netdev - Re: [PATCH net-next v2] ip_tunnel: Add dst_cache management lwtunnel

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:   Fri, 22 Feb 2019 12:14:01 +0800
From:   wenxu <wenxu@...oud.cn>
To:     David Ahern <dsahern@...il.com>, davem@...emloft.net,
        netdev@...r.kernel.org
Subject: Re: [PATCH net-next v2] ip_tunnel: Add dst_cache management
 lwtunnel_state of ip tunnel


On 2/22/2019 11:39 AM, David Ahern wrote:
> On 2/21/19 10:00 PM, wenxu@...oud.cn wrote:
>> From: wenxu <wenxu@...oud.cn>
>>
>> The lwtunnel_state is not init the dst_cache Which make the
>> ip_md_tunnel_xmit can't use the dst_cache. It will lookup
>> route table every packets.
>>
>> Signed-off-by: wenxu <wenxu@...oud.cn>
>> ---
>>  net/ipv4/ip_tunnel_core.c | 14 ++++++++++++++
>>  1 file changed, 14 insertions(+)
>>
>> diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
>> index 9a0e67b..f91ef7a 100644
>> --- a/net/ipv4/ip_tunnel_core.c
>> +++ b/net/ipv4/ip_tunnel_core.c
>> @@ -252,6 +252,12 @@ static int ip_tun_build_state(struct nlattr *attr,
>>  
>>  	tun_info = lwt_tun_info(new_state);
>>  
>> +	err = dst_cache_init(&tun_info->dst_cache, GFP_ATOMIC);
> build_state is called with rtnl held. Unless I am missing something, you
> don't need ATOMIC here.

If set GFP_KERNEL shows following:

Feb 22 07:07:03 A1_V5_02 kernel: BUG: sleeping function called from invalid context at kernel/locking/mutex.c:1301
Feb 22 07:07:03 A1_V5_02 kernel: in_atomic(): 1, irqs_disabled(): 0, pid: 7950, name: ip
Feb 22 07:07:03 A1_V5_02 kernel: CPU: 21 PID: 7950 Comm: ip Kdump: loaded Not tainted 5.0.0-rc6-next-20190215 #14
Feb 22 07:07:03 A1_V5_02 kernel: Hardware name: Inspur SA5112M5/YZMB-00870-101, BIOS 4.0.3 05/22/2018
Feb 22 07:07:03 A1_V5_02 kernel: Call Trace:
Feb 22 07:07:03 A1_V5_02 kernel: dump_stack+0x5a/0x73
Feb 22 07:07:03 A1_V5_02 kernel: ___might_sleep+0xfc/0x120
Feb 22 07:07:03 A1_V5_02 kernel: mutex_lock_killable+0x1c/0x42
Feb 22 07:07:03 A1_V5_02 kernel: pcpu_alloc+0x390/0x6f0
Feb 22 07:07:03 A1_V5_02 kernel: ? __nla_parse+0xf8/0x160
Feb 22 07:07:03 A1_V5_02 kernel: dst_cache_init+0x1d/0x40
Feb 22 07:07:03 A1_V5_02 kernel: ip_tun_build_state+0x75/0x140
Feb 22 07:07:03 A1_V5_02 kernel: lwtunnel_build_state+0x92/0xf0

Feb 22 07:07:03 A1_V5_02 kernel: fib_create_info+0x74e/0x1020


build_state in the rcu_read_lock and disable the preempt

 rcu_read_lock();
    ops = rcu_dereference(lwtun_encaps[encap_type]);
    if (likely(ops && ops->build_state && try_module_get(ops->owner))) {
        found = true;
        ret = ops->build_state(encap, family, cfg, lws, extack);
        if (ret)
            module_put(ops->owner);
    }  
    rcu_read_unlock();