Re: [RFC 0/3] mm: Discard lazily freed pages when migrating

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

Mel Gorman <mgorman@suse.de> writes:

> On Tue, Mar 03, 2020 at 09:51:56AM +0800, Huang, Ying wrote:
>> Mel Gorman <mgorman@suse.de> writes:
>> > On Mon, Mar 02, 2020 at 07:23:12PM +0800, Huang, Ying wrote:
>> >> If some applications cannot tolerate the latency incurred by the memory
>> >> allocation and zeroing.  Then we cannot discard instead of migrate
>> >> always.  While in some situations, less memory pressure can help.  So
>> >> it's better to let the administrator and the application choose the
>> >> right behavior in the specific situation?
>> >> 
>> >
>> > Is there an application you have in mind that benefits from discarding
>> > MADV_FREE pages instead of migrating them?
>> >
>> > Allowing the administrator or application to tune this would be very
>> > problematic. An application would require an update to the system call
>> > to take advantage of it and then detect if the running kernel supports
>> > it. An administrator would have to detect that MADV_FREE pages are being
>> > prematurely discarded leading to a slowdown and that is hard to detect.
>> > It could be inferred from monitoring compaction stats and checking
>> > if compaction activity is correlated with higher minor faults in the
>> > target application. Proving the correlation would require using the perf
>> > software event PERF_COUNT_SW_PAGE_FAULTS_MIN and matching the addresses
>> > to MADV_FREE regions that were freed prematurely. That is not an obvious
>> > debugging step to take when an application detects latency spikes.
>> >
>> > Now, you could add a counter specifically for MADV_FREE pages freed for
>> > reasons other than memory pressure and hope the administrator knows about
>> > the counter and what it means. That type of knowledge could take a long
>> > time to spread so it's really very important that there is evidence of
>> > an application that suffers due to the current MADV_FREE and migration
>> > behaviour.
>> 
>> OK.  I understand that this patchset isn't a universal win, so we need
>> some way to justify it.  I will try to find some application for that.
>> 
>> Another thought, as proposed by David Hildenbrand, it's may be a
>> universal win to discard clean MADV_FREE pages when migrating if there are
>> already memory pressure on the target node.  For example, if the free
>> memory on the target node is lower than high watermark?
>> 
>
> That is an extremely specific corner case that is not likely to occur.
> NUMA balancing is not going to migrate a MADV_FREE page under these
> circumstances as a write cancels MADV_FREE is read attempt will probably
> fail to allocate a destination page in alloc_misplaced_dst_page so the
> data gets lost instead of remaining remote. sys_movepages is a possibility
> but the circumstances of an application delibertly trying to migrate to
> a loaded node is low. Compaction never migrates cross-node so the state
> of a remote node under pressure do not matter.
>
> Once again, there needs to be a reasonable use case to be able to
> meaningfully balance between the benefits and risks of changing the
> MADV_FREE semantics.

OK.  Will try to find some workloads for this.

Best Regards,
Huang, Ying