| 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
| ||
|
Date: Fri, 23 Apr 2010 13:41:41 +0900
From: Daisuke Nishimura <nishimura@....nes.nec.co.jp>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
Li Zefan <lizf@...fujitsu.com>
Cc: Balbir Singh <balbir@...ux.vnet.ibm.com>,
"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
"linux-mm@...ck.org" <linux-mm@...ck.org>,
LKML <linux-kernel@...r.kernel.org>,
"akpm@...ux-foundation.org" <akpm@...ux-foundation.org>,
Daisuke Nishimura <nishimura@....nes.nec.co.jp>
Subject: Re: [BUGFIX][PATCH] memcg rcu lock fix v3
On Fri, 23 Apr 2010 13:03:49 +0900, KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com> wrote:
> On Fri, 23 Apr 2010 12:58:14 +0900
> KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com> wrote:
>
> > On Fri, 23 Apr 2010 11:55:16 +0800
> > Li Zefan <lizf@...fujitsu.com> wrote:
> >
> > > Li Zefan wrote:
> > > > KAMEZAWA Hiroyuki wrote:
> > > >> On Fri, 23 Apr 2010 11:00:41 +0800
> > > >> Li Zefan <lizf@...fujitsu.com> wrote:
> > > >>
> > > >>> with CONFIG_PROVE_RCU=y, I saw this warning, it's because
> > > >>> css_id() is not under rcu_read_lock().
> > > >>>
> > > >> Ok. Thank you for reporting.
> > > >> This is ok ?
> > > >
> > > > Yes, and I did some more simple tests on memcg, no more warning
> > > > showed up.
> > > >
> > >
> > > oops, after trigging oom, I saw 2 more warnings:
> > >
> >
> > Thank you for good testing.
> v3 here...sorry too rapid posting...
>
Thank you for your report & patch.
(and I'm sorry that I've not been active these days ;( )
This patch looks good to me and, IIUC, would be enough to fix this bug.
Reviewed-by: Daisuke Nishimura <nishimura@....nes.nec.co.jp>
BTW, it wouldn't cause any problem, I think former rcu_read_lock()/unlock()
in task_in_mem_cgroup() is unnecessary, because try_get_mem_cgroup_from_mm()
calls them for itself.
Thanks,
Daisuke Nishimura.
> ==
> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
>
> css_id() should be called under rcu_read_lock().
> Following is a report from Li Zefan.
> ==
> ===================================================
> [ INFO: suspicious rcu_dereference_check() usage. ]
> ---------------------------------------------------
> kernel/cgroup.c:4438 invoked rcu_dereference_check() without protection!
>
> other info that might help us debug this:
>
>
> rcu_scheduler_active = 1, debug_locks = 1
> 1 lock held by kswapd0/31:
> #0: (swap_lock){+.+.-.}, at: [<c05058bb>] swap_info_get+0x4b/0xd0
>
> stack backtrace:
> Pid: 31, comm: kswapd0 Not tainted 2.6.34-rc5-tip+ #13
> Call Trace:
> [<c083c5d6>] ? printk+0x1d/0x1f
> [<c0480744>] lockdep_rcu_dereference+0x94/0xb0
> [<c049d6ed>] css_id+0x5d/0x60
> [<c05165a5>] mem_cgroup_uncharge_swapcache+0x45/0xa0
> [<c0505e4f>] swapcache_free+0x3f/0x60
> [<c04e79e2>] __remove_mapping+0xb2/0xf0
> [<c04e7cbb>] shrink_page_list+0x26b/0x490
> [<c047f85d>] ? put_lock_stats+0xd/0x30
> [<c083fd67>] ? _raw_spin_unlock_irq+0x27/0x50
> [<c0482566>] ? trace_hardirqs_on_caller+0xb6/0x220
> [<c04e8158>] shrink_inactive_list+0x278/0x620
> [<c04729e1>] ? sched_clock_cpu+0x121/0x180
> [<c047e9b8>] ? trace_hardirqs_off_caller+0x18/0x130
> [<c047eadb>] ? trace_hardirqs_off+0xb/0x10
> [<c0843438>] ? sub_preempt_count+0x8/0x90
> [<c047f85d>] ? put_lock_stats+0xd/0x30
> [<c04e8704>] shrink_zone+0x204/0x3c0
> [<c083fcac>] ? _raw_spin_unlock+0x2c/0x50
> [<c04e951e>] kswapd+0x61e/0x7c0
> [<c04e6ed0>] ? isolate_pages_global+0x0/0x1f0
> [<c046bae0>] ? autoremove_wake_function+0x0/0x50
> [<c04e8f00>] ? kswapd+0x0/0x7c0
> [<c046b5e4>] kthread+0x74/0x80
> [<c046b570>] ? kthread+0x0/0x80
> [<c04035ba>] kernel_thread_helper+0x6/0x10
>
> And css_is_ancestor() should be called under rcu_read_lock().
