Re: [Qemu-devel] [PATCH 09/12] netfilter: add a netbuffer filter

[Jason Wang]

On 07/30/2015 02:47 PM, Yang Hongyang wrote:
> On 07/30/2015 01:13 PM, Jason Wang wrote:
> [...]
>>> +
>>> +#include "net/filter.h"
>>> +#include "net/queue.h"
>>> +#include "filters.h"
>>> +#include "qemu-common.h"
>>> +#include "qemu/error-report.h"
>>> +
>>> +typedef struct FILTERBUFFERState {
>>> +    NetFilterState nf;
>>> +    NetClientState dummy; /* used to send buffered packets */
>>
>> Why need this? Couldn't we just infer this from NetFilterState?
>
> Because we use existing API qemu_send_packet_async/raw to send
> packet, it takes an NetClientState as the first argument sender,
> and use sender->peer->incoming_queue as the dest queue, so in order to
> make this API work, we need to use this dummy NC and init it's
> peer to our dest(which is the network backend)
> Another way is to call qemu_net_queue_send(netdev->incoming_queue,...)
> directly, we still need a NetClientState *sender param, can not
> use NetFilterState.

I think this is my meaning. Use NetFilterState->netdev.

> This dummy NC also been checked in filter_buffer_receive to avoid
> buffering
> packet been sent by ourself.
>

I don't get why this is needed. Who is going to queue a packet in dummy
NC, consider it was not peered by any others?
>>
>>> +    NetQueue *incoming_queue;
>>> +    NetQueue *inflight_queue;
>>> +} FILTERBUFFERState;
>>> +
>>> +static void packet_send_completed(NetClientState *nc, ssize_t len)
>>> +{
>>> +    return;
>>> +}
>>> +
>>> +static void filter_buffer_flush(NetFilterState *nf)
>>> +{
>>> +    FILTERBUFFERState *s = DO_UPCAST(FILTERBUFFERState, nf, nf);
>>> +    NetQueue *queue = s->inflight_queue;
>>> +    NetPacket *packet;
>>> +
>>> +    while (queue && !QTAILQ_EMPTY(&queue->packets)) {
>>> +        packet = QTAILQ_FIRST(&queue->packets);
>>> +        QTAILQ_REMOVE(&queue->packets, packet, entry);
>>> +        queue->nq_count--;
>>> +
>>> +        if (packet->flags & QEMU_NET_PACKET_FLAG_RAW) {
>>> +            qemu_send_packet_raw(&s->dummy, packet->data,
>>> packet->size);
>>> +        } else {
>>> +            qemu_send_packet_async(&s->dummy, packet->data,
>>> packet->size,
>>> +                                   packet->sent_cb);
>>> +        }
>>> +
>>> +        /*
>>> +         * now that we pass the packet to
>>> sender->peer->incoming_queue, we
>>> +         * don't care the reture value here, because the peer's
>>> queue will
>>> +         * take care of this packet
>>> +         */
>>> +        g_free(packet);
>>> +    }
>>> +
>>> +    if (queue && QTAILQ_EMPTY(&queue->packets)) {
>>
>> Under what condition queue->packet could be not empty here?
>
> You're right, this check is nonsense. can be dropped.
>
>> And I don't
>> get the point that why a inflight_queue is needed.
>
> maybe squash the interval patch here will make it clear?
>

Yes, maybe.

