【OVS2.5.0源码分析】normal action精确流表生成和刷新过程分析

最新推荐文章于 2022-11-22 07:16:54 发布

one_clouder

最新推荐文章于 2022-11-22 07:16:54 发布

阅读量5.8k

点赞数 1

分类专栏： OVS2.5.0源码分析文章标签：网络 OVS openvswitch

本文链接：https://blog.csdn.net/one_clouder/article/details/52186661

版权

本文详细探讨了OVS2.5.0中，首包到达时如何基于目的MAC生成精确流表规则，并在后续过程中如何通过upcall_xlate函数和revalidate线程实现流表的刷新，确保网络流量的正确转发。

摘要由CSDN通过智能技术生成

一、首包精确流表生成

首包到达OVS交换机时，尚未建立基于目的mac的流表规则，需要upcall到用户态进行学习，此时生成的规则是把报文flood到其他端口。

1、upcall_xlate函数

    upcall->dump_seq = seq_read(udpif->dump_seq);
    upcall->reval_seq = seq_read(udpif->reval_seq);
    xlate_actions(&xin, &upcall->xout); 
    upcall->xout_initialized = true;

2、xlate_actions函数

<pre name="code" class="cpp">   /* The bridge is now known so obtain its table version. */
    ctx.tables_version = ofproto_dpif_get_tables_version(ctx.xbridge->ofproto);

    if (!xin->ofpacts && !ctx.rule) {
        ctx.rule = rule_dpif_lookup_from_table(       //检索流表
            ctx.xbridge->ofproto, ctx.tables_version, flow, xin->wc,
            ctx.xin->resubmit_stats, &ctx.table_id,
            flow->in_port.ofp_port, true, true);
        if (ctx.xin->resubmit_stats) {
            rule_dpif_credit_stats(ctx.rule, ctx.xin->resubmit_stats);
        }
        if (ctx.xin->xcache) {
            struct xc_entry *entry;

            entry = xlate_cache_add_entry(ctx.xin->xcache, XC_RULE);
            entry->u.rule = ctx.rule;
            rule_dpif_ref(ctx.rule);
        }

        if (OVS_UNLIKELY(ctx.xin->resubmit_hook)) {
            ctx.xin->resubmit_hook(ctx.xin, ctx.rule, 0);
        }
    }
    xout->fail_open = ctx.rule && rule_dpif_is_fail_open(ctx.rule);

    /* Get the proximate input port of the packet.  (If xin->recirc,
     * flow->in_port is the ultimate input port of the packet.) */
    struct xport *in_port = get_ofp_port(xbridge,
                                         ctx.base_flow.in_port.ofp_port);

    /* Tunnel stats only for non-recirculated packets. */
    if (!xin->recirc && in_port && in_port->is_tunnel) {
        if (ctx.xin->resubmit_stats) {
            netdev_vport_inc_rx(in_port->netdev, ctx.xin->resubmit_stats);
            if (in_port->bfd) {
                bfd_account_rx(in_port->bfd, ctx.xin->resubmit_stats);
            }
        }
        if (ctx.xin->xcache) {
            struct xc_entry *entry;

            entry = xlate_cache_add_entry(ctx.xin->xcache, XC_NETDEV);
            entry->u.dev.rx = netdev_ref(in_port->netdev);
            entry->u.dev.bfd = bfd_ref(in_port->bfd);
        }
    }

    if (!xin->recirc && process_special(&ctx, in_port)) {
        /* process_special() did all the processing for this packet.
         *
         * We do not perform special processing on recirculated packets, as
         * recirculated packets are not really received by the bridge.*/
    } else if (in_port && in_port->xbundle
               && xbundle_mirror_out(xbridge, in_port->xbundle)) {
        if (ctx.xin->packet != NULL) {
            static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
            VLOG_WARN_RL(&rl, "bridge %s: dropping packet received on port "
                         "%s, which is reserved exclusively for mirroring",
                         ctx.xbridge->name, in_port->xbundle->name);
        }
    } else {
        /* Sampling is done only for packets really received by th