On 27.05.2019 17:21, Michal Hocko wrote:
On Mon 27-05-19 16:12:23, Michal Hocko wrote:
[Cc linux-api. Please always cc this list when proposing a new user
visible api. Keeping the rest of the email intact for reference]
On Mon 27-05-19 13:05:58, Konstantin Khlebnikov wrote:
[...]
This implements manual kswapd-style memory reclaim initiated by userspace.
It reclaims both physical memory and cgroup pages. It works in context of
task who calls syscall madvise thus cpu time is accounted correctly.
I do not follow. Does this mean that the madvise always reclaims from
the memcg the process is member of?
First it reclaims in its own memcg while limit - usage < requested.
Then repeats this in parent memcg and so on. And at least pokes global
direct reclaimer while system wide free memory is less than requested.
So, if machine is divided into containers without overcommit global
reclaim will never happens - memcg will free enough memory.
