From: Hailiang Zhang <zhang.zhanghailiang@huawei.com>
To: Jason Wang <jasowang@redhat.com>, qemu-devel@nongnu.org
Cc: xiecl.fnst@cn.fujitsu.com, zhangchen.fnst@cn.fujitsu.com,
lizhijian@cn.fujitsu.com, quintela@redhat.com,
yunhong.jiang@intel.com, eddie.dong@intel.com,
peter.huangpeng@huawei.com, dgilbert@redhat.com,
arei.gonglei@huawei.com, stefanha@redhat.com,
amit.shah@redhat.com, hongyang.yang@easystack.cn
Subject: Re: [Qemu-devel] [PATCH COLO-Frame v13 34/39] net/filter-buffer: Add default filter-buffer for each netdev
Date: Wed, 20 Jan 2016 15:14:38 +0800 [thread overview]
Message-ID: <569F33DE.2080809@huawei.com> (raw)
In-Reply-To: <569EF351.2020608@redhat.com>
On 2016/1/20 10:39, Jason Wang wrote:
>
>
> On 01/19/2016 04:39 PM, Hailiang Zhang wrote:
>> Hi Jason,
>>
>> Thanks for your review.
>>
>> On 2016/1/19 11:19, Jason Wang wrote:
>>>
>>>
>>> On 12/29/2015 03:09 PM, zhanghailiang wrote:
>>>> We add each netdev (except vhost-net) a default filter-buffer,
>>>> which will be used for COLO or Micro-checkpoint to buffer VM's packets.
>>>> The name of default filter-buffer is 'nop'.
>>>> For the default filter-buffer, it will not buffer any packets in
>>>> default.
>>>> So it has no side effect for the netdev.
>>>>
>>>> Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
>>>> Cc: Jason Wang <jasowang@redhat.com>
>>>> Cc: Yang Hongyang <hongyang.yang@easystack.cn>
>>>
>>> This patch did three things:
>>>
>>> 1) the ability to enable or disable a netfilter
>>> 2) the ability to add a default filter
>>> 3) default filter attaching for filter-buffer
>>>
>>> Better to split them into separate small patches.
>>>
>>> And several questions:
>>>
>>> For 1), I'm not sure this is real needed, we can in fact disable a
>>> filter by removing it.
>>
>> If we do like this, do we also need to _enable_ the buffer filter by
>> add it dynamically instead of attaching the default filter ?
>> Just like what we do in V10 ?
>> (In that series, you think have a default filter may be better.
>> The main reason for that is to support
>> hot-add nic. Since we didn't support hot-add nic during COLO,
>> it will be OK to add default filter dynamically)
>
> Aha, right. So rethinking of this, enabling/disabling a filter looks ok
> to me.
>
>>
>>> For 2), Instead of a specific code just for filter buffer, I think we
>>> need a generic method for an arbitrary filter to be used as default.
>>
>> Good idea.
>>
>>> And if we can achieve 2), 3) is not needed any more.
>>>
>>>> ---
>>>> v12:
>>>> - Skip vhost-net when add default filter
>>>> - Don't go through filter layer if the filter is disabled.
