > > > > > > > > memory node near node 0,
> > > > > > > > >
> > > > > > > > > available: 3 nodes (0-2)
> > > > > > > > > node 0 cpus: 0 1
> > > > > > > > > node 0 size: n MB
> > > > > > > > > node 0 free: n MB
> > > > > > > > > node 1 cpus:
> > > > > > > > > node 1 size: n MB
> > > > > > > > > node 1 free: n MB
> > > > > > > > > node 2 cpus: 2 3
> > > > > > > > > node 2 size: n MB
> > > > > > > > > node 2 free: n MB
> > > > > > > > > node distances:
> > > > > > > > > node 0 1 2
> > > > > > > > > 0: 10 40 20
> > > > > > > > > 1: 40 10 80
> > > > > > > > > 2: 20 80 10
> > > > > > > > >
> > > > > > > > > We have 2 choices,
> > > > > > > > >
> > > > > > > > > a)
> > > > > > > > > node demotion targets
> > > > > > > > > 0 1
> > > > > > > > > 2 1
> > > > > > > > >
> > > > > > > > > b)
> > > > > > > > > node demotion targets
> > > > > > > > > 0 1
> > > > > > > > > 2 X
> > > > > > > > >
> > > > > > > > > a) is good to take advantage of PMEM. b) is good to reduce cross-socket
> > > > > > > > > traffic. Both are OK as defualt configuration. But some users may
> > > > > > > > > prefer the other one. So we need a user space ABI to override the
> > > > > > > > > default configuration.
> > > > > > > >
> > > > > > > > I think 2(a) should be the system-wide configuration and 2(b) can be
> > > > > > > > achieved with NUMA mempolicy (which needs to be added to demotion).
> > > > > > > >
> > > > > > > > In general, we can view the demotion order in a way similar to
> > > > > > > > allocation fallback order (after all, if we don't demote or demotion
> > > > > > > > lags behind, the allocations will go to these demotion target nodes
> > > > > > > > according to the allocation fallback order anyway). If we initialize
> > > > > > > > the demotion order in that way (i.e. every node can demote to any node
> > > > > > > > in the next tier, and the priority of the target nodes is sorted for
> > > > > > > > each source node), we don't need per-node demotion order override from
> > > > > > > > the userspace. What we need is to specify what nodes should be in
> > > > > > > > each tier and support NUMA mempolicy in demotion.
> > > > > > > >
> > > > > > >
> > > > > > > I have been wondering how we would handle this. For ex: If an
> > > > > > > application has specified an MPOL_BIND policy and restricted the
> > > > > > > allocation to be from Node0 and Node1, should we demote pages allocated
> > > > > > > by that application
> > > > > > > to Node10? The other alternative for that demotion is swapping. So from
> > > > > > > the page point of view, we either demote to a slow memory or pageout to
> > > > > > > swap. But then if we demote we are also breaking the MPOL_BIND rule.
> > > > > >
> > > > > > IMHO, the MPOL_BIND policy should be respected and demotion should be
> > > > > > skipped in such cases. Such MPOL_BIND policies can be an important
> > > > > > tool for applications to override and control their memory placement
> > > > > > when transparent memory tiering is enabled. If the application
> > > > > > doesn't want swapping, there are other ways to achieve that (e.g.
> > > > > > mlock, disabling swap globally, setting memcg parameters, etc).
> > > > > >
> > > > > >
> > > > > > > The above says we would need some kind of mem policy interaction, but
> > > > > > > what I am not sure about is how to find the memory policy in the
> > > > > > > demotion path.
> > > > > >
> > > > > > This is indeed an important and challenging problem. One possible
> > > > > > approach is to retrieve the allowed demotion nodemask from
> > > > > > page_referenced() similar to vm_flags.
> > > > >
> > > > > This works for mempolicy in struct vm_area_struct, but not for that in
> > > > > struct task_struct. Mutiple threads in a process may have different
> > > > > mempolicy.
> > > >
> > > > From vm_area_struct, we can get to mm_struct and then to the owner
> > > > task_struct, which has the process mempolicy.
> > > >
> > > > It is indeed a problem when a page is shared by different threads or
> > > > different processes that have different thread default mempolicy
> > > > values.
> > >
> > > Sorry for chiming in late, this is a known issue when we were working
> > > on demotion. Yes, it is hard to handle the shared pages and multi
> > > threads since mempolicy is applied to each thread so each thread may
> > > have different mempolicy. And I don't think this case is rare. And not
> > > only mempolicy but also may cpuset settings cause the similar problem,
> > > different threads may have different cpuset settings for cgroupv1.
> > >
> > > If this is really a problem for real life workloads, we may consider
> > > tackling it for exclusively owned pages first. Thanks to David's
> > > patches, now we have dedicated flags to tell exclusively owned pages.
> >
> > One of the problems with demotion when I last looked is it does almost exactly
> > the opposite of what we want on systems like POWER9 where GPU memory is a
> > CPU-less memory node.
> >
> > On those systems users tend to use MPOL_BIND or MPOL_PREFERRED to allocate
> > memory on the GPU node. Under memory pressure demotion should migrate GPU
> > allocations to the CPU node and finally other slow memory nodes or swap.
> >
> > Currently though demotion considers the GPU node slow memory (because it is
> > CPU-less) so will demote CPU memory to GPU memory which is a limited resource.
> > And when trying to allocate GPU memory with MPOL_BIND/PREFERRED it will swap
> > everything to disk rather than demote to CPU memory (which would be preferred).
> >
> > I'm still looking at this series but as I understand it it will help somewhat
> > because we could make GPU memory the top-tier so nothing gets demoted to it.
>
> Yes. If we have a way to put GPU memory in top-tier (tier 0) and
> CPU+DRAM in tier 1. Your requirement can be satisfied. One way is to
> override the auto-generated demotion order via some user space tool.
> Another way is to change the GPU driver (I guess where the GPU memory is
> enumerated and onlined?) to change the tier of GPU memory node.
>
> > However I wouldn't want to see demotion skipped entirely when a memory policy
> > such as MPOL_BIND is specified. For example most memory on a GPU node will have
> > some kind of policy specified and IMHO it would be better to demote to another
> > node in the mempolicy nodemask rather than going straight to swap, particularly
> > as GPU memory capacity tends to be limited in comparison to CPU memory
> > capacity.
> > >
>
> Can you use MPOL_PREFERRED? Even if we enforce MPOL_BIND as much as
> possible, we will not stop demoting from GPU to DRAM with
> MPOL_PREFERRED. And in addition to demotion, allocation fallbacking can
> be used too to avoid allocation latency caused by demotion.
I expect that MPOL_BIND can be used to either prevent demotion or
select a particular demotion node/nodemask. It all depends on the
mempolicy nodemask specified by MPOL_BIND.
> This is another example of a system with 3 tiers if PMEM is installed in
> this machine too.
>
> Best Regards,
> Huang, Ying
>
> > > > On the other hand, it can already support most interesting use cases
> > > > for demotion (e.g. selecting the demotion node, mbind to prevent
> > > > demotion) by respecting cpuset and vma mempolicies.
> > > >
> > > > > Best Regards,
> > > > > Huang, Ying
> > > > >
> > > > > > >
> > > > > > > > Cross-socket demotion should not be too big a problem in practice
> > > > > > > > because we can optimize the code to do the demotion from the local CPU
> > > > > > > > node (i.e. local writes to the target node and remote read from the
> > > > > > > > source node). The bigger issue is cross-socket memory access onto the
> > > > > > > > demoted pages from the applications, which is why NUMA mempolicy is
> > > > > > > > important here.
> > > > > > > >
> > > > > > > >
> > > > > > > -aneesh
> > > > >
> > > > >
> > >
> > >
> >
> >
> >
> >
>
>
next prev parent reply other threads:[~2022-04-29 2:59 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-13 9:22 [PATCH v2 0/5] mm: demotion: Introduce new node state N_DEMOTION_TARGETS Jagdish Gediya
2022-04-13 9:22 ` [PATCH v2 1/5] mm: demotion: Set demotion list differently Jagdish Gediya
2022-04-14 7:09 ` ying.huang
2022-04-14 8:48 ` Jagdish Gediya
2022-04-14 8:57 ` ying.huang
2022-04-14 8:55 ` Baolin Wang
2022-04-14 9:02 ` Jonathan Cameron
2022-04-14 10:40 ` Jagdish Gediya
2022-04-21 6:13 ` ying.huang
2022-04-13 9:22 ` [PATCH v2 2/5] mm: demotion: Add new node state N_DEMOTION_TARGETS Jagdish Gediya
2022-04-21 4:33 ` Wei Xu
2022-04-13 9:22 ` [PATCH v2 3/5] mm: demotion: Add support to set targets from userspace Jagdish Gediya
2022-04-21 4:26 ` Wei Xu
2022-04-22 9:13 ` Jagdish Gediya
2022-04-21 5:31 ` Wei Xu
2022-04-13 9:22 ` [PATCH v2 4/5] device-dax/kmem: Set node state as N_DEMOTION_TARGETS Jagdish Gediya
2022-04-13 9:22 ` [PATCH v2 5/5] mm: demotion: Build demotion list based on N_DEMOTION_TARGETS Jagdish Gediya
2022-04-13 21:44 ` [PATCH v2 0/5] mm: demotion: Introduce new node state N_DEMOTION_TARGETS Andrew Morton
2022-04-14 10:16 ` Jagdish Gediya
2022-04-14 7:00 ` ying.huang
2022-04-14 10:19 ` Jagdish Gediya
2022-04-21 3:11 ` Yang Shi
2022-04-21 5:41 ` Wei Xu
2022-04-21 6:24 ` ying.huang
2022-04-21 6:49 ` Wei Xu
2022-04-21 7:08 ` ying.huang
2022-04-21 7:29 ` Wei Xu
2022-04-21 7:45 ` ying.huang
2022-04-21 18:26 ` Wei Xu
2022-04-22 0:58 ` ying.huang
2022-04-22 4:46 ` Wei Xu
2022-04-22 5:40 ` ying.huang
2022-04-22 6:13 ` Wei Xu
2022-04-22 6:21 ` ying.huang
2022-04-22 11:00 ` Jagdish Gediya
2022-04-22 16:43 ` Wei Xu
2022-04-22 17:29 ` Yang Shi
2022-04-24 3:02 ` ying.huang
2022-04-25 3:50 ` Aneesh Kumar K.V
2022-04-25 6:10 ` ying.huang
2022-04-25 8:09 ` Aneesh Kumar K V
2022-04-25 8:54 ` Aneesh Kumar K V
2022-04-25 20:17 ` Davidlohr Bueso
2022-04-26 8:42 ` ying.huang
2022-04-26 9:02 ` Aneesh Kumar K V
2022-04-26 9:44 ` ying.huang
2022-04-27 4:27 ` Wei Xu
2022-04-25 7:26 ` Jagdish Gediya
2022-04-25 16:56 ` Wei Xu
2022-04-27 5:06 ` Aneesh Kumar K V
2022-04-27 18:27 ` Wei Xu
2022-04-28 0:56 ` ying.huang
2022-04-28 4:11 ` Wei Xu
2022-04-28 17:14 ` Yang Shi
2022-04-29 1:27 ` Alistair Popple
2022-04-29 2:21 ` ying.huang
2022-04-29 2:58 ` Wei Xu [this message]
2022-04-29 3:27 ` ying.huang
2022-04-29 4:45 ` Alistair Popple
2022-04-29 18:53 ` Yang Shi
2022-04-29 18:52 ` Yang Shi
2022-04-27 7:11 ` ying.huang
2022-04-27 16:27 ` Wei Xu
2022-04-28 8:37 ` ying.huang
[not found] ` <DM6PR11MB4107867291AFE0C210D9052ADCFD9@DM6PR11MB4107.namprd11.prod.outlook.com>
2022-04-30 2:21 ` Wei Xu
2022-04-21 17:56 ` Yang Shi
2022-04-21 23:48 ` ying.huang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAAPL-u8cdpi6r3wM6UN9qT22-xyd6J0-a_8Kgmdc7mT1smAq-A@mail.gmail.com \
--to=weixugc@google.com \
--cc=akpm@linux-foundation.org \
--cc=aneesh.kumar@linux.ibm.com \
--cc=apopple@nvidia.com \
--cc=baolin.wang@linux.alibaba.com \
--cc=brice.goglin@gmail.com \
--cc=dan.j.williams@intel.com \
--cc=dave.hansen@linux.intel.com \
--cc=dave@stgolabs.net \
--cc=feng.tang@intel.com \
--cc=gthelen@google.com \
--cc=jvgediya@linux.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=shy828301@gmail.com \
--cc=ying.huang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).