From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:48325) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZNHPM-0003gf-0U for qemu-devel@nongnu.org; Thu, 06 Aug 2015 05:24:17 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZNHPH-0004in-MB for qemu-devel@nongnu.org; Thu, 06 Aug 2015 05:24:15 -0400 Received: from [59.151.112.132] (port=18857 helo=heian.cn.fujitsu.com) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZNHPG-0004i7-MN for qemu-devel@nongnu.org; Thu, 06 Aug 2015 05:24:11 -0400 Message-ID: <55C327A5.9060802@cn.fujitsu.com> Date: Thu, 6 Aug 2015 17:23:49 +0800 From: Yang Hongyang MIME-Version: 1.0 References: <1438677044-13030-1-git-send-email-yanghy@cn.fujitsu.com> <1438677044-13030-10-git-send-email-yanghy@cn.fujitsu.com> <55C30AE6.8080909@redhat.com> <55C3187C.2050503@cn.fujitsu.com> <55C32448.1030107@redhat.com> In-Reply-To: <55C32448.1030107@redhat.com> Content-Type: text/plain; charset="windows-1252"; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH v4 09/11] netfilter: add a netbuffer filter List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jason Wang , qemu-devel@nongnu.org Cc: thuth@redhat.com, zhang.zhanghailiang@huawei.com, lizhijian@cn.fujitsu.com, eddie.dong@intel.com, mrhines@linux.vnet.ibm.com, stefanha@redhat.com On 08/06/2015 05:09 PM, Jason Wang wrote: > > > On 08/06/2015 04:19 PM, Yang Hongyang wrote: >> >> >> On 08/06/2015 03:21 PM, Jason Wang wrote: >>> >>> >>> On 08/04/2015 04:30 PM, Yang Hongyang wrote: >>>> This filter is to buffer/release packets, this feature can be used >>>> when using MicroCheckpointing, or other Remus like VM FT solutions, you >>>> can also use it to simulate the network delay. >>>> It has an interval option, if supplied, this filter will release >>>> packets by interval. >>>> >>>> Usage: >>>> -netdev tap,id=bn0 >>>> -netfilter buffer,id=f0,netdev=bn0,chain=in,interval=1000 >>>> >>>> NOTE: >>>> the scale of interval is microsecond. >>>> >>>> Signed-off-by: Yang Hongyang >>>> --- >>>> v4: remove bh >>>> pass the packet to next filter instead of receiver >>>> v3: check packet's sender and sender->peer when flush it >>>> --- >>>> net/Makefile.objs | 1 + >>>> net/filter-buffer.c | 120 >>>> ++++++++++++++++++++++++++++++++++++++++++++++++++++ >>>> net/filter.c | 2 + >>>> net/filters.h | 17 ++++++++ >>>> qapi-schema.json | 18 +++++++- >>>> 5 files changed, 157 insertions(+), 1 deletion(-) >>>> create mode 100644 net/filter-buffer.c >>>> create mode 100644 net/filters.h >>>> >>>> diff --git a/net/Makefile.objs b/net/Makefile.objs >>>> index 914aec0..5fa2f97 100644 >>>> --- a/net/Makefile.objs >>>> +++ b/net/Makefile.objs >>>> @@ -14,3 +14,4 @@ common-obj-$(CONFIG_SLIRP) += slirp.o >>>> common-obj-$(CONFIG_VDE) += vde.o >>>> common-obj-$(CONFIG_NETMAP) += netmap.o >>>> common-obj-y += filter.o >>>> +common-obj-y += filter-buffer.o >>>> diff --git a/net/filter-buffer.c b/net/filter-buffer.c >>>> new file mode 100644 >>>> index 0000000..fd467db >>>> --- /dev/null >>>> +++ b/net/filter-buffer.c >>>> @@ -0,0 +1,120 @@ >>>> +/* >>>> + * Copyright (c) 2015 FUJITSU LIMITED >>>> + * Author: Yang Hongyang >>>> + * >>>> + * This work is licensed under the terms of the GNU GPL, version 2 or >>>> + * later. See the COPYING file in the top-level directory. >>>> + */ >>>> + >>>> +#include "net/filter.h" >>>> +#include "net/queue.h" >>>> +#include "filters.h" >>>> +#include "qemu-common.h" >>>> +#include "qemu/timer.h" >>>> +#include "qemu/iov.h" >>>> + >>>> +typedef struct FILTERBUFFERState { >>>> + NetFilterState nf; >>>> + NetQueue *incoming_queue; >>>> + int64_t interval; >>> >>> Can interval be negative? >> >> No. >> >>> If not please uint. And not sure you really >>> need a 64 bit integer, if not, please use uint32_t. >> >> Ok, will use uint32_t instead. >> >>> >>>> + QEMUTimer release_timer; >>>> +} FILTERBUFFERState; >>> >>> Filter buffer is not abbreviation. So better name it as >>> FilterBufferState. >> >> Ok, thanks. >> >>> >>>> + >>>> +static void packet_send_completed(NetClientState *nc, ssize_t len) >>>> +{ >>>> + return; >>>> +} >>> >>> Why need a dummy sent cb? >> >> Need to work around with queue_append, if there's no sent_cb, >> queue_append >> will simply drop the packet...Even we provide a sent_cb param to >> receive_iov, >> this dummy might still be needed because sent_cb might be null. > > Dropping happens only when the number of queued packet exceeds queue > limitation and no sent_cb. Isn't this just what we want? Yes, you are right. > And like we've > discussed, we need track packet->sent_cb so it was not a problem? Ok, we can drop this dummy callback, thank you. > >> >>> >>>> + >>>> +static void filter_buffer_flush(NetFilterState *nf) >>>> +{ >>>> + FILTERBUFFERState *s = DO_UPCAST(FILTERBUFFERState, nf, nf); >>>> + NetQueue *queue = s->incoming_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->sender && packet->sender->peer) { >>>> + qemu_netfilter_pass_to_next(nf, packet->sender, >>>> packet->flags, >>>> + packet->data, packet->size); >>>> + } >>>> + >>>> + /* >>>> + * now that we pass the packet to next filter, we don't >>>> care the >>>> + * reture value here, because the filter layer or other filter >>>> + * will take care of this packet >>>> + */ >>>> + g_free(packet); >>> >>> This seems wrong, since packet could be queued into incoming queue. >>> Doing this may cause use after free. >> >> The incoming queue will make a copy of packet data when queued a packet. >> So I think it's ok to free this packet here...because this packet is >> alloced by this filter when calling qemu_net_queue_append_iov. >> > > You're right. I see. > >>> >>>> + } >>>> +} >>>> + >>>> +static void filter_buffer_release_timer(void *opaque) >>>> +{ >>>> + FILTERBUFFERState *s = opaque; >>>> + filter_buffer_flush(&s->nf); >>>> + timer_mod(&s->release_timer, >>>> + qemu_clock_get_us(QEMU_CLOCK_VIRTUAL) + s->interval); >>>> +} >>>> + >>>> +/* filter APIs */ >>>> +static ssize_t filter_buffer_receive_iov(NetFilterState *nf, >>>> + NetClientState *sender, >>>> + unsigned flags, >>>> + const struct iovec *iov, >>>> + int iovcnt) >>>> +{ >>>> + FILTERBUFFERState *s = DO_UPCAST(FILTERBUFFERState, nf, nf); >>>> + NetQueue *queue = s->incoming_queue; >>>> + >>>> + qemu_net_queue_append_iov(queue, sender, flags, iov, iovcnt, >>>> + packet_send_completed); >>>> + return iov_size(iov, iovcnt); >>> >>> So if interval is zero, packet will be blocked forever and memory will >>> be exhausted. >> >> Yes, but this supposed to be used by FT solutions, so it will >> be released periodically. Currently only used with interval makes sense. >> > > Then you need terminate the initialization when interval is zero. Ok, but I'd like to also write a comment with it to mention when there're users, the interval check can be dropped. > > [...] > . > -- Thanks, Yang.