>>>> v11:
>>>> - New patch
>>>> ---
>>>> include/net/filter.h | 10 +++++++
>>>> net/filter-buffer.c | 82
>>>> ++++++++++++++++++++++++++++++++++++++++++++++++++++
>>>> net/filter.c | 6 +++-
>>>> net/net.c | 12 ++++++++
>>>> 4 files changed, 109 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/include/net/filter.h b/include/net/filter.h
>>>> index 2deda36..40aa38c 100644
>>>> --- a/include/net/filter.h
>>>> +++ b/include/net/filter.h
>>>> @@ -56,6 +56,8 @@ struct NetFilterState {
>>>> NetClientState *netdev;
>>>> NetFilterDirection direction;
>>>> char info_str[256];
>>>> + bool is_default;
>>>> + bool enabled;
>>>> QTAILQ_ENTRY(NetFilterState) next;
>>>> };
>>>>
>>>> @@ -74,4 +76,12 @@ ssize_t
>>>> qemu_netfilter_pass_to_next(NetClientState *sender,
>>>> int iovcnt,
>>>> void *opaque);
>>>>
>>>> +static inline bool qemu_need_skip_netfilter(NetFilterState *nf)
>>>> +{
>>>> + return nf->enabled ? false : true;
>>>> +}
>>>> +
>>>> +void netdev_add_default_filter_buffer(const char *netdev_id,
>>>> + NetFilterDirection direction,
>>>> + Error **errp);
>>>> #endif /* QEMU_NET_FILTER_H */
>>>> diff --git a/net/filter-buffer.c b/net/filter-buffer.c
>>>> index 57be149..9cf3544 100644
>>>> --- a/net/filter-buffer.c
>>>> +++ b/net/filter-buffer.c
>>>> @@ -14,6 +14,13 @@
>>>> #include "qapi/qmp/qerror.h"
>>>> #include "qapi-visit.h"
>>>> #include "qom/object.h"
>>>> +#include "net/net.h"
>>>> +#include "qapi/qmp/qdict.h"
>>>> +#include "qapi/qmp-output-visitor.h"
>>>> +#include "qapi/qmp-input-visitor.h"
>>>> +#include "monitor/monitor.h"
>>>> +#include "qmp-commands.h"
>>>> +#include "net/vhost_net.h"
>>>>
>>>> #define TYPE_FILTER_BUFFER "filter-buffer"
>>>>
>>>> @@ -102,6 +109,7 @@ static void filter_buffer_cleanup(NetFilterState
>>>> *nf)
>>>> static void filter_buffer_setup(NetFilterState *nf, Error **errp)
>>>> {
>>>> FilterBufferState *s = FILTER_BUFFER(nf);
>>>> + char *path = object_get_canonical_path_component(OBJECT(nf));
>>>>
>>>> /*
>>>> * We may want to accept zero interval when VM FT solutions
>>>> like MC
>>>> @@ -114,6 +122,14 @@ static void filter_buffer_setup(NetFilterState
>>>> *nf, Error **errp)
>>>> }
>>>>
>>>> s->incoming_queue =
>>>> qemu_new_net_queue(qemu_netfilter_pass_to_next, nf);
>>>> + nf->is_default = !strcmp(path, "nop");
>>>> + /*
>>>> + * For the default buffer filter, it will be disabled by default,
>>>> + * So it will not buffer any packets.
>>>> + */
>>>> + if (nf->is_default) {
>>>> + nf->enabled = false;
>>>> + }
>>>> if (s->interval) {
>>>> timer_init_us(&s->release_timer, QEMU_CLOCK_VIRTUAL,
>>>> filter_buffer_release_timer, nf);
>>>> @@ -163,6 +179,72 @@ out:
>>>> error_propagate(errp, local_err);
>>>> }
>>>>
>>>> +/*
>>>> +* This will be used by COLO or MC FT, for which they will need
>>>> +* to buffer the packets of VM's net devices, Here we add a default
>>>> +* buffer filter for each netdev. The name of default buffer filter is
>>>> +* 'nop'
>>>> +*/
>>>> +void netdev_add_default_filter_buffer(const char *netdev_id,
>>>> + NetFilterDirection direction,
>>>> + Error **errp)
>>>> +{
>>>
>>> Need a more generic way to add an arbitrary filter as default. E.g
>>> during netdev init, query if there's a default and do the initialization
>>> there.
>>>
>>
>> We call it in net_client_init1(), i don't find a better place to call it,
>> what's your suggestion ?
>
> net_client_init1() is ok, just need a generic way. E.g a string which
> stores the default filter and its parameters which could be queried by
> default filter initialization function.
