Re: [PATCH v1 09/14] mm: multigenerational lru: mm_struct list

["Huang, Ying" <ying.huang@intel.com>]

Yu Zhao <yuzhao@google.com> writes:

> On Wed, Mar 24, 2021 at 12:58 AM Huang, Ying <ying.huang@intel.com> wrote:
>>
>> Yu Zhao <yuzhao@google.com> writes:
>>
>> > On Mon, Mar 22, 2021 at 11:13:19AM +0800, Huang, Ying wrote:
>> >> Yu Zhao <yuzhao@google.com> writes:
>> >>
>> >> > On Wed, Mar 17, 2021 at 11:37:38AM +0800, Huang, Ying wrote:
>> >> >> Yu Zhao <yuzhao@google.com> writes:
>> >> >>
>> >> >> > On Tue, Mar 16, 2021 at 02:44:31PM +0800, Huang, Ying wrote:
>> >> >> > The scanning overhead is only one of the two major problems of the
>> >> >> > current page reclaim. The other problem is the granularity of the
>> >> >> > active/inactive (sizes). We stopped using them in making job
>> >> >> > scheduling decision a long time ago. I know another large internet
>> >> >> > company adopted a similar approach as ours, and I'm wondering how
>> >> >> > everybody else is coping with the discrepancy from those counters.
>> >> >>
>> >> >> From intuition, the scanning overhead of the full page table scanning
>> >> >> appears higher than that of the rmap scanning for a small portion of
>> >> >> system memory.  But form your words, you think the reality is the
>> >> >> reverse?  If others concern about the overhead too, finally, I think you
>> >> >> need to prove the overhead of the page table scanning isn't too higher,
>> >> >> or even lower with more data and theory.
>> >> >
>> >> > There is a misunderstanding here. I never said anything about full
>> >> > page table scanning. And this is not how it's done in this series
>> >> > either. I guess the misunderstanding has something to do with the cold
>> >> > memory tracking you are thinking about?
>> >>
>> >> If my understanding were correct, from the following code path in your
>> >> patch 10/14,
>> >>
>> >> age_active_anon
>> >>   age_lru_gens
>> >>     try_walk_mm_list
>> >>       walk_mm_list
>> >>         walk_mm
>> >>
>> >> So, in kswapd(), the page tables of many processes may be scanned
>> >> fully.  If the number of processes that are active are high, the
>> >> overhead may be high too.
>> >
>> > That's correct. Just in case we have different definitions of what we
>> > call "full":
>> >
>> >   I understand it as the full range of the address space of a process
>> >   that was loaded by switch_mm() at least once since the last scan.
>> >   This is not the case because we don't scan the full range -- we skip
>> >   holes and VMAs that are unevictable, as well as PTE tables that have
>> >   no accessed entries on x86_64, by should_skip_vma() and
>> >   CONFIG_HAVE_ARCH_PARENT_PMD_YOUNG.
>> >
>> >   If you are referring to the full range of PTE tables that have at
>> >   least one accessed entry, i.e., other 511 are not none  but have not
>> >   been accessed either since the last scan on x86_64, then yes, you
>> >   are right again :) This is the worse case scenario.
>>
>> OK.  So there's no fundamental difference between us on this.
>>
>> >> > This series uses page tables to discover page accesses when a system
>> >> > has run out of inactive pages. Under such a situation, the system is
>> >> > very likely to have a lot of page accesses, and using the rmap is
>> >> > likely to cost a lot more because its poor memory locality compared
>> >> > with page tables.
>> >>
>> >> This is the theory.  Can you verify this with more data?  Including the
>> >> CPU cycles or time spent scanning page tables?
>> >
>> > Yes, I'll be happy to do so as I should, because page table scanning
>> > is counterintuitive. Let me add more theory in case it's still unclear
>> > to others.
>> >
>> > From my understanding, the two fundamental questions we need to
>> > consider in terms of page reclaim are:
>> >
>> >   What are the sizes of hot clusters (spatial locality) should we
>> >   expect under memory pressure?
>> >
>> >   On smaller systems with 4GB memory, our observations are that the
>> >   average size of hot clusters found during each scan is 32KB. On
>> >   larger systems with hundreds of gigabytes of memory, it's well
>> >   above this value -- 512KB or larger. These values vary under
>> >   different workloads and with different memory allocators. Unless
>> >   done deliberately by memory allocators, e.g., Scudo as I've
>> >   mentioned earlier, it's safe to say if a PTE entry has been
>> >   accessed, its neighbors are likely to have been accessed too.
>> >
>> >   What's hot memory footprint (total size of hot clusters) should we
>> >   expect when we have run out of inactive pages?
>> >
>> >   Some numbers first: on large and heavily overcommitted systems, we
>> >   have observed close to 90% during a scan. Those systems have
>> >   millions of pages and using the rmap to find out which pages to
>> >   reclaim will just blow kswapd. On smaller systems with less memory
>> >   pressure (due to their weaker CPUs), this number is more reasonable,
>> >   ~50%. Here is some kswapd profiles from a smaller systems running
>> >   5.11:
>> >
>> >    the rmap                                 page table scan
>> >    ---------------------------------------------------------------------
>> >    31.03%  page_vma_mapped_walk             49.36%  lzo1x_1_do_compress
