netdev - [bridge] sane minimal values for multicast_query_interval/multicast_startup_query

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>] [day] [month] [year] [list]

Message-ID: <e8b9ce41-57b9-b6e2-a46a-ff9c791cf0ba@gmail.com>
Date:   Tue, 21 Dec 2021 05:48:34 -0800
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
Cc:     netdev@...r.kernel.org, edumazet@...gle.com
Subject: [bridge] sane minimal values for
 multicast_query_interval/multicast_startup_query_interval

Hi Nikolay

I was looking at a syzbot report, crashing a host simply by creating

hundreds of bridge devices, with IFLA_BR_MCAST_QUERY_INTVL set to 0.

For each bridge device, br_multicast_send_query() is invoked 1000 times 
per second (on HZ=1000 build) and machine eventually crashes hard.

What would be a reasonable minimal value for 
multicast_query_interval/multicast_startup_query_interval,

and should we enforce some kind of global rate limit to avoid future 
syzbot reports ?

INFO: rcu detected stall in br_multicast_query_expired

rcu: INFO: rcu_preempt self-detected stall on CPU
rcu:    0-...!: (1 GPs behind) idle=6eb/1/0x4000000000000000 
softirq=14214/14215 fqs=16
         (t=10500 jiffies g=15505 q=102699)
rcu: rcu_preempt kthread starved for 9059 jiffies! g15505 f0x0 
RCU_GP_WAIT_FQS(5) ->state=0x0 ->cpu=1
rcu:    Unless rcu_preempt kthread gets sufficient CPU time, OOM is now 
expected behavior.
rcu: RCU grace-period kthread stack dump:
task:rcu_preempt     state:R  running task     stack:28688 pid: 14 
ppid:     2 flags:0x00004000