>
Hmm, i'm still confused about how to realize this, do you mean change this helper
to a more generic function ? Just like :
void netdev_add_filter(const char *netdev_id,
const char *filter_type, char *id,
Error **errp)
Could you please give me more detail about how to realize it ? :)
>>
>>>> + QmpOutputVisitor *qov;
>>>> + QmpInputVisitor *qiv;
>>>> + Visitor *ov, *iv;
>>>> + QObject *obj = NULL;
>>>> + QDict *qdict;
>>>> + void *dummy = NULL;
>>>> + const char *id = "nop";
>>>> + char *queue = g_strdup(NetFilterDirection_lookup[direction]);
>>>> + NetClientState *nc = qemu_find_netdev(netdev_id);
>>>> + Error *err = NULL;
>>>> +
>>>> + /* FIXME: Not support multiple queues */
>>>> + if (!nc || nc->queue_index > 1) {
>>>> + g_free(queue);
>>>> + return;
>>>> + }
>>>> + /* Not support vhost-net */
>>>> + if (get_vhost_net(nc)) {
>>>> + g_free(queue);
>>>> + return;
>>>> + }
>>>> + qov = qmp_output_visitor_new();
>>>> + ov = qmp_output_get_visitor(qov);
>>>> + visit_start_struct(ov, &dummy, NULL, NULL, 0, &err);
>>>> + if (err) {
>>>> + goto out;
>>>> + }
>>>> + visit_type_str(ov, &nc->name, "netdev", &err);
>>>> + if (err) {
>>>> + goto out;
>>>> + }
>>>> + visit_type_str(ov, &queue, "queue", &err);
>>>> + if (err) {
>>>> + goto out;
>>>> + }
>>>> + visit_end_struct(ov, &err);
>>>> + if (err) {
>>>> + goto out;
>>>> + }
>>>> + obj = qmp_output_get_qobject(qov);
>>>> + g_assert(obj != NULL);
>>>> + qdict = qobject_to_qdict(obj);
>>>> + qmp_output_visitor_cleanup(qov);
>>>> +
>>>> + qiv = qmp_input_visitor_new(obj);
>>>> + iv = qmp_input_get_visitor(qiv);
>>>> + object_add(TYPE_FILTER_BUFFER, id, qdict, iv, &err);
>>>> + qmp_input_visitor_cleanup(qiv);
>>>> + qobject_decref(obj);
>>>> +out:
>>>> + g_free(queue);
>>>> + if (err) {
>>>> + error_propagate(errp, err);
>>>> + }
>>>> +}
>>>> +
>>>> static void filter_buffer_init(Object *obj)
>>>> {
>>>> object_property_add(obj, "interval", "int",
>>>> diff --git a/net/filter.c b/net/filter.c
>>>> index 1365bad..0b1e408 100644
>>>> --- a/net/filter.c
>>>> +++ b/net/filter.c
>>>> @@ -163,7 +163,8 @@ static void netfilter_complete(UserCreatable
>>>> *uc, Error **errp)
>>>> }
>>>>
>>>> nf->netdev = ncs[0];
>>>> -
>>>> + nf->is_default = false;
>>>> + nf->enabled = true;
>>>
>>> If we really need something like enabled, need more code for user to
>>> control this bit.
>>>
>>
>> OK, i will fix it.
>
> Or if you want to minimize the change set. You can make something like
> enabled locally to filter buffer, and let it to be used by colo only.
>
I have convert it to use object property to set/get the value, is that OK ?