>> >    25.59%  lzo1x_1_do_compress               4.54%  page_vma_mapped_walk
>> >     4.63%  do_raw_spin_lock                  4.45%  memset_erms
>> >     3.89%  vma_interval_tree_iter_next       3.47%  walk_pte_range
>> >     3.33%  vma_interval_tree_subtree_search  2.88%  zram_bvec_rw
>> >
>> >   The page table scan is only twice as fast. Only larger systems,
>> >   it's usually more than 4 times, without THP. With THP, both are
>> >   negligible (<1% CPU usage). I can grab profiles from our servers
>> >   too if you are interested in seeing them on 4.15 kernel.
>>
>> Yes.  On a heavily overcommitted systems with high-percent hot pages,
>> the page table scanning works much better.  Because almost all pages
>> (and their mappings) will be scanned finally.
>>
>> But on a not-so-heavily overcommitted system with low-percent hot pages,
>> it's possible that rmap scanning works better.  That is, only a small
>> fraction of the pages need to be scanned.  I know that the page table
>> scanning may still work better in many cases.
>>
>> And another possibility, on a system with cool instead of completely
>> cold pages, that is, some pages are accessed at quite low frequency, but
>> not 0, there will be always some low-bandwidth memory reclaiming.  That
>> is, it's impossible to find a perfect solution with one or two full
>> scanning.  But we need to reclaim some pages periodically.  And I guess
>> there are no perfect (or very good) page reclaiming solutions for some
>> other situations too. Where what we can do are,
>>
>> - Avoid OOM, that is, reclaim some pages if possible.
>>
>> - Control the overhead of the page reclaiming.
>>
>> But this is theory only.  If anyone can point out that they are not
>> realistic at all, it's good too :-)
>>
>> >> > But, page tables can be sparse too, in terms of hot memory tracking.
>> >> > Dave has asked me to test the worst case scenario, which I'll do.
>> >> > And I'd be happy to share more data. Any specific workload you are
>> >> > interested in?
>> >>
>> >> We can start with some simple workloads that are easier to be reasoned.
>> >> For example,
>> >>
>> >> 1. Run the workload with hot and cold pages, when the free memory
>> >> becomes lower than the low watermark, kswapd will be waken up to scan
>> >> and reclaim some cold pages.  How long will it take to do that?  It's
>> >> expected that almost all pages need to be scanned, so that page table
>> >
>> > A typical scenario. Otherwise why would we have run out of cold pages
>> > and still be under memory? Because what's in memory is hot and
>> > therefore most of the them need to be scanned :)
>> >
>> >> scanning is expected to have less overhead.  We can measure how well it
>> >> is.
>> >
>> > Sounds good to me.
>> >
>> >> 2. Run the workload with hot and cold pages, if the whole working-set
>> >> cannot fit in DRAM, that is, the cold pages will be reclaimed and
>> >> swapped in regularly (for example tens MB/s).  It's expected that less
>> >> pages may be scanned with rmap, but the speed of page table scanning is
>> >> faster.
>> >
>> > So IIUC, this is a sustained memory pressure, i.e., servers constantly
>> > running under memory pressure?
>>
>> Yes.  The system can accommodate more workloads at the cost of
>> performance, as long as the end-user latency isn't unacceptable.  Or we
>> need some time to schedule more computing resources, so we need to run
>> in this condition for some while.
>>
>> But again, this is theory only.  I am glad if people can tell me that
>> this is unrealistic.
>>
>> >> 3. Run the workload with hot and cold pages, the system is
>> >> overcommitted, that is, some cold pages will be placed in swap.  But the
>> >> cold pages are cold enough, so there's almost no thrashing.  Then the
>> >> hot working-set of the workload changes, that is, some hot pages become
>> >> cold, while some cold pages becomes hot, so page reclaiming and swapin
>> >> will be triggered.
>> >
>> > This is usually what we see on clients, i.e., bursty workloads when
>> > switching from an active app to an inactive one.
>>
>> Thanks for your information.  Now I know a typical realistic use case :-)
>>
>> >> For each cases, we can use some different parameters.  And we can
>> >> measure something like the number of pages scanned, the time taken to
>> >> scan them, the number of page reclaimed and swapped in, etc.
>> >
>> > Thanks, I appreciate these -- very well thought test cases. I'll look
>> > into them and probably write some synthetic test cases. If you have
>> > some already, I'd love to get my hands one them.
>>
>> Sorry.  I have no test cases in hand.  Maybe we can add some into
>> Fengguang's vm-scalability test suite as follows.
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/wfg/vm-scalability.git/
>
> Hi Ying,
>
> I'm still investigating the test cases you suggested. I'm also
> wondering if it's possible to test the next version, which I'll post
> soon, with Intel's 0-Day infra.

Sure.  But now 0-Day has only quite limited coverage for swap testing.
Including the swap test in vm-scalability.git, and several test cases
with pmbench.  I think it's good to improve the coverage of 0-Day for
swap.  But it needs some time.

Best Regards,
Huang, Ying