From: Alexander Duyck <email@example.com> To: Michal Hocko <firstname.lastname@example.org> Cc: "Pasha Tatashin" <email@example.com>, linux-nvdimm <firstname.lastname@example.org>, "Dave Hansen" <email@example.com>, "Linux Kernel Mailing List" <firstname.lastname@example.org>, "Ingo Molnar" <email@example.com>, "Linux MM" <firstname.lastname@example.org>, "Jérôme Glisse" <email@example.com>, firstname.lastname@example.org, "Andrew Morton" <email@example.com>, "Kirill A. Shutemov" <firstname.lastname@example.org> Subject: Re: [PATCH v5 4/4] mm: Defer ZONE_DEVICE page initialization to the point where we init pgmap Date: Mon, 29 Oct 2018 08:59:34 -0700 [thread overview] Message-ID: <email@example.com> (raw) In-Reply-To: <20181029141210.GJ32673@dhcp22.suse.cz> On Mon, 2018-10-29 at 15:12 +0100, Michal Hocko wrote: > On Wed 17-10-18 08:02:20, Alexander Duyck wrote: > > On 10/17/2018 12:52 AM, Michal Hocko wrote: > > > On Thu 11-10-18 10:38:39, Alexander Duyck wrote: > > > > On 10/11/2018 1:55 AM, Michal Hocko wrote: > > > > > On Wed 10-10-18 20:52:42, Michal Hocko wrote: > > > > > [...] > > > > > > My recollection was that we do clear the reserved bit in > > > > > > move_pfn_range_to_zone and we indeed do in __init_single_page. But then > > > > > > we set the bit back right afterwards. This seems to be the case since > > > > > > d0dc12e86b319 which reorganized the code. I have to study this some more > > > > > > obviously. > > > > > > > > > > so my recollection was wrong and d0dc12e86b319 hasn't really changed > > > > > much because __init_single_page wouldn't zero out the struct page for > > > > > the hotplug contex. A comment in move_pfn_range_to_zone explains that we > > > > > want the reserved bit because pfn walkers already do see the pfn range > > > > > and the page is not fully associated with the zone until it is onlined. > > > > > > > > > > I am thinking that we might be overzealous here. With the full state > > > > > initialized we shouldn't actually care. pfn_to_online_page should return > > > > > NULL regardless of the reserved bit and normal pfn walkers shouldn't > > > > > touch pages they do not recognize and a plain page with ref. count 1 > > > > > doesn't tell much to anybody. So I _suspect_ that we can simply drop the > > > > > reserved bit setting here. > > > > > > > > So this has me a bit hesitant to want to just drop the bit entirely. If > > > > nothing else I think I may wan to make that a patch onto itself so that if > > > > we aren't going to set it we just drop it there. That way if it does cause > > > > issues we can bisect it to that patch and pinpoint the cause. > > > > > > Yes a patch on its own make sense for bisectability. > > > > For now I think I am going to back off of this. There is a bunch of other > > changes that need to happen in order for us to make this work. As far as I > > can tell there are several places that are relying on this reserved bit. > > Please be more specific. Unless I misremember, I have added this > PageReserved just to be sure (f1dd2cd13c4bb) because pages where just > half initialized back then. I am not aware anybody is depending on this. > If there is somebody then be explicit about that. The last thing I want > to see is to preserve a cargo cult and build a design around it. It is mostly just a matter of going through and auditing all the places that are using PageReserved to identify pages that they aren't supposed to touch for whatever reason. >From what I can tell the issue appears to be the fact that the reserved bit is used to identify if a region of memory is "online" or "offline". So for example the call "online_pages_range" doesn't invoke the online_page_callback unless the first pfn in the range is marked as reserved. Another example Dan had pointed out was the