On Wed, May 11, 2022 at 07:32:18PM +0200, Michal Koutný wrote: > Hello. > > On Tue, May 10, 2022 at 11:28:47AM -0400, Johannes Weiner <hannes@xxxxxxxxxxx> wrote: > > +void obj_cgroup_charge_zswap(struct obj_cgroup *objcg, size_t size) > > +{ > > + struct mem_cgroup *memcg; > > + > > + VM_WARN_ON_ONCE(!(current->flags & PF_MEMALLOC)); > > + > > + /* PF_MEMALLOC context, charging must succeed */ > > + if (obj_cgroup_charge(objcg, GFP_KERNEL, size)) > > + VM_WARN_ON_ONCE(1); > > IIUC, the objcg is derived from the compressed page, i.e. same memcg > (reparenting neglected for now). This memcg's memory.current is then > charged with the compressed object size. Correct. After which the uncompressed page is reclaimed and uncharged. So the zswapout process will reduce the charge bottom line. > Do I get it right that memory.zswap.current is a subset of memory.current? > > (And that zswap is limited both by memory.max and memory.zswap.max?) Yes. Zswap is a memory consumer, and we want the compressed part of a workload's memory to show up in the total memory footprint. memory.zswap.* are there to configure zswap policy, within the boundaries of available memory - it's by definition a subset.