Re: [net-next:master 641/694] include/net/request_sock.h:198:9: sparse: context imbalance in 'tcp_conn_request' - unexpected unlock

[Eric Dumazet]

On Tue, 2015-10-06 at 00:14 +0800, kbuild test robot wrote:
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git 
> master
> head:   4560cdff03a76348ee5fae48e3c7914e4de2db5b
> commit: 7656d842de93fd2d2de7b403062cad757cadf1df [641/694] tcp: fix fastopen 
> races vs lockless listener
> reproduce:
>         # apt-get install sparse
>         git checkout 7656d842de93fd2d2de7b403062cad757cadf1df
>         make ARCH=x86_64 allmodconfig
>         make C=1 CF=-D__CHECK_ENDIAN__
> 
> 
> sparse warnings: (new ones prefixed by >>)
> 
> >> include/net/request_sock.h:198:9: sparse: context imbalance in 
> >> 'tcp_conn_request' - unexpected unlock
> 
> vim +/tcp_conn_request +198 include/net/request_sock.h
> 
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  182  }
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  183  
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  184  static inline void 
> reqsk_queue_add(struct request_sock_queue *queue,
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  185                             
>    struct request_sock *req,
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  186                             
>    struct sock *parent,
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  187                             
>    struct sock *child)
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  188  {
> fff1f300 Eric Dumazet             2015-10-02  189     
> spin_lock(&queue->rskq_lock);
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  190     req->sk = child;
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  191     
> sk_acceptq_added(parent);
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  192  
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  193     if 
> (queue->rskq_accept_head == NULL)
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  194             
> queue->rskq_accept_head = req;
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  195     else
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  196             
> queue->rskq_accept_tail->dl_next = req;
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  197  
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18 @198     queue->rskq_accept_tail 
> = req;
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  199     req->dl_next = NULL;
> fff1f300 Eric Dumazet             2015-10-02  200     
> spin_unlock(&queue->rskq_lock);
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  201  }
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  202  
> fff1f300 Eric Dumazet             2015-10-02  203  static inline struct 
> request_sock *reqsk_queue_remove(struct request_sock_queue *queue,
> fff1f300 Eric Dumazet             2015-10-02  204                             
>                       struct sock *parent)
> 0e87506f Arnaldo Carvalho de Melo 2005-06-18  205  {
> fff1f300 Eric Dumazet             2015-10-02  206     struct request_sock 
> *req;

Seems a false positive, or a rather useless message ?

I do not see any obvious problem in this code.

static inline void reqsk_queue_add(struct request_sock_queue *queue,
                                   struct request_sock *req,
                                   struct sock *parent,
                                   struct sock *child)
{
        spin_lock(&queue->rskq_lock);
        ...
        spin_unlock(&queue->rskq_lock);
}

Thanks.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html