saveable_page function in kernel/power/snapshot.c. > > > > > Regarding the post initialization required by devm_memremap_pages and > > > > > potentially others. Can we update the altmap which is already a way how > > > > > to get alternative struct pages by a constructor which we could call > > > > > from memmap_init_zone and do the post initialization? This would reduce > > > > > the additional loop in the caller while it would still fit the overall > > > > > design of the altmap and the core hotplug doesn't have to know anything > > > > > about DAX or whatever needs a special treatment. > > > > > > > > > > Does that make any sense? > > > > > > > > I think the only thing that is currently using the altmap is the ZONE_DEVICE > > > > memory init. Specifically I think it is only really used by the > > > > devm_memremap_pages version of things, and then only under certain > > > > circumstances. Also the HMM driver doesn't pass an altmap. What we would > > > > really need is a non-ZONE_DEVICE users of the altmap to really justify > > > > sticking with that as the preferred argument to pass. > > > > > > I am not aware of any upstream HMM user so I am not sure what are the > > > expectations there. But I thought that ZONE_DEVICE users use altmap. If > > > that is not generally true then we certainly have to think about a > > > better interface. > > > > I'm just basing my statement on the use of the move_pfn_range_to_zone call. > > The only caller that is actually passing the altmap is devm_memremap_pages > > and if I understand things correctly that is only used when we want to stare > > the vmmemmap on the same memory that we just hotplugged. > > Yes, and that is what I've called as allocator callback earlier. I am really not a fan of the callback approach. It just means we will have to do everything multiple times in terms of initialization. > > That is why it made more sense to me to just create a ZONE_DEVICE specific > > function for handling the page initialization because the one value I do > > have to pass is the dev_pagemap in both HMM and memremap case, and that has > > the altmap already embedded inside of it. > > And I have argued that this is a wrong approach to the problem. If you > need a very specific struct page initialization then create a init > (constructor) callback. The callback solution just ends up being more expensive because we lose multiple layers of possible optimization. By putting everything into on initization function we are able to let the compiler go through and optimize things to the point where we are essentially just doing something akin to one bit memcpy/memset where we are able to construct one set of page values and write that to every single page we have to initialize within a given page block. My concern is that we are going to see a 2-4x regression if I were to update the current patches I have to improve init performance in order to achieve the purity of the page initilization functions that you are looking for. I feel we are much better off having one function that can handle all cases and do so with high performance, than trying to construct a set of functions that end up having to reinitialize the same memory from the previous step and end up with us wasting cycles and duplicating overhead in multiple spots. In my mind the memmap_init_zone_device function is essentially just bringing the pages "online" after they have been mapped. That is why I am thinking it is probably okay to clear the reseved bit for the DAX pages at least. Once we have a hard go/no-go on Dan's patches that were consolidating the HMM functionality we could look at seeing if we need to move some functions around and what we can do to make it so that all the ZONE_DEVICE code can be moved as far out of the generic page init as possible while still maintaining reasonable initialization performance. _______________________________________________ Linux-nvdimm mailing list Linuxfirstname.lastname@example.org https://lists.01.org/mailman/listinfo/linux-nvdimm
