| Message ID | 20240928021620.8369-5-kanchana.p.sridhar@intel.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm: zswap swap-out of large folios | expand |
On Fri, Sep 27, 2024 at 7:16 PM Kanchana P Sridhar <kanchana.p.sridhar@intel.com> wrote: > > With the introduction of zswap_store() swapping out large folios, > we need to efficiently update the objcg's memcg events once per > successfully stored folio. For instance, the 'ZSWPOUT' event needs > to be incremented by folio_nr_pages(). > > Signed-off-by: Kanchana P Sridhar <kanchana.p.sridhar@intel.com> > --- > include/linux/memcontrol.h | 20 ++++++++++++++++++++ > 1 file changed, 20 insertions(+) > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h > index 15c2716f9aa3..f47fd00c5eea 100644 > --- a/include/linux/memcontrol.h > +++ b/include/linux/memcontrol.h > @@ -1778,6 +1778,21 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, > rcu_read_unlock(); > } > > +static inline void count_objcg_events(struct obj_cgroup *objcg, > + enum vm_event_item idx, > + unsigned long count) > +{ > + struct mem_cgroup *memcg; > + > + if (!memcg_kmem_online()) > + return; > + > + rcu_read_lock(); > + memcg = obj_cgroup_memcg(objcg); > + count_memcg_events(memcg, idx, count); > + rcu_read_unlock(); > +} Instead of replicating the code in count_objcg_event(), we should change count_objcg_event() to become count_objcg_events() (i.e. add a count parameter). The existing callers can pass in 1, there's only 3 of them anyway (2 after patch 6), and they are all in zswap. > + > #else > static inline bool mem_cgroup_kmem_disabled(void) > { > @@ -1834,6 +1849,11 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, > { > } > > +static inline void count_objcg_events(struct obj_cgroup *objcg, > + enum vm_event_item idx, > + unsigned long count) > +{ > +} > #endif /* CONFIG_MEMCG */ > > #if defined(CONFIG_MEMCG) && defined(CONFIG_ZSWAP) > -- > 2.27.0 >
On 2024/9/28 11:02, Yosry Ahmed wrote: > On Fri, Sep 27, 2024 at 7:16 PM Kanchana P Sridhar > <kanchana.p.sridhar@intel.com> wrote: >> >> With the introduction of zswap_store() swapping out large folios, >> we need to efficiently update the objcg's memcg events once per >> successfully stored folio. For instance, the 'ZSWPOUT' event needs >> to be incremented by folio_nr_pages(). >> >> Signed-off-by: Kanchana P Sridhar <kanchana.p.sridhar@intel.com> >> --- >> include/linux/memcontrol.h | 20 ++++++++++++++++++++ >> 1 file changed, 20 insertions(+) >> >> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h >> index 15c2716f9aa3..f47fd00c5eea 100644 >> --- a/include/linux/memcontrol.h >> +++ b/include/linux/memcontrol.h >> @@ -1778,6 +1778,21 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, >> rcu_read_unlock(); >> } >> >> +static inline void count_objcg_events(struct obj_cgroup *objcg, >> + enum vm_event_item idx, >> + unsigned long count) >> +{ >> + struct mem_cgroup *memcg; >> + >> + if (!memcg_kmem_online()) >> + return; >> + >> + rcu_read_lock(); >> + memcg = obj_cgroup_memcg(objcg); >> + count_memcg_events(memcg, idx, count); >> + rcu_read_unlock(); >> +} > > Instead of replicating the code in count_objcg_event(), we should > change count_objcg_event() to become count_objcg_events() (i.e. add a > count parameter). The existing callers can pass in 1, there's only 3 > of them anyway (2 after patch 6), and they are all in zswap. Right, agree. > >> + >> #else >> static inline bool mem_cgroup_kmem_disabled(void) >> { >> @@ -1834,6 +1849,11 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, >> { >> } >> >> +static inline void count_objcg_events(struct obj_cgroup *objcg, >> + enum vm_event_item idx, >> + unsigned long count) >> +{ >> +} >> #endif /* CONFIG_MEMCG */ >> >> #if defined(CONFIG_MEMCG) && defined(CONFIG_ZSWAP) >> -- >> 2.27.0 >>