Thanks,
Hailiang
>>
>>>> if (nfc->setup) {
>>>> nfc->setup(nf, &local_err);
>>>> if (local_err) {
>>>> @@ -190,6 +191,9 @@ static void netfilter_complete(UserCreatable
>>>> *uc, Error **errp)
>>>> g_free(info);
>>>> }
>>>> object_property_iter_free(iter);
>>>> + info = g_strdup_printf(",status=%s", nf->enabled ? "on" : "off");
>>>> + g_strlcat(nf->info_str, info, sizeof(nf->info_str));
>>>> + g_free(info);
>>>> }
>>>>
>>>> static void netfilter_finalize(Object *obj)
>>>> diff --git a/net/net.c b/net/net.c
>>>> index 87dd356..fd53cfc 100644
>>>> --- a/net/net.c
>>>> +++ b/net/net.c
>>>> @@ -581,6 +581,10 @@ static ssize_t
>>>> filter_receive_iov(NetClientState *nc,
>>>> NetFilterState *nf = NULL;
>>>>
>>>> QTAILQ_FOREACH(nf, &nc->filters, next) {
>>>> + /* Don't go through filter if it is off */
>>>> + if (qemu_need_skip_netfilter(nf)) {
>>>> + continue;
>>>> + }
>>>
>>> Better move the above to qemu_netfilter_receive()
>>>
>>
>> OK.
>>
>>>> ret = qemu_netfilter_receive(nf, direction, sender, flags,
>>>> iov,
>>>> iovcnt, sent_cb);
>>>> if (ret) {
>>>> @@ -1028,6 +1032,14 @@ static int net_client_init1(const void
>>>> *object, int is_netdev, Error **errp)
>>>> }
>>>> return -1;
>>>> }
>>>> +
>>>> + if (is_netdev) {
>>>> + const Netdev *netdev = object;
>>>> +
>>>> + netdev_add_default_filter_buffer(netdev->id,
>>>> + NET_FILTER_DIRECTION_RX,
>>>> + errp);
>>>
>>> Several issues here:
>>>
>>> - Generic code should know nothing about a specific kind of netfilter.
>>> So the default filter should be pointed out by user instead of
>>> hard-coding it like this.
>>> - Need to deal with the netdev hot add. It should still has the default
>>> filter.
>>>
>>
>> OK, then, here just answer my above question, we still need to add the
>> default filter,
>> so we still need the ability to enable or disable a netfilter. :)
>
> Right. So the issue is just to have a way to let an arbitrary filter to
> be default :).
>
> Thanks
>
>>
>> Thanks,
>> Hailiang
>>
>>>> + }
>>>> return 0;
>>>> }
>>>>
>>>
>>>
>>> .
>>>
>>
>>
>>
>
>
> .
>
next prev parent reply other threads:[~2016-01-20 7:15 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-29 7:08 [Qemu-devel] [PATCH COLO-Frame v13 00/39] COarse-grain LOck-stepping(COLO) Virtual Machines for Non-stop Service (FT) zhanghailiang
2015-12-29 7:08 ` [Qemu-devel] [PATCH COLO-Frame v13 01/39] configure: Add parameter for configure to enable/disable COLO support zhanghailiang
2015-12-29 7:08 ` [Qemu-devel] [PATCH COLO-Frame v13 02/39] migration: Introduce capability 'x-colo' to migration zhanghailiang
2015-12-29 7:08 ` [Qemu-devel] [PATCH COLO-Frame v13 03/39] COLO: migrate colo related info to secondary node zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 04/39] migration: Export migrate_set_state() zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 05/39] migration: Add state records for migration incoming zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 06/39] migration: Integrate COLO checkpoint process into migration zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 07/39] migration: Integrate COLO checkpoint process into loadvm zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 08/39] migration: Rename the'file' member of MigrationState zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 09/39] COLO/migration: Create a new communication path from destination to source zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 10/39] COLO: Implement colo checkpoint protocol zhanghailiang
2016-01-29 13:08 ` Dr. David Alan Gilbert
2016-01-30 8:51 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 11/39] COLO: Add a new RunState RUN_STATE_COLO zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 12/39] QEMUSizedBuffer: Introduce two help functions for qsb zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 13/39] COLO: Save PVM state to secondary side when do checkpoint zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 14/39] ram: Split host_from_stream_offset() into two helper functions zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 15/39] COLO: Load PVM's dirty pages into SVM's RAM cache temporarily zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 16/39] ram/COLO: Record the dirty pages that SVM received zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 17/39] COLO: Load VMState into qsb before restore it zhanghailiang
2016-01-04 19:00 ` Dr. David Alan Gilbert
2016-01-11 1:16 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 18/39] COLO: Flush PVM's cached RAM into SVM's memory zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 19/39] COLO: Add checkpoint-delay parameter for migrate-set-parameters zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 20/39] COLO: synchronize PVM's state to SVM periodically zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 21/39] COLO failover: Introduce a new command to trigger a failover zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 22/39] COLO failover: Introduce state to record failover process zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 23/39] COLO: Implement failover work for Primary VM zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 24/39] COLO: Implement failover work for Secondary VM zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 25/39] qmp event: Add COLO_EXIT event to notify users while exited from COLO zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 26/39] COLO failover: Shutdown related socket fd when do failover zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 27/39] COLO failover: Don't do failover during loading VM's state zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 28/39] COLO: Process shutdown command for VM in COLO state zhanghailiang
2016-01-26 19:55 ` Dr. David Alan Gilbert
2016-01-27 9:54 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 29/39] COLO: Update the global runstate after going into colo state zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 30/39] savevm: Split load vm state function qemu_loadvm_state zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 31/39] savevm: Introduce two helper functions for save/find loadvm_handlers entry zhanghailiang
2016-01-26 19:59 ` Dr. David Alan Gilbert
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 32/39] COLO: Separate the process of saving/loading ram and device state zhanghailiang
2016-01-27 14:14 ` Dr. David Alan Gilbert
2016-01-30 10:23 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 33/39] COLO: Split qemu_savevm_state_begin out of checkpoint process zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 34/39] net/filter-buffer: Add default filter-buffer for each netdev zhanghailiang
2016-01-11 1:26 ` Hailiang Zhang
2016-01-19 1:46 ` Hailiang Zhang
2016-01-19 3:19 ` Jason Wang
2016-01-19 8:39 ` Hailiang Zhang
2016-01-20 2:39 ` Jason Wang
2016-01-20 7:14 ` Hailiang Zhang [this message]
2016-01-20 9:15 ` Jason Wang
2016-01-20 9:27 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 35/39] filter-buffer: Accept zero interval zhanghailiang
2016-01-19 3:21 ` Jason Wang
2016-01-19 8:40 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 36/39] filter-buffer: Introduce a helper function to enable/disable default filter zhanghailiang
2016-01-19 3:35 ` Jason Wang
2016-01-19 8:44 ` Hailiang Zhang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 37/39] filter-buffer: Introduce a helper function to release packets zhanghailiang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 38/39] colo: Use default buffer-filter to buffer and " zhanghailiang
2016-01-19 3:59 ` Jason Wang
2015-12-29 7:09 ` [Qemu-devel] [PATCH COLO-Frame v13 39/39] COLO: Add block replication into colo process zhanghailiang
2015-12-29 7:14 ` [Qemu-devel] [PATCH COLO-Frame v13 00/39] COarse-grain LOck-stepping(COLO) Virtual Machines for Non-stop Service (FT) Hailiang Zhang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=569F33DE.2080809@huawei.com \
--to=zhang.zhanghailiang@huawei.com \
--cc=amit.shah@redhat.com \
--cc=arei.gonglei@huawei.com \
--cc=dgilbert@redhat.com \
--cc=eddie.dong@intel.com \
--cc=hongyang.yang@easystack.cn \
--cc=jasowang@redhat.com \
--cc=lizhijian@cn.fujitsu.com \
--cc=peter.huangpeng@huawei.com \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
--cc=stefanha@redhat.com \
--cc=xiecl.fnst@cn.fujitsu.com \
--cc=yunhong.jiang@intel.com \
--cc=zhangchen.fnst@cn.fujitsu.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).