On Thu, Jul 12, 2012 at 05:10:43PM +0800, Wanpeng Li wrote: > On Thu, Jul 12, 2012 at 04:54:07PM +0800, Wanpeng Li wrote: > >On Wed, Jul 11, 2012 at 07:02:13PM +0200, Johannes Weiner wrote: > >>Compaction (and page migration in general) can currently be hindered > >>through pages being owned by memory cgroups that are at their limits > >>and unreclaimable. > >> > >>The reason is that the replacement page is being charged against the > >>limit while the page being replaced is also still charged. But this > >>seems unnecessary, given that only one of the two pages will still be > >>in use after migration finishes. > >> > >>This patch changes the memcg migration sequence so that the > >>replacement page is not charged. Whatever page is still in use after > >>successful or failed migration gets to keep the charge of the page > >>that was going to be replaced. > >> > >>The replacement page will still show up temporarily in the rss/cache > >>statistics, this can be fixed in a later patch as it's less urgent. > >> > > > >So I want to know after this patch be merged if mem_cgroup_wait_acct_move > >still make sense, if the answer is no, I will send a patch to remove it. > > And if this still make sense, I want to change check in > mem_cgroup_do_charge: > > if (mem_cgroup_wait_acct_move(mem_over_limit)) > return CHARGE_RETRY; > > => > > if (mem_cgroup_wait_acct_move(mem_over_limit) && > mem_cgroup_margin(mem_over_limit) >= nr_pages) > return CHARGE_RETRY; > > Since mem_cgroup_relcaim can reclaim some pages, but in > mem_cgroup_reclaim function there are some exit condition: > > total += try_to_free_mem_cgroup_pages(memcg, gfp_mask, noswap); > if(total && (flag & MEM_CGROUP_RECLAIM_SHRINK)) > break; > > and > > if (mem_cgroup_margin(memcg)) > break; > > So maybe mem_cgroup_reclaim not reclaim enough pages >= nr_pages, this > time we should go to mem_cgroup_handle_oom instead of return > CHARGE_RETRY. > > Hopefully, you can verify if my idea make sense. Sorry, but this is a waste of your time, my time, and that of everybody else in this thread. I will ignore any subsequent proposals from you unless they start with a coherent description of an actual problem. Something that has impact on userspace, or significant impact on kernel development. If there is a bug I don't see in your description above, than please explain how it affects userspace. If the code or comments are cryptic and can be simplified or clarified, please explain. -- To unsubscribe from this list: send the line "unsubscribe cgroups" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
