| Message ID | 20221227121341.981235373@redhat.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Ensure quiet_vmstat() is called when returning to userpace and when idle tick is stopped | expand |
On Tue, Dec 27, 2022 at 09:11:39AM -0300, Marcelo Tosatti wrote: > @@ -606,6 +608,7 @@ static inline void mod_zone_state(struct > > if (z) > zone_page_state_add(z, zone, item); > + vmstat_mark_dirty(); > } > > void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, > @@ -674,6 +677,7 @@ static inline void mod_node_state(struct > > if (z) > node_page_state_add(z, pgdat, item); > + vmstat_mark_dirty(); Looking at this further, about the two above chunks, there is a risk to mark the wrong CPU dirty because those functions are preemptible and rely on this_cpu_cmpxchg() to deal with preemption. Thanks.
On Fri, Dec 30, 2022 at 02:21:32PM +0100, Frederic Weisbecker wrote: > On Tue, Dec 27, 2022 at 09:11:39AM -0300, Marcelo Tosatti wrote: > > @@ -606,6 +608,7 @@ static inline void mod_zone_state(struct > > > > if (z) > > zone_page_state_add(z, zone, item); > > + vmstat_mark_dirty(); > > } > > > > void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, > > @@ -674,6 +677,7 @@ static inline void mod_node_state(struct > > > > if (z) > > node_page_state_add(z, pgdat, item); > > + vmstat_mark_dirty(); > > Looking at this further, about the two above chunks, there is a risk to > mark the wrong CPU dirty because those functions are preemptible and rely > on this_cpu_cmpxchg() to deal with preemption. > > Thanks. Hi Frederic, Yes, good catch: if the CPU is preempted after this_cpu_cmpxchg(), but before vmstat_mark_dirty, then one ends up with a CPU with per-CPU vm counters dirty and the per-CPU vmstat dirty bit unset. This could cause a CPU to remain with the per-CPU vm counters dirty for longer than sysctl_stat_interval. Should move vmstat_mark_dirty() above "if (z)", then do preempt_disable() on function entry and preempt_enable() after vmstat_mark_dirty. Luckily preempt_disable()/preempt_enable() is much cheaper than local_irq_disable()/local_irq_enable().
On Fri 2022-12-30 14:21 +0100, Frederic Weisbecker wrote: > On Tue, Dec 27, 2022 at 09:11:39AM -0300, Marcelo Tosatti wrote: > > @@ -606,6 +608,7 @@ static inline void mod_zone_state(struct > > > > if (z) > > zone_page_state_add(z, zone, item); > > + vmstat_mark_dirty(); > > } > > > > void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, > > @@ -674,6 +677,7 @@ static inline void mod_node_state(struct > > > > if (z) > > node_page_state_add(z, pgdat, item); > > + vmstat_mark_dirty(); > > Looking at this further, about the two above chunks, there is a risk to > mark the wrong CPU dirty because those functions are preemptible and rely > on this_cpu_cmpxchg() to deal with preemption. Hi Frederic, Agreed. Kind regards,
Index: linux-2.6/mm/vmstat.c =================================================================== --- linux-2.6.orig/mm/vmstat.c +++ linux-2.6/mm/vmstat.c @@ -381,6 +381,7 @@ void __mod_zone_page_state(struct zone * x = 0; } __this_cpu_write(*p, x); + vmstat_mark_dirty(); preempt_enable_nested(); } @@ -417,6 +418,7 @@ void __mod_node_page_state(struct pglist x = 0; } __this_cpu_write(*p, x); + vmstat_mark_dirty(); preempt_enable_nested(); } @@ -606,6 +608,7 @@ static inline void mod_zone_state(struct if (z) zone_page_state_add(z, zone, item); + vmstat_mark_dirty(); } void mod_zone_page_state(struct zone *zone, enum zone_stat_item item, @@ -674,6 +677,7 @@ static inline void mod_node_state(struct if (z) node_page_state_add(z, pgdat, item); + vmstat_mark_dirty(); } void mod_node_page_state(struct pglist_data *pgdat, enum node_stat_item item, @@ -828,6 +832,14 @@ static int refresh_cpu_vm_stats(bool do_ int global_node_diff[NR_VM_NODE_STAT_ITEMS] = { 0, }; int changes = 0; + /* + * Clear vmstat_dirty before clearing the percpu vmstats. + * If interrupts are enabled, it is possible that an interrupt + * or another task modifies a percpu vmstat, which will + * set vmstat_dirty to true. + */ + vmstat_clear_dirty(); + for_each_populated_zone(zone) { struct per_cpu_zonestat __percpu *pzstats = zone->per_cpu_zonestats; #ifdef CONFIG_NUMA @@ -1957,35 +1969,6 @@ static void vmstat_update(struct work_st } /* - * Check if the diffs for a certain cpu indicate that - * an update is needed. - */ -static bool need_update(int cpu) -{ - pg_data_t *last_pgdat = NULL; - struct zone *zone; - - for_each_populated_zone(zone) { - struct per_cpu_zonestat *pzstats = per_cpu_ptr(zone->per_cpu_zonestats, cpu); - struct per_cpu_nodestat *n; - - /* - * The fast way of checking if there are any vmstat diffs. - */ - if (memchr_inv(pzstats->vm_stat_diff, 0, sizeof(pzstats->vm_stat_diff))) - return true; - - if (last_pgdat == zone->zone_pgdat) - continue; - last_pgdat = zone->zone_pgdat; - n = per_cpu_ptr(zone->zone_pgdat->per_cpu_nodestats, cpu); - if (memchr_inv(n->vm_node_stat_diff, 0, sizeof(n->vm_node_stat_diff))) - return true; - } - return false; -} - -/* * Switch off vmstat processing and then fold all the remaining differentials * until the diffs stay at zero. The function is used by NOHZ and can only be * invoked when tick processing is not active. @@ -1995,10 +1978,7 @@ void quiet_vmstat(void) if (system_state != SYSTEM_RUNNING) return; - if (!delayed_work_pending(this_cpu_ptr(&vmstat_work))) - return; - - if (!need_update(smp_processor_id())) + if (!is_vmstat_dirty()) return; /* @@ -2029,7 +2009,7 @@ static void vmstat_shepherd(struct work_ for_each_online_cpu(cpu) { struct delayed_work *dw = &per_cpu(vmstat_work, cpu); - if (!delayed_work_pending(dw) && need_update(cpu)) + if (!delayed_work_pending(dw) && per_cpu(vmstat_dirty, cpu)) queue_delayed_work_on(cpu, mm_percpu_wq, dw, 0); cond_resched();