On Wed, 21 Nov 2012 15:01:50 +0000 Mel Gorman mgorman@suse.de wrote:
On Tue, Nov 20, 2012 at 10:12:28AM -0800, David Rientjes wrote:
On Mon, 19 Nov 2012, Anton Vorontsov wrote:
We try to make userland freeing resources when the system becomes low on memory. Once we're short on memory, sometimes it's better to discard (free) data, rather than let the kernel to drain file caches or even start swapping.
To add another usecase: its possible to modify our version of malloc (or any malloc) so that memory that is free()'d can be released back to the kernel only when necessary, i.e. when keeping the extra memory around starts to have a detremental effect on the system, memcg, or cpuset. When there is an abundance of memory available such that allocations need not defragment or reclaim memory to be allocated, it can improve performance to keep a memory arena from which to allocate from immediately without calling the kernel.
A potential third use case is a variation of the first for batch systems. If it's running low priority tasks and a high priority task starts that results in memory pressure then the job scheduler may decide to move the low priority jobs elsewhere (or cancel them entirely).
A similar use case is monitoring systems running high priority workloads that should never swap. It can be easily detected if the system starts swapping but a pressure notification might act as an early warning system that something is happening on the system that might cause the primary workload to start swapping.
I hope Anton's writing all of this down ;)
The proposed API bugs me a bit. It seems simplistic. I need to have a quality think about this. Maybe the result of that think will be to suggest an interface which can be extended in a back-compatible fashion later on, if/when the simplistic nature becomes a problem.