linux-kernel - Re: [PATCH v3] mm: memcontrol: Don't flood OOM messages with no eligible task.

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 18 Oct 2018 19:37:18 +0900
From:   Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>
To:     Michal Hocko <mhocko@...nel.org>
Cc:     Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Johannes Weiner <hannes@...xchg.org>, linux-mm@...ck.org,
        syzkaller-bugs@...glegroups.com, guro@...com,
        kirill.shutemov@...ux.intel.com, linux-kernel@...r.kernel.org,
        rientjes@...gle.com, yang.s@...baba-inc.com,
        Andrew Morton <akpm@...ux-foundation.org>,
        Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        syzbot <syzbot+77e6b28a7a7106ad0def@...kaller.appspotmail.com>
Subject: Re: [PATCH v3] mm: memcontrol: Don't flood OOM messages with no
 eligible task.

On 2018/10/18 15:55, Michal Hocko wrote:
> On Thu 18-10-18 11:46:50, Tetsuo Handa wrote:
>> This is essentially a ratelimit approach, roughly equivalent with:
>>
>>   static DEFINE_RATELIMIT_STATE(oom_no_victim_rs, 60 * HZ, 1);
>>   oom_no_victim_rs.flags |= RATELIMIT_MSG_ON_RELEASE;
>>
>>   if (__ratelimit(&oom_no_victim_rs)) {
>>     dump_header(oc, NULL);
>>     pr_warn("Out of memory and no killable processes...\n");
>>     oom_no_victim_rs.begin = jiffies;
>>   }
> 
> Then there is no reason to reinvent the wheel. So use the standard
> ratelimit approach. Or put it in other words, this place is no special
> to any other that needs some sort of printk throttling. We surely do not
> want an ad-hoc solutions all over the kernel.

netdev_wait_allrefs() in net/core/dev.c is doing the same thing. Since
out_of_memory() is serialized by oom_lock mutex, there is no need to use
"struct ratelimit_state"->lock field. Plain "unsigned long" is enough.

> 
> And once you realize that the ratelimit api is the proper one (put aside
> any potential improvements in the implementation of this api) then you
> quickly learn that we already do throttle oom reports and it would be
> nice to unify that and ... we are back to a naked patch. So please stop
> being stuborn and try to cooperate finally.

I don't think that ratelimit API is the proper one, for I am touching
"struct ratelimit_state"->begin field which is not exported by ratelimit API.
But if you insist on ratelimit API version, I can tolerate with below one.

 mm/oom_kill.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index f10aa53..7c6118e 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -1106,6 +1106,12 @@ bool out_of_memory(struct oom_control *oc)
 	select_bad_process(oc);
 	/* Found nothing?!?! */
 	if (!oc->chosen) {
+		static DEFINE_RATELIMIT_STATE(no_eligible_rs, 60 * HZ, 1);
+
+		ratelimit_set_flags(&no_eligible_rs, RATELIMIT_MSG_ON_RELEASE);
+		if ((is_sysrq_oom(oc) || is_memcg_oom(oc)) &&
+		    !__ratelimit(&no_eligible_rs))
+			return false;
 		dump_header(oc, NULL);
 		pr_warn("Out of memory and no killable processes...\n");
 		/*
@@ -1115,6 +1121,7 @@ bool out_of_memory(struct oom_control *oc)
 		 */
 		if (!is_sysrq_oom(oc) && !is_memcg_oom(oc))
 			panic("System is deadlocked on memory\n");
+		no_eligible_rs.begin = jiffies;
 	}
 	if (oc->chosen && oc->chosen != (void *)-1UL)
 		oom_kill_process(oc, !is_memcg_oom(oc) ? "Out of memory" :
-- 
1.8.3.1