| Message ID | 1547955021-11520-3-git-send-email-duanxiongchun@bytedance.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | fix offline memcgroup still hold in memory | expand |
On Sat 19-01-19 22:30:18, Xiongchun Duan wrote: > Add timer to trigger workqueue which will scan offline memcgroup and call trigger > memcgroup force_empty worker to force_empty itself. This requires much more explanation but it looks like a complete hack to me. Why do we need a timer at all? Why is the hardcoded timeout a generic enough? > Signed-off-by: Xiongchun Duan <duanxiongchun@bytedance.com> > --- > include/linux/memcontrol.h | 1 + > mm/memcontrol.c | 23 +++++++++++++++++++++++ > 2 files changed, 24 insertions(+) > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h > index d6fbb77..0a29f7f 100644 > --- a/include/linux/memcontrol.h > +++ b/include/linux/memcontrol.h > @@ -313,6 +313,7 @@ struct mem_cgroup { > > int max_retry; > int current_retry; > + unsigned long timer_jiffies; > > struct mem_cgroup_per_node *nodeinfo[0]; > /* WARNING: nodeinfo must be the last member here */ > diff --git a/mm/memcontrol.c b/mm/memcontrol.c > index 2b13c2b..4db08b7 100644 > --- a/mm/memcontrol.c > +++ b/mm/memcontrol.c > @@ -81,6 +81,8 @@ > int sysctl_cgroup_default_retry_min; > int sysctl_cgroup_default_retry_max = 16; > > +struct timer_list empty_trigger; > + > struct mem_cgroup *root_mem_cgroup __read_mostly; > > #define MEM_CGROUP_RECLAIM_RETRIES 5 > @@ -2933,6 +2935,11 @@ static ssize_t mem_cgroup_force_empty_write(struct kernfs_open_file *of, > return mem_cgroup_force_empty(memcg) ?: nbytes; > } > > +static void add_force_empty_list(struct mem_cgroup *memcg) > +{ > + > +} > + > static u64 mem_cgroup_hierarchy_read(struct cgroup_subsys_state *css, > struct cftype *cft) > { > @@ -4566,11 +4573,26 @@ static int mem_cgroup_css_online(struct cgroup_subsys_state *css) > return 0; > } > > +void empty_timer_trigger(struct timer_list *t) > +{ > + > +} > + > static void mem_cgroup_css_offline(struct cgroup_subsys_state *css) > { > struct mem_cgroup *memcg = mem_cgroup_from_css(css); > struct mem_cgroup_event *event, *tmp; > > + if (memcg->max_retry != 0) { > + memcg->current_retry = 1; > + mem_cgroup_force_empty(memcg); > + if (page_counter_read(&memcg->memory) && > + memcg->max_retry != 1) { > + memcg->timer_jiffies = jiffies + HZ; > + add_force_empty_list(memcg); > + } > + } > + > /* > * Unregister events and notify userspace. > * Notify userspace about cgroup removing only after rmdir of cgroup > @@ -6368,6 +6390,7 @@ static int __init mem_cgroup_init(void) > memcg_kmem_cache_wq = alloc_workqueue("memcg_kmem_cache", 0, 1); > BUG_ON(!memcg_kmem_cache_wq); > #endif > + timer_setup(&empty_trigger, empty_timer_trigger, 0); > > cpuhp_setup_state_nocalls(CPUHP_MM_MEMCQ_DEAD, "mm/memctrl:dead", NULL, > memcg_hotplug_cpu_dead); > -- > 1.8.3.1
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index d6fbb77..0a29f7f 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -313,6 +313,7 @@ struct mem_cgroup { int max_retry; int current_retry; + unsigned long timer_jiffies; struct mem_cgroup_per_node *nodeinfo[0]; /* WARNING: nodeinfo must be the last member here */ diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 2b13c2b..4db08b7 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -81,6 +81,8 @@ int sysctl_cgroup_default_retry_min; int sysctl_cgroup_default_retry_max = 16; +struct timer_list empty_trigger; + struct mem_cgroup *root_mem_cgroup __read_mostly; #define MEM_CGROUP_RECLAIM_RETRIES 5 @@ -2933,6 +2935,11 @@ static ssize_t mem_cgroup_force_empty_write(struct kernfs_open_file *of, return mem_cgroup_force_empty(memcg) ?: nbytes; } +static void add_force_empty_list(struct mem_cgroup *memcg) +{ + +} + static u64 mem_cgroup_hierarchy_read(struct cgroup_subsys_state *css, struct cftype *cft) { @@ -4566,11 +4573,26 @@ static int mem_cgroup_css_online(struct cgroup_subsys_state *css) return 0; } +void empty_timer_trigger(struct timer_list *t) +{ + +} + static void mem_cgroup_css_offline(struct cgroup_subsys_state *css) { struct mem_cgroup *memcg = mem_cgroup_from_css(css); struct mem_cgroup_event *event, *tmp; + if (memcg->max_retry != 0) { + memcg->current_retry = 1; + mem_cgroup_force_empty(memcg); + if (page_counter_read(&memcg->memory) && + memcg->max_retry != 1) { + memcg->timer_jiffies = jiffies + HZ; + add_force_empty_list(memcg); + } + } + /* * Unregister events and notify userspace. * Notify userspace about cgroup removing only after rmdir of cgroup @@ -6368,6 +6390,7 @@ static int __init mem_cgroup_init(void) memcg_kmem_cache_wq = alloc_workqueue("memcg_kmem_cache", 0, 1); BUG_ON(!memcg_kmem_cache_wq); #endif + timer_setup(&empty_trigger, empty_timer_trigger, 0); cpuhp_setup_state_nocalls(CPUHP_MM_MEMCQ_DEAD, "mm/memctrl:dead", NULL, memcg_hotplug_cpu_dead);
Add timer to trigger workqueue which will scan offline memcgroup and call trigger memcgroup force_empty worker to force_empty itself. Signed-off-by: Xiongchun Duan <duanxiongchun@bytedance.com> --- include/linux/memcontrol.h | 1 + mm/memcontrol.c | 23 +++++++++++++++++++++++ 2 files changed, 24 insertions(+)