From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:42034) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dX1x7-0007om-Q2 for qemu-devel@nongnu.org; Mon, 17 Jul 2017 05:04:31 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dX1x4-00048p-J7 for qemu-devel@nongnu.org; Mon, 17 Jul 2017 05:04:29 -0400 Received: from [59.151.112.132] (port=6713 helo=heian.cn.fujitsu.com) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dX1x3-00047d-Km for qemu-devel@nongnu.org; Mon, 17 Jul 2017 05:04:26 -0400 References: <1499925175-21218-1-git-send-email-zhangchen.fnst@cn.fujitsu.com> <1499925175-21218-4-git-send-email-zhangchen.fnst@cn.fujitsu.com> <5468b372-eec3-1969-e927-236859a6b079@redhat.com> From: Zhang Chen Message-ID: <79fc29f3-b08f-c3c1-4006-3a0167010c2d@cn.fujitsu.com> Date: Mon, 17 Jul 2017 17:06:45 +0800 MIME-Version: 1.0 In-Reply-To: <5468b372-eec3-1969-e927-236859a6b079@redhat.com> Content-Type: text/plain; charset="utf-8"; format=flowed Content-Transfer-Encoding: 8bit Content-Language: en-US Subject: Re: [Qemu-devel] [PATCH V2 3/4] net/colo-compare.c: Optimize unpredictable tcp options comparison List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jason Wang , qemu devel Cc: zhangchen.fnst@cn.fujitsu.com, zhanghailiang , Li Zhijian On 07/14/2017 11:33 AM, Jason Wang wrote: > > > On 2017年07月13日 13:52, Zhang Chen wrote: >> When network is busy, some tcp options(like sack) will unpredictable >> occur in primary side or secondary side. it will make packet size >> not same, but the two packet's payload is identical. colo just >> care about packet payload, so we skip the option field. > > A question is, if SACK were not same, does it mean e.g some packet > were lost just for primary or secondary? If yes, we will be out of > sync soon. Is it really better to delay the checkpoint here? The SACK is designed to optimize TCP fast retransmit, but in COLO situation, this TCP options field will make COLO-compare trigger checkpoint frequently, and we use normal TCP ACK to do retransmit job will get better performance. In the worst situation, some skipped TCP options will make primary and secondary send different packet after that, it will trigger checkpoint very soon. So, I think we no need care the SACK field in COLO situation. Thanks Zhang Chen > > Thanks > >> >> Signed-off-by: Zhang Chen >> --- >> net/colo-compare.c | 34 +++++++++++++++++++++++++--------- >> 1 file changed, 25 insertions(+), 9 deletions(-) >> >> diff --git a/net/colo-compare.c b/net/colo-compare.c >> index 2caeb80..6406c4a 100644 >> --- a/net/colo-compare.c >> +++ b/net/colo-compare.c >> @@ -183,7 +183,10 @@ 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, >> int offset) >> +static int colo_packet_compare_common(Packet *ppkt, >> + Packet *spkt, >> + int poffset, >> + int soffset) >> { >> if (trace_event_get_state(TRACE_COLO_COMPARE_MISCOMPARE)) { >> char pri_ip_src[20], pri_ip_dst[20], sec_ip_src[20], >> sec_ip_dst[20]; >> @@ -198,9 +201,10 @@ static int colo_packet_compare_common(Packet >> *ppkt, Packet *spkt, int offset) >> sec_ip_src, sec_ip_dst); >> } >> - if (ppkt->size == spkt->size) { >> - return memcmp(ppkt->data + offset, spkt->data + offset, >> - spkt->size - offset); >> + if (ppkt->size == spkt->size || poffset != soffset) { >> + return memcmp(ppkt->data + poffset, >> + spkt->data + soffset, >> + spkt->size - soffset); >> } else { >> trace_colo_compare_main("Net packet size are not the same"); >> return -1; >> @@ -263,12 +267,22 @@ static int colo_packet_compare_tcp(Packet >> *spkt, Packet *ppkt) >> * so we just need skip this field. >> */ >> if (ptcp->th_off > 5) { >> - ptrdiff_t tcp_offset; >> - tcp_offset = ppkt->transport_header - (uint8_t *)ppkt->data >> + ptrdiff_t ptcp_offset, stcp_offset; >> + >> + ptcp_offset = ppkt->transport_header - (uint8_t *)ppkt->data >> + (ptcp->th_off * 4); >> - res = colo_packet_compare_common(ppkt, spkt, tcp_offset); >> + stcp_offset = spkt->transport_header - (uint8_t *)spkt->data >> + + (stcp->th_off * 4); >> + >> + /* >> + * When network is busy, some tcp options(like sack) will >> unpredictable >> + * occur in primary side or secondary side. it will make >> packet size >> + * not same, but the two packet's payload is identical. colo >> just >> + * care about packet payload, so we skip the option field. >> + */ >> + res = colo_packet_compare_common(ppkt, spkt, ptcp_offset, >> stcp_offset); >> } else if (ptcp->th_sum == stcp->th_sum) { >> - res = colo_packet_compare_common(ppkt, spkt, ETH_HLEN); >> + res = colo_packet_compare_common(ppkt, spkt, ETH_HLEN, >> ETH_HLEN); >> } else { >> res = -1; >> } >> @@ -328,6 +342,7 @@ static int colo_packet_compare_udp(Packet *spkt, >> Packet *ppkt) >> * the ip payload here. >> */ >> ret = colo_packet_compare_common(ppkt, spkt, >> + network_header_length + ETH_HLEN, >> network_header_length + >> ETH_HLEN); >> if (ret) { >> @@ -365,6 +380,7 @@ static int colo_packet_compare_icmp(Packet *spkt, >> Packet *ppkt) >> * the ip payload here. >> */ >> if (colo_packet_compare_common(ppkt, spkt, >> + network_header_length + ETH_HLEN, >> network_header_length + >> ETH_HLEN)) { >> trace_colo_compare_icmp_miscompare("primary pkt size", >> ppkt->size); >> @@ -402,7 +418,7 @@ static int colo_packet_compare_other(Packet >> *spkt, Packet *ppkt) >> sec_ip_src, sec_ip_dst); >> } >> - return colo_packet_compare_common(ppkt, spkt, 0); >> + return colo_packet_compare_common(ppkt, spkt, 0, 0); >> } >> static int colo_old_packet_check_one(Packet *pkt, int64_t >> *check_time) > > > > . > -- Thanks Zhang Chen