All of lore.kernel.org
 help / color / mirror / Atom feed
From: Hailiang Zhang <zhang.zhanghailiang@huawei.com>
To: Zhang Chen <zhangchen.fnst@cn.fujitsu.com>, qemu-devel@nongnu.org
Cc: xuquan8@huawei.com
Subject: Re: [Qemu-devel] [PATCH 2/3] COLO-compare: Optimize colo_packet_compare_common
Date: Mon, 27 Feb 2017 15:28:37 +0800	[thread overview]
Message-ID: <58B3D525.6080104@huawei.com> (raw)
In-Reply-To: <31f0e89c-13c0-247a-1026-9fc28ee2d6f4@cn.fujitsu.com>

On 2017/2/27 15:03, Zhang Chen wrote:
>
>
> On 02/25/2017 02:58 PM, Hailiang Zhang wrote:
>> On 2017/2/25 11:32, Zhang Chen wrote:
>>> Add offset args for colo_packet_compare_common, optimize
>>> colo_packet_compare_icmp() and colo_packet_compare_udp()
>>> just compare the IP payload.
>>>
>>> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com>
>>> ---
>>>    net/colo-compare.c | 28 +++++++++++++++++++++-------
>>>    1 file changed, 21 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/net/colo-compare.c b/net/colo-compare.c
>>> index e75f0ae..9853232 100644
>>> --- a/net/colo-compare.c
>>> +++ b/net/colo-compare.c
>>> @@ -180,7 +180,7 @@ static int packet_enqueue(CompareState *s, int mode)
>>>     * return:    0  means packet same
>>>     *            > 0 || < 0 means packet different
>>>     */
>>> -static int colo_packet_compare_common(Packet *ppkt, Packet *spkt)
>>> +static int colo_packet_compare_common(Packet *ppkt, Packet *spkt,
>>> int offset)
>>>    {
>>>        trace_colo_compare_ip_info(ppkt->size,
>>> inet_ntoa(ppkt->ip->ip_src),
>>> inet_ntoa(ppkt->ip->ip_dst), spkt->size,
>>> @@ -188,7 +188,8 @@ static int colo_packet_compare_common(Packet
>>> *ppkt, Packet *spkt)
>>> inet_ntoa(spkt->ip->ip_dst));
>>>
>>>        if (ppkt->size == spkt->size) {
>>> -        return memcmp(ppkt->data, spkt->data, spkt->size);
>>> +        return memcmp(ppkt->data + offset, spkt->data + offset,
>>> +                      spkt->size - offset);
>>>        } else {
>>>            trace_colo_compare_main("Net packet size are not the same");
>>>            return -1;
>>> @@ -237,8 +238,7 @@ static int colo_packet_compare_tcp(Packet *spkt,
>>> Packet *ppkt)
>>>            spkt->ip->ip_sum = ppkt->ip->ip_sum;
>>>        }
>>>
>>> -    res = memcmp(ppkt->data + ETH_HLEN, spkt->data + ETH_HLEN,
>>> -                (spkt->size - ETH_HLEN));
>>> +    res = colo_packet_compare_common(ppkt, spkt, ETH_HLEN);
>>>
>>
>> For tcp packets check, why not ignore the ip headers, just like udp
>> packets check ?
>> Besides, here, can we compare the checksum stored in headers of tcp
>> and udp
>> before call colo_packet_compare_common(), which i think will improve
>> the comparing
>> performance.
>
> That's another way to compare the packet suggested by Dr. David Alan
> Gilbert,
> It makes two packets IP header be same firstly, then compare all IP packet.
> This way can tell people why we ignore IP header explicitly in other
> packets check like udp.
> For performance, If we ignore the IP header that will reduce at least 20
> byte not to be compared.
> So, ignore ip header have better comparing performance.
>

OK, here, i think we can re-use the checksum value stored in headers of tcp
or udp, comparing it first before comparing the complete payload of tcp or udp
packets is another way to improve the performance, it is only 2 bytes.

Thanks.