> -----Original Message----- > From: Yosry Ahmed <yosryahmed@google.com> > Sent: Friday, September 27, 2024 8:02 PM > To: Sridhar, Kanchana P <kanchana.p.sridhar@intel.com> > Cc: linux-kernel@vger.kernel.org; linux-mm@kvack.org; > hannes@cmpxchg.org; nphamcs@gmail.com; chengming.zhou@linux.dev; > usamaarif642@gmail.com; shakeel.butt@linux.dev; ryan.roberts@arm.com; > Huang, Ying <ying.huang@intel.com>; 21cnbao@gmail.com; akpm@linux- > foundation.org; Zou, Nanhai <nanhai.zou@intel.com>; Feghali, Wajdi K > <wajdi.k.feghali@intel.com>; Gopal, Vinodh <vinodh.gopal@intel.com> > Subject: Re: [PATCH v8 4/8] mm: Provide a new count_objcg_events() API for > batch event updates. > > On Fri, Sep 27, 2024 at 7:16 PM Kanchana P Sridhar > <kanchana.p.sridhar@intel.com> wrote: > > > > With the introduction of zswap_store() swapping out large folios, > > we need to efficiently update the objcg's memcg events once per > > successfully stored folio. For instance, the 'ZSWPOUT' event needs > > to be incremented by folio_nr_pages(). > > > > Signed-off-by: Kanchana P Sridhar <kanchana.p.sridhar@intel.com> > > --- > > include/linux/memcontrol.h | 20 ++++++++++++++++++++ > > 1 file changed, 20 insertions(+) > > > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h > > index 15c2716f9aa3..f47fd00c5eea 100644 > > --- a/include/linux/memcontrol.h > > +++ b/include/linux/memcontrol.h > > @@ -1778,6 +1778,21 @@ static inline void count_objcg_event(struct > obj_cgroup *objcg, > > rcu_read_unlock(); > > } > > > > +static inline void count_objcg_events(struct obj_cgroup *objcg, > > + enum vm_event_item idx, > > + unsigned long count) > > +{ > > + struct mem_cgroup *memcg; > > + > > + if (!memcg_kmem_online()) > > + return; > > + > > + rcu_read_lock(); > > + memcg = obj_cgroup_memcg(objcg); > > + count_memcg_events(memcg, idx, count); > > + rcu_read_unlock(); > > +} > > Instead of replicating the code in count_objcg_event(), we should > change count_objcg_event() to become count_objcg_events() (i.e. add a > count parameter). The existing callers can pass in 1, there's only 3 > of them anyway (2 after patch 6), and they are all in zswap. Thanks Yosry. This makes sense. I will incorporate this in v9. Thanks, Kanchana > > > + > > #else > > static inline bool mem_cgroup_kmem_disabled(void) > > { > > @@ -1834,6 +1849,11 @@ static inline void count_objcg_event(struct > obj_cgroup *objcg, > > { > > } > > > > +static inline void count_objcg_events(struct obj_cgroup *objcg, > > + enum vm_event_item idx, > > + unsigned long count) > > +{ > > +} > > #endif /* CONFIG_MEMCG */ > > > > #if defined(CONFIG_MEMCG) && defined(CONFIG_ZSWAP) > > -- > > 2.27.0 > >
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 15c2716f9aa3..f47fd00c5eea 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -1778,6 +1778,21 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, rcu_read_unlock(); } +static inline void count_objcg_events(struct obj_cgroup *objcg, + enum vm_event_item idx, + unsigned long count) +{ + struct mem_cgroup *memcg; + + if (!memcg_kmem_online()) + return; + + rcu_read_lock(); + memcg = obj_cgroup_memcg(objcg); + count_memcg_events(memcg, idx, count); + rcu_read_unlock(); +} + #else static inline bool mem_cgroup_kmem_disabled(void) { @@ -1834,6 +1849,11 @@ static inline void count_objcg_event(struct obj_cgroup *objcg, { } +static inline void count_objcg_events(struct obj_cgroup *objcg, + enum vm_event_item idx, + unsigned long count) +{ +} #endif /* CONFIG_MEMCG */ #if defined(CONFIG_MEMCG) && defined(CONFIG_ZSWAP)
With the introduction of zswap_store() swapping out large folios, we need to efficiently update the objcg's memcg events once per successfully stored folio. For instance, the 'ZSWPOUT' event needs to be incremented by folio_nr_pages(). Signed-off-by: Kanchana P Sridhar <kanchana.p.sridhar@intel.com> --- include/linux/memcontrol.h | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)