| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20100728161301.c9cfd3e6.kamezawa.hiroyu@jp.fujitsu.com>
Date: Wed, 28 Jul 2010 16:13:01 +0900
From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
To: Greg Thelen <gthelen@...gle.com>
Cc: "linux-mm\@kvack.org" <linux-mm@...ck.org>,
"nishimura\@mxp.nes.nec.co.jp" <nishimura@....nes.nec.co.jp>,
"balbir\@linux.vnet.ibm.com" <balbir@...ux.vnet.ibm.com>,
m-ikeda@...jp.nec.com,
"akpm\@linux-foundation.org" <akpm@...ux-foundation.org>,
"linux-kernel\@vger.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [RFC][PATCH 5/7][memcg] memcg lockless update of file mapped
On Wed, 28 Jul 2010 00:09:21 -0700
Greg Thelen <gthelen@...gle.com> wrote:
> KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com> writes:
>
> > From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
> >
> > At accounting file events per memory cgroup, we need to find memory cgroup
> > via page_cgroup->mem_cgroup. Now, we use lock_page_cgroup().
> >
> > But, considering the context which page-cgroup for files are accessed,
> > we can use alternative light-weight mutual execusion in the most case.
> > At handling file-caches, the only race we have to take care of is "moving"
> > account, IOW, overwriting page_cgroup->mem_cgroup. Because file status
> > update is done while the page-cache is in stable state, we don't have to
> > take care of race with charge/uncharge.
> >
> > Unlike charge/uncharge, "move" happens not so frequently. It happens only when
> > rmdir() and task-moving (with a special settings.)
> > This patch adds a race-checker for file-cache-status accounting v.s. account
> > moving. The new per-cpu-per-memcg counter MEM_CGROUP_ON_MOVE is added.
> > The routine for account move
> > 1. Increment it before start moving
> > 2. Call synchronize_rcu()
> > 3. Decrement it after the end of moving.
> > By this, file-status-counting routine can check it needs to call
> > lock_page_cgroup(). In most case, I doesn't need to call it.
> >
> > Note: update_file_mapped is safe against charge/uncharge even if it's
> > not under address_space->tree_lock or lock_page(). Because it's under
> > page_table_lock(), anyone can't unmap it...then, anyone can't uncharge().
> >
> >
> >
> > Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
> > ---
> > mm/memcontrol.c | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++----
> > 1 file changed, 60 insertions(+), 4 deletions(-)
> >
> > Index: mmotm-0719/mm/memcontrol.c
> > ===================================================================
> > --- mmotm-0719.orig/mm/memcontrol.c
> > +++ mmotm-0719/mm/memcontrol.c
> > @@ -89,6 +89,7 @@ enum mem_cgroup_stat_index {
> > MEM_CGROUP_STAT_PGPGOUT_COUNT, /* # of pages paged out */
> > MEM_CGROUP_STAT_SWAPOUT, /* # of pages, swapped out */
> > MEM_CGROUP_EVENTS, /* incremented at every pagein/pageout */
> > + MEM_CGROUP_ON_MOVE, /* A check for locking move account/status */
> >
> > MEM_CGROUP_STAT_NSTATS,
> > };
> > @@ -1071,7 +1072,48 @@ static unsigned int get_swappiness(struc
> > return swappiness;
> > }
> >
> > -/* A routine for testing mem is not under move_account */
> > +static void mem_cgroup_start_move(struct mem_cgroup *mem)
> > +{
> > + int cpu;
> > + /* for fast checking in mem_cgroup_update_file_stat() etc..*/
> > + spin_lock(&mc.lock);
> > + for_each_possible_cpu(cpu)
> > + per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) += 1;
> > + spin_unlock(&mc.lock);
> > +
> > + synchronize_rcu();
> > +}
> > +
> > +static void mem_cgroup_end_move(struct mem_cgroup *mem)
> > +{
> > + int cpu;
> > +
> > + if (!mem)
> > + return;
> > + /* for fast checking in mem_cgroup_update_file_stat() etc..*/
> > + spin_lock(&mc.lock);
> > + for_each_possible_cpu(cpu)
> > + per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) -= 1;
> > + spin_unlock(&mc.lock);
> > +}
> > +
> > +/*
> > + * mem_cgroup_is_moved -- checking a cgroup is mc.from target or not.