> Thanks
> Zhang Chen
>
>
>>
>> Thanks.
>> Hailiang
>>
>>>        if (res != 0 &&
>>> trace_event_get_state(TRACE_COLO_COMPARE_MISCOMPARE)) {
>>> trace_colo_compare_pkt_info_src(inet_ntoa(ppkt->ip->ip_src),
>>> @@ -277,7 +277,14 @@ static int colo_packet_compare_udp(Packet *spkt,
>>> Packet *ppkt)
>>>            return -1;
>>>        }
>>>
>>> -    ret = colo_packet_compare_common(ppkt, spkt);
>>> +    /*
>>> +     * Because of ppkt and spkt are both in the same connection,
>>> +     * The ppkt's src ip, dst ip, src port, dst port, ip_proto all are
>>> +     * same with spkt. In addition, IP header's Identification is a
>>> random
>>> +     * field, we can handle it in IP fragmentation function later.
>>> +     * So we just compare the ip payload here.
>>> +     */
>>> +    ret = colo_packet_compare_common(ppkt, spkt, network_length +
>>> ETH_HLEN);
>>>
>>>        if (ret) {
>>>            trace_colo_compare_udp_miscompare("primary pkt size",
>>> ppkt->size);
>>> @@ -304,7 +311,14 @@ static int colo_packet_compare_icmp(Packet
>>> *spkt, Packet *ppkt)
>>>            return -1;
>>>        }
>>>
>>> -    if (colo_packet_compare_common(ppkt, spkt)) {
>>> +    /*
>>> +     * Because of ppkt and spkt are both in the same connection,
>>> +     * The ppkt's src ip, dst ip, src port, dst port, ip_proto all are
>>> +     * same with spkt. In addition, IP header's Identification is a
>>> random
>>> +     * field, we can handle it in IP fragmentation function later.
>>> +     * So we just compare the ip payload here.
>>> +     */
>>> +    if (colo_packet_compare_common(ppkt, spkt, network_length +
>>> ETH_HLEN)) {
>>>            trace_colo_compare_icmp_miscompare("primary pkt size",
>>>                                               ppkt->size);
>>>            qemu_hexdump((char *)ppkt->data, stderr, "colo-compare",
>>> @@ -330,7 +344,7 @@ static int colo_packet_compare_other(Packet
>>> *spkt, Packet *ppkt)
>>> inet_ntoa(ppkt->ip->ip_dst), spkt->size,
>>> inet_ntoa(spkt->ip->ip_src),
>>> inet_ntoa(spkt->ip->ip_dst));
>>> -    return colo_packet_compare_common(ppkt, spkt);
>>> +    return colo_packet_compare_common(ppkt, spkt, 0);
>>>    }
>>>
>>>    static int colo_old_packet_check_one(Packet *pkt, int64_t *check_time)
>>>
>>
>>
>>
>>
>> .
>>
>

  reply	other threads:[~2017-02-27  7:28 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-25  3:32 [Qemu-devel] [PATCH 0/3] COLO-compare: Optimize the code and fix some bug Zhang Chen
2017-02-25  3:32 ` [Qemu-devel] [PATCH 1/3] COLO-compare: Add minimum packet size check and some fix Zhang Chen
2017-02-25  6:43   ` Hailiang Zhang
2017-02-27  6:39     ` Zhang Chen
2017-02-25  3:32 ` [Qemu-devel] [PATCH 1/3] COLO-compare: Add " Zhang Chen
2017-02-25  3:43   ` Zhang Chen
2017-02-28  3:22     ` Jason Wang
2017-02-28  3:27       ` Zhang Chen
2017-02-25  3:32 ` [Qemu-devel] [PATCH 2/3] COLO-compare: Optimize colo_packet_compare_common Zhang Chen
2017-02-25  6:58   ` Hailiang Zhang
2017-02-27  7:03     ` Zhang Chen
2017-02-27  7:28       ` Hailiang Zhang [this message]
2017-02-27  7:34         ` Zhang Chen
2017-02-27  8:43           ` Hailiang Zhang
2017-02-27  9:02             ` Zhang Chen
2017-02-25  7:26   ` Hailiang Zhang
2017-02-27  7:06     ` Zhang Chen
2017-02-25  3:32 ` [Qemu-devel] [PATCH 3/3] COLO-compare: Make qemu_hexdump() print data when trace-event ON Zhang Chen

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=58B3D525.6080104@huawei.com \
    --to=zhang.zhanghailiang@huawei.com \
    --cc=qemu-devel@nongnu.org \
    --cc=xuquan8@huawei.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.