RE: [kbuild-all] Re: drivers/vhost/net.c:1010 vhost_net_rx_peek_head_len() error: double unlocked 'sk->sk_receive_queue.lock' (orig line 1002)

[Xia, Hui]

>-----Original Message-----
>From: Michael S. Tsirkin <mst@xxxxxxxxxx>
>Sent: 2020年8月10日 20:34
>To: kernel test robot <lkp@xxxxxxxxx>
>Cc: Jason Wang <jasowang@xxxxxxxxxx>; kbuild-all@xxxxxxxxxxxx; linux-
>kernel@xxxxxxxxxxxxxxx
>Subject: [kbuild-all] Re: drivers/vhost/net.c:1010 vhost_net_rx_peek_head_len()
>error: double unlocked 'sk->sk_receive_queue.lock' (orig line 1002)
>
>On Sat, Aug 08, 2020 at 03:37:07PM +0800, kernel test robot wrote:
>> Hi Jason,
>>
>> First bad commit (maybe != root cause):
>>
>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
>> head:   449dc8c97089a6e09fb2dac4d92b1b7ac0eb7c1e
>> commit: 20c384f1ea1a0bc7320bc445c72dd02d2970d594 vhost: refine vhost
>and vringh kconfig
>> date:   4 months ago
>> config: arm-randconfig-m031-20200808 (attached as .config)
>> compiler: arm-linux-gnueabi-gcc (GCC) 9.3.0
>>
>> If you fix the issue, kindly add following tag as appropriate
>> Reported-by: kernel test robot <lkp@xxxxxxxxx>
>>
>> New smatch warnings:
>> drivers/vhost/net.c:1010 vhost_net_rx_peek_head_len() error: double
>> unlocked 'sk->sk_receive_queue.lock' (orig line 1002)
>> drivers/vhost/net.c:1010 vhost_net_rx_peek_head_len() error: double
>> unlocked 'sk->sk_receive_queue.lock' (orig line 1002)
>
>Weird. Can't figure out where the double unlock is ...
Please ignore this report. It is a false positive. 
We have changed the rule to avoid send out this warning type directly. Sorry for inconvenient.

>
>
>> Old smatch warnings:
>> drivers/vhost/net.c:1528 vhost_net_set_backend() warn: passing a valid pointer
>to 'PTR_ERR'
>
>this stumped me too...
Ignore this one too. It is old warnings.

Thanks,
Hui
>
>> vim +1010 drivers/vhost/net.c
>>
>> 8dd014adfea6f17 David Stevens   2010-07-27   994
>> be294a51adfc1e1 Toshiaki Makita 2018-07-03   995  static int
>vhost_net_rx_peek_head_len(struct vhost_net *net, struct sock *sk,
>> be294a51adfc1e1 Toshiaki Makita 2018-07-03   996
>	      bool *busyloop_intr)
>> 030881372460654 Jason Wang      2016-03-04   997  {
>> 28b9b33b983f4de Toshiaki Makita 2018-07-03   998  	struct
>vhost_net_virtqueue *rnvq = &net->vqs[VHOST_NET_VQ_RX];
>> 28b9b33b983f4de Toshiaki Makita 2018-07-03   999  	struct
>vhost_net_virtqueue *tnvq = &net->vqs[VHOST_NET_VQ_TX];
>> 6369fec5be0aad4 Toshiaki Makita 2018-07-03  1000  	struct vhost_virtqueue
>*rvq = &rnvq->vq;
>> 28b9b33b983f4de Toshiaki Makita 2018-07-03  1001  	struct vhost_virtqueue
>*tvq = &tnvq->vq;
>> 28b9b33b983f4de Toshiaki Makita 2018-07-03 @1002  	int len =
>peek_head_len(rnvq, sk);
>> 030881372460654 Jason Wang      2016-03-04  1003
>> dc151282bbdaed0 Tonghao Zhang   2018-09-25  1004  	if (!len && rvq-
>>busyloop_timeout) {
>> f5a4941aa6d190e Jason Wang      2018-05-29  1005  		/* Flush
>batched heads first */
>> 09c3248938c3e3b Jason Wang      2018-07-20  1006
>	vhost_net_signal_used(rnvq);
>> 030881372460654 Jason Wang      2016-03-04  1007  		/* Both tx vq
>and rx socket were polled here */
>> dc151282bbdaed0 Tonghao Zhang   2018-09-25  1008
>	vhost_net_busy_poll(net, rvq, tvq, busyloop_intr, true);
>> 030881372460654 Jason Wang      2016-03-04  1009
>> 28b9b33b983f4de Toshiaki Makita 2018-07-03 @1010  		len =
>peek_head_len(rnvq, sk);
>> 030881372460654 Jason Wang      2016-03-04  1011  	}
>> 030881372460654 Jason Wang      2016-03-04  1012
>> 030881372460654 Jason Wang      2016-03-04  1013  	return len;
>> 030881372460654 Jason Wang      2016-03-04  1014  }
>> 030881372460654 Jason Wang      2016-03-04  1015
>>
>> :::::: The code at line 1010 was first introduced by commit
>> :::::: 28b9b33b983f4de3ce9e660e3efe1e08adabf779 vhost_net: Rename
>> local variables in vhost_net_rx_peek_head_len
>>
>> :::::: TO: Toshiaki Makita <makita.toshiaki@xxxxxxxxxxxxx>
>> :::::: CC: David S. Miller <davem@xxxxxxxxxxxxx>
>>
>> ---
>> 0-DAY CI Kernel Test Service, Intel Corporation
>> https://lists.01.org/hyperkitty/list/kbuild-all@xxxxxxxxxxxx
>
>_______________________________________________
>kbuild-all mailing list -- kbuild-all@xxxxxxxxxxxx To unsubscribe send an email to
>kbuild-all-leave@xxxxxxxxxxxx