> > + * used for avoiding race.
> > + * mem_cgroup_under_move -- checking a cgroup is mc.from or mc.to or
> > + * under hierarchy of them. used for waiting at
> > + * memory pressure.
> > + * Result of is_moved can be trusted until the end of rcu_read_unlock().
> > + * The caller must do
> > + * rcu_read_lock();
> > + * result = mem_cgroup_is_moved();
> > + * .....make use of result here....
> > + * rcu_read_unlock();
> > + */
> > +static bool mem_cgroup_is_moved(struct mem_cgroup *mem)
> > +{
>
> Could we add an assertion to confirm locking contract is upheld:
> VM_BUG_ON(!rcu_read_lock_held());
>
Hmm. there is an only one caller...I'll add one or I don't make
this as a funciton.
> > + return this_cpu_read(mem->stat->count[MEM_CGROUP_ON_MOVE]) > 0;
> > +}
> >
> > static bool mem_cgroup_under_move(struct mem_cgroup *mem)
> > {
> > @@ -1470,13 +1512,21 @@ void mem_cgroup_update_file_mapped(struc
> > {
> > struct mem_cgroup *mem;
> > struct page_cgroup *pc;
> > + bool need_lock = false;
> >
> > pc = lookup_page_cgroup(page);
> > if (unlikely(!pc))
> > return;
> > -
> > - lock_page_cgroup(pc);
> > + rcu_read_lock();
> > mem = id_to_mem(pc->mem_cgroup);
> > + if (!mem)
> > + goto done;
> > + need_lock = mem_cgroup_is_moved(mem);
> > + if (need_lock) {
> > + /* need to serialize with move_account */
> > + lock_page_cgroup(pc);
> > + mem = id_to_mem(pc->mem_cgroup);
> > + }
> > if (!mem || !PageCgroupUsed(pc))
> > goto done;
>
> Could we add a preemption() check here to ensure that the
> __this_cpu_xxx() is safe to use?
>
Hmm, ok.
Thanks,
-Kame
> /*
> * Preemption is already disabled. We can use __this_cpu_xxx
> */
> + VM_BUG_ON(preemptible());
>
> > @@ -1492,7 +1542,9 @@ void mem_cgroup_update_file_mapped(struc
> > }
> >
> > done:
> > - unlock_page_cgroup(pc);
> > + if (need_lock)
> > + unlock_page_cgroup(pc);
> > + rcu_read_unlock();
> > }
> >
> > /*
> > @@ -3024,6 +3076,7 @@ move_account:
> > lru_add_drain_all();
> > drain_all_stock_sync();
> > ret = 0;
> > + mem_cgroup_start_move(mem);
> > for_each_node_state(node, N_HIGH_MEMORY) {
> > for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) {
> > enum lru_list l;
> > @@ -3037,6 +3090,7 @@ move_account:
> > if (ret)
> > break;
> > }
> > + mem_cgroup_end_move(mem);
> > memcg_oom_recover(mem);
> > /* it seems parent cgroup doesn't have enough mem */
> > if (ret == -ENOMEM)
> > @@ -4503,6 +4557,7 @@ static void mem_cgroup_clear_mc(void)
> > mc.to = NULL;
> > mc.moving_task = NULL;
> > spin_unlock(&mc.lock);
> > + mem_cgroup_end_move(from);
> > memcg_oom_recover(from);
> > memcg_oom_recover(to);
> > wake_up_all(&mc.waitq);
> > @@ -4533,6 +4588,7 @@ static int mem_cgroup_can_attach(struct
> > VM_BUG_ON(mc.moved_charge);
> > VM_BUG_ON(mc.moved_swap);
> > VM_BUG_ON(mc.moving_task);
> > + mem_cgroup_start_move(from);
> > spin_lock(&mc.lock);
> > mc.from = from;
> > mc.to = mem;
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists