Re: [Qemu-devel] [RFC PATCH V2 2/3] filter-rewriter: track connection and parse packet

[Zhang Chen]

On 07/04/2016 11:04 AM, Jason Wang wrote:
> On 2016年07月02日 14:22, Zhang Chen wrote:
> > We use colo-base.h to track connection and parse packet
> >
> > Signed-off-by: Zhang Chen <address@hidden>
> > Signed-off-by: Li Zhijian <address@hidden>
> > Signed-off-by: Wen Congyang <address@hidden>
> > ---
> >   net/filter-rewriter.c | 52 
> > +++++++++++++++++++++++++++++++++++++++++++++++++++
> >   1 file changed, 52 insertions(+)
> >
> > diff --git a/net/filter-rewriter.c b/net/filter-rewriter.c
> > index 08b015d..c38ab24 100644
> > --- a/net/filter-rewriter.c
> > +++ b/net/filter-rewriter.c
> > @@ -50,6 +50,20 @@ typedef struct RewriterState {
> >       uint32_t hashtable_size;
> >   } RewriterState;
> >   +/*
> > + * Return 1 on success, if return 0 means the pkt
> > + * is not TCP packet
> > + */
> > +static int is_tcp_packet(Packet *pkt)
> > +{
> > +    if (!parse_packet_early(pkt) &&
> > +        pkt->ip->ip_p == IPPROTO_TCP) {
> > +        return 1;
> > +    } else {
> > +        return 0;
> > +    }
> > +}
> > +
> >   static ssize_t colo_rewriter_receive_iov(NetFilterState *nf,
> >                                            NetClientState *sender,
> >                                            unsigned flags,
> > @@ -57,11 +71,49 @@ static ssize_t 
> > colo_rewriter_receive_iov(NetFilterState *nf,
> >                                            int iovcnt,
> >                                            NetPacketSent *sent_cb)
> >   {
> > +    RewriterState *s = FILTER_COLO_REWRITER(nf);
> > +    Connection *conn;
> > +    ConnectionKey key = {{ 0 } };
> > +    Packet *pkt;
> > +    ssize_t size = iov_size(iov, iovcnt);
> > +    char *buf = g_malloc0(size);
> > +
> > +    iov_to_buf(iov, iovcnt, 0, buf, size);
> > +    pkt = packet_new(buf, size);
> > +
> >       /*
> >        * if we get tcp packet
> >        * we will rewrite it to make secondary guest's
> >        * connection established successfully
> >        */
> > +    if (is_tcp_packet(pkt)) {
> > +        if (sender == nf->netdev) {
> > +            fill_connection_key(pkt, &key, SECONDARY);
> > +        } else {
> > +            fill_connection_key(pkt, &key, PRIMARY);
> > +        }
> > +
> > +        conn = connection_get(s->connection_track_table,
> > +                              &key,
> > +                              &s->hashtable_size);
> > +        if (!conn->processing) {
> > +            qemu_mutex_lock(&s->conn_list_lock);
> > +            g_queue_push_tail(&s->conn_list, conn);
> > +            qemu_mutex_unlock(&s->conn_list_lock);
>
> conn_list was never used in this series, and I fail to understand why 
> conn_list is needed?

When filter-rewriter work with colo-frame we need this do some 
synchronization job.
like colo do checkpoint,we will flush connection.
Maybe I should move this lock to other patch (like filter-rewriter work 
with colo-frame).

Thanks
Zhang Chen

> > +            conn->processing = true;
> > +        }
> > +
> > +        if (sender == nf->netdev) {
> > +            /* NET_FILTER_DIRECTION_TX */
> > +            /* handle_primary_tcp_pkt */
> > +        } else {
> > +            /* NET_FILTER_DIRECTION_RX */
> > +            /* handle_secondary_tcp_pkt */
> > +        }
> > +    }
> > +
> > +    packet_destroy(pkt, NULL);
> > +    pkt = NULL;
> >       return 0;
> >   }
>
>
>
> .

--
Thanks
zhangchen