>
>
> Reported-by: Li Zefan <lizf@...fujitsu.com>
> Cc: Daisuke Nishimura <nishimura@....nes.nec.co.jp>
> Cc: Balbir Singh <balbir@...ux.vnet.ibm.com>
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
> ---
> mm/memcontrol.c | 18 ++++++++++++++++--
> 1 file changed, 16 insertions(+), 2 deletions(-)
>
> Index: linux-2.6.34-rc5-mm1/mm/memcontrol.c
> ===================================================================
> --- linux-2.6.34-rc5-mm1.orig/mm/memcontrol.c
> +++ linux-2.6.34-rc5-mm1/mm/memcontrol.c
> @@ -838,10 +838,12 @@ int task_in_mem_cgroup(struct task_struc
> * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
> * hierarchy(even if use_hierarchy is disabled in "mem").
> */
> + rcu_read_lock();
> if (mem->use_hierarchy)
> ret = css_is_ancestor(&curr->css, &mem->css);
> else
> ret = (curr == mem);
> + rcu_read_unlock();
> css_put(&curr->css);
> return ret;
> }
> @@ -1360,9 +1362,13 @@ static int memcg_oom_wake_function(wait_
> * Both of oom_wait_info->mem and wake_mem are stable under us.
> * Then we can use css_is_ancestor without taking care of RCU.
> */
> + rcu_read_lock();
> if (!css_is_ancestor(&oom_wait_info->mem->css, &wake_mem->css) &&
> - !css_is_ancestor(&wake_mem->css, &oom_wait_info->mem->css))
> + !css_is_ancestor(&wake_mem->css, &oom_wait_info->mem->css)) {
> + rcu_read_unlock();
> return 0;
> + }
> + rcu_read_unlock();
>
> wakeup:
> return autoremove_wake_function(wait, mode, sync, arg);
> @@ -2401,7 +2407,9 @@ mem_cgroup_uncharge_swapcache(struct pag
>
> /* record memcg information */
> if (do_swap_account && swapout && memcg) {
> + rcu_read_lock();
> swap_cgroup_record(ent, css_id(&memcg->css));
> + rcu_read_unlock();
> mem_cgroup_get(memcg);
> }
> if (swapout && memcg)
> @@ -2458,8 +2466,10 @@ static int mem_cgroup_move_swap_account(
> {
> unsigned short old_id, new_id;
>
> + rcu_read_lock();
> old_id = css_id(&from->css);
> new_id = css_id(&to->css);
> + rcu_read_unlock();
>
> if (swap_cgroup_cmpxchg(entry, old_id, new_id) == old_id) {
> mem_cgroup_swap_statistics(from, false);
> @@ -4303,7 +4313,11 @@ static int is_target_pte_for_mc(struct v
> }
> /* Threre is a swap entry and a page doesn't exist or isn't charged */
> if (ent.val && !ret) {
> - if (css_id(&mc.from->css) == lookup_swap_cgroup(ent)) {
> + unsigned short id;
> + rcu_read_lock();
> + id = css_id(&mc.from->css);
> + rcu_read_unlock();
> + if (id == lookup_swap_cgroup(ent)) {
> ret = MC_TARGET_SWAP;
> if (target)
> target->ent = ent;
>
>
--
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