next prev parent reply other threads:[~2018-10-29 15:59 UTC|newest] Thread overview: 77+ messages / expand[flat|nested] mbox.gz Atom feed top 2018-09-25 20:18 [PATCH v5 0/4] Address issues slowing persistent memory initialization Alexander Duyck 2018-09-25 20:19 ` [PATCH v5 1/4] mm: Remove now defunct NO_BOOTMEM from depends list for deferred init Alexander Duyck 2018-09-25 21:05 ` Mike Rapoport 2018-09-25 20:20 ` [PATCH v5 2/4] mm: Provide kernel parameter to allow disabling page init poisoning Alexander Duyck 2018-09-25 20:26 ` Dave Hansen 2018-09-25 20:38 ` Alexander Duyck 2018-09-25 22:14 ` Dave Hansen 2018-09-25 22:27 ` Alexander Duyck 2018-09-26 7:38 ` Michal Hocko 2018-09-26 15:24 ` Alexander Duyck 2018-09-26 15:39 ` Michal Hocko 2018-09-26 15:41 ` Dave Hansen 2018-09-26 16:18 ` Alexander Duyck 2018-09-26 15:36 ` Dave Hansen 2018-09-26 22:36 ` Andrew Morton 2018-09-25 20:20 ` [PATCH v5 3/4] mm: Create non-atomic version of SetPageReserved for init use Alexander Duyck 2018-09-25 20:21 ` [PATCH v5 4/4] mm: Defer ZONE_DEVICE page initialization to the point where we init pgmap Alexander Duyck 2018-09-26 7:55 ` Michal Hocko 2018-09-26 18:25 ` Alexander Duyck 2018-09-26 18:52 ` Dan Williams 2018-09-27 11:20 ` Michal Hocko 2018-09-27 11:09 ` Michal Hocko 2018-09-27 12:25 ` Oscar Salvador 2018-09-27 13:13 ` Michal Hocko 2018-09-27 14:50 ` Oscar Salvador 2018-09-27 15:41 ` David Hildenbrand 2018-09-28 8:12 ` Oscar Salvador 2018-09-28 8:44 ` Oscar Salvador 2018-09-28 15:50 ` Dan Williams 2018-09-27 12:32 ` Oscar Salvador 2018-10-08 21:01 ` Dan Williams 2018-10-08 21:38 ` Alexander Duyck 2018-10-08 22:00 ` Dan Williams 2018-10-08 22:07 ` Alexander Duyck 2018-10-08 22:36 ` Alexander Duyck 2018-10-08 22:59 ` Dan Williams 2018-10-08 23:34 ` [mm PATCH] memremap: Fix reference count for pgmap in devm_memremap_pages Alexander Duyck 2018-10-09 0:20 ` Dan Williams 2018-10-09 17:00 ` [PATCH v5 4/4] mm: Defer ZONE_DEVICE page initialization to the point where we init pgmap Yi Zhang 2018-10-09 18:04 ` Dan Williams 2018-10-09 20:26 ` Alexander Duyck 2018-10-09 21:19 ` Dan Williams 2018-10-10 12:52 ` Yi Zhang 2018-10-10 15:27 ` Alexander Duyck 2018-10-11 8:17 ` Yi Zhang 2018-10-10 9:58 ` Michal Hocko 2018-10-10 16:39 ` Alexander Duyck 2018-10-10 17:24 ` Michal Hocko 2018-10-10 17:39 ` Alexander Duyck 2018-10-10 17:53 ` Michal Hocko 2018-10-10 18:13 ` Alexander Duyck 2018-10-10 18:52 ` Michal Hocko 2018-10-11 8:55 ` Michal Hocko 2018-10-11 17:38 ` Alexander Duyck 2018-10-11 18:22 ` Dan Williams 2018-10-17 7:52 ` Michal Hocko 2018-10-17 15:02 ` Alexander Duyck 2018-10-29 14:12 ` Michal Hocko 2018-10-29 15:59 ` Alexander Duyck [this message] 2018-10-29 16:35 ` Michal Hocko 2018-10-29 17:01 ` Alexander Duyck 2018-10-29 17:24 ` Michal Hocko 2018-10-29 17:34 ` Dan Williams 2018-10-29 17:45 ` Michal Hocko 2018-10-29 17:42 ` Alexander Duyck 2018-10-29 18:18 ` Michal Hocko 2018-10-29 19:59 ` Alexander Duyck 2018-10-30 6:29 ` Michal Hocko 2018-10-30 6:55 ` Dan Williams 2018-10-30 8:17 ` Michal Hocko 2018-10-30 15:57 ` Dan Williams 2018-10-30 8:05 ` Oscar Salvador 2018-10-29 15:49 ` Dan Williams 2018-10-29 15:56 ` Michal Hocko 2018-10-10 18:18 ` Dan Williams 2018-10-11 8:39 ` Yi Zhang 2018-10-11 15:38 ` Alexander Duyck
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 \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --subject='Re: [PATCH v5 4/4] mm: Defer ZONE_DEVICE page initialization to the point where we init pgmap' \ /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
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).