>>
>>> +        g_free(queue);
>>> +        s->inflight_queue = NULL;
>>> +    }
>>> +}
>>> +
>>> +/* filter APIs */
>>> +static ssize_t filter_buffer_receive(NetFilterState *nf,
>>> +                                     NetClientState *sender,
>>> +                                     unsigned flags,
>>> +                                     const uint8_t *data,
>>> +                                     size_t size)
>>> +{
>>> +    FILTERBUFFERState *s = DO_UPCAST(FILTERBUFFERState, nf, nf);
>>> +    NetQueue *queue = s->incoming_queue;
>>> +
>>> +    if (!sender->info) {
>>> +        /* This must be a dummy NetClientState, do nothing */
>>> +        return 0;
>>> +    }
>>> +
>>> +    if (sender->info->type == NET_CLIENT_OPTIONS_KIND_NIC) {
>>> +        /* we only buffer guest output packets */
>>> +        qemu_net_queue_append(queue, sender, flags, data, size,
>>> +                              packet_send_completed);
>>
>> This may brings some confusion for user. Since the name is 'netbuffer'.
>> Maybe we can change the filter to be ingress or out or both? Like:
>>
>> -device virtio-net-pci,id=virtio0
>> -netfilter buffer,id=filter0,dev=virtio0,interval=1000,type=out
>>
>> Then we can just try to enqueue the packet when virtio-net-pci is
>> sender?
>>
>>> +        /* Now that we have buffered the packet, return sucess */
>>> +        return size;
>>> +    }
>>> +
>>> +    return 0;
>>> +}
>>> +
>>> +static void filter_buffer_cleanup(NetFilterState *nf)
>>> +{
>>> +    FILTERBUFFERState *s = DO_UPCAST(FILTERBUFFERState, nf, nf);
>>> +
>>> +    /* flush inflight packets */
>>> +    filter_buffer_flush(nf);
>>> +    /* flush incoming packets */
>>> +    s->inflight_queue = s->incoming_queue;
>>> +    s->incoming_queue = NULL;
>>> +    filter_buffer_flush(nf);
>>> +
>>> +    return;
>>> +}
>>> +
>>> +
>>> +static NetFilterInfo net_filter_buffer_info = {
>>> +    .type = NET_FILTER_OPTIONS_KIND_BUFFER,
>>> +    .size = sizeof(FILTERBUFFERState),
>>> +    .receive = filter_buffer_receive,
>>> +    .cleanup = filter_buffer_cleanup,
>>> +};
>>> +
>>> +int net_init_filter_buffer(const NetFilterOptions *opts, const char
>>> *name,
>>> +                           NetClientState *netdev, Error **errp)
>>> +{
>>> +    NetFilterState *nf;
>>> +    FILTERBUFFERState *s;
>>> +
>>> +    assert(opts->kind == NET_FILTER_OPTIONS_KIND_BUFFER);
>>> +
>>> +    nf = qemu_new_net_filter(&net_filter_buffer_info, netdev,
>>> "buffer", name);
>>> +    s = DO_UPCAST(FILTERBUFFERState, nf, nf);
>>> +    /*
>>> +     * we need the dummy NetClientState to send packets in order to
>>> avoid
>>> +     * receive packets again.
>>> +     * we are buffering guest output packets, our buffered packets
>>> should be
>>> +     * sent to real network backend, so dummy's peer should be that
>>> backend.
>>> +     */
>>> +    s->dummy.peer = netdev;
>>> +    s->incoming_queue = qemu_new_net_queue(nf);
>>> +
>>> +    return 0;
>>> +}
>>> diff --git a/net/filter.c b/net/filter.c
>>> index 50fb837..e741e2a 100644
>>> --- a/net/filter.c
>>> +++ b/net/filter.c
>>> @@ -18,6 +18,7 @@
>>>
>>>   #include "net/filter.h"
>>>   #include "net/net.h"
>>> +#include "filters.h"
>>>
>>>   static QTAILQ_HEAD(, NetFilterState) net_filters;
>>>
>>> @@ -152,6 +153,7 @@ typedef int (NetFilterInit)(const
>>> NetFilterOptions *opts,
>>>
>>>   static
>>>   NetFilterInit * const
>>> net_filter_init_fun[NET_FILTER_OPTIONS_KIND_MAX] = {
>>> +    [NET_FILTER_OPTIONS_KIND_BUFFER] = net_init_filter_buffer,
>>>   };
>>>
>>>   static int net_filter_init1(const NetFilter *netfilter, Error **errp)
>>> diff --git a/net/filters.h b/net/filters.h
>>> new file mode 100644
>>> index 0000000..6c249b8
>>> --- /dev/null
>>> +++ b/net/filters.h
>>> @@ -0,0 +1,17 @@
>>> +/*
>>> + * Copyright (c) 2015 FUJITSU LIMITED
>>> + *
>>> + * This work is licensed under the terms of the GNU GPL, version 2 or
>>> + * later.  See the COPYING file in the top-level directory.
>>> + */
>>> +
>>> +#ifndef QEMU_NET_FILTERS_H
>>> +#define QEMU_NET_FILTERS_H
>>> +
>>> +#include "net/net.h"
>>> +#include "net/filter.h"
>>> +
>>> +int net_init_filter_buffer(const NetFilterOptions *opts, const char
>>> *name,
>>> +                           NetClientState *netdev, Error **errp);
>>> +
>>> +#endif /* QEMU_NET_FILTERS_H */
>>> diff --git a/qapi-schema.json b/qapi-schema.json
>>> index 1fc6390..67e00a0 100644
>>> --- a/qapi-schema.json
>>> +++ b/qapi-schema.json
>>> @@ -2577,6 +2577,16 @@
>>>   { 'command': 'netfilter_del', 'data': {'id': 'str'} }
>>>
>>>   ##
>>> +# @NetFilterBufferOptions
>>> +#
>>> +# a netbuffer filter for network backend.
>>> +#
>>> +# Since 2.5
>>> +##
>>> +{ 'struct': 'NetFilterBufferOptions',
>>> +  'data': { } }
>>> +
>>> +##
>>>   # @NetFilterOptions
>>>   #
>>>   # A discriminated record of network filters.
>>> @@ -2585,7 +2595,8 @@
>>>   #
>>>   ##
>>>   { 'union': 'NetFilterOptions',
>>> -  'data': { } }
>>> +  'data': {
>>> +    'buffer':     'NetFilterBufferOptions'} }
>>>
>>>   ##
>>>   # @NetFilter
>>
>> .
>>
>