Michal Hocko <mhocko@xxxxxxxx> writes:
> On Tue 25-02-25 22:22:58, Gabriel Krisman Bertazi wrote:
>> Commit 96a5c186efff ("mm/page_alloc.c: don't show protection in zone's
>> ->lowmem_reserve[] for empty zone") removes the protection of lower
>> zones from allocations targeting memory-less high zones. This had an
>> unintended impact on the pattern of reclaims because it makes the
>> high-zone-targeted allocation more likely to succeed in lower zones,
>> which adds pressure to said zones. I.e, the following corresponding
>> checks in zone_watermark_ok/zone_watermark_fast are less likely to
>> trigger:
>>
>> if (free_pages <= min + z->lowmem_reserve[highest_zoneidx])
>> return false;
>>
>> As a result, we are observing an increase in reclaim and kswapd scans,
>> due to the increased pressure. This was initially observed as increased
>> latency in filesystem operations when benchmarking with fio on a machine
>> with some memory-less zones, but it has since been associated with
>> increased contention in locks related to memory reclaim. By reverting
>> this patch, the original performance was recovered on that machine.
>
> I think it would be nice to show the memory layout on that machine (is
> there any movable or device zone)?
>
> Exact reclaim patterns are really hard to predict and it is little bit
> surprising the said patch has caused an increased kswapd activity
> because I would expect that there will be more reclaim with the lowmem
> reserves in place. But it is quite possible that the higher zone memory
> pressure is just tipping over and increase the lowmem pressure enough
> that it shows up.
For reference, I collected vmstat with and without this patch on a
freshly booted system running intensive randread io from an nvme for 5
minutes. I got:
rpm-6.12.0-slfo.1.2 -> pgscan_kswapd 5629543865
Patched -> pgscan_kswapd 33580844
33M scans is similar to what we had in kernels predating this patch.
These numbers is fairly representative of the workload on this machine, as
measured in several runs. So we are talking about a 2-order of
magnitude increase.
Attached is the zoneinfo with my revert patch applied.
Attachment:
zoneinfo
Description: Binary data
-- Gabriel Krisman Bertazi
