DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Wiles, Keith" <keith.wiles@intel.com>
To: Shailja Pandey <csz168117@iitd.ac.in>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] How to update the sequence number in TCP packets using	DPDK-pktgen?
Date: Mon, 9 Oct 2017 17:12:45 +0000	[thread overview]
Message-ID: <99AABA1F-637E-45FA-8F38-3688749B048D@intel.com> (raw)
In-Reply-To: <c8dde074-7192-e7dd-b8b8-857f370e4649@iitd.ac.in>



> On Oct 9, 2017, at 8:42 AM, Shailja Pandey <csz168117@iitd.ac.in> wrote:
> 
> Hi,
> 
> In the DPDK application, I am maintaining some state and due to that I need to generate the packets with monotonically increasing sequence numbers. As shown in the code below,
> 
> /tip->tcp.seq        = htonl(DEFAULT_PKT_NUMBER);/
> 
> DPDK is putting some default number as a sequence number in the TCP packet and not using as defined by TCP protocol. I tried various workarounds but due to multi-threaded nature of the pktgen application, I am facing some issues and unable to put sequence numbers in increasing order. I tried thread_local and pthread_mutex_lock etc to generate packets with increasing sequence number.
> 
> I am not very sure what am I missing, Please help me in this matter.

The way pktgen was written was for performance, which means I setup the mbufs at start time and do not touch them again till another start command. This causes packets to be sent out of order as I do not know when mbufs are returned to the free pool.

The only way to send a given sequence is to modified each packet as it is sent, which will impact performance. The random mode does just this type of packet mods and it would be a good model to look at if you want to modify the code.

> 
> _Function:_
> 
> void
> pktgen_tcp_hdr_ctor(pkt_seq_t *pkt, tcpip_t *tip, int type __rte_unused)
> {
>         uint16_t tlen;
> 
>         /* Zero out the header space */
>         memset((char *)tip, 0, sizeof(tcpip_t));
> 
>         /* Create the TCP header */
>         tip->ip.src         = htonl(pkt->ip_src_addr.addr.ipv4.s_addr);
>         tip->ip.dst         = htonl(pkt->ip_dst_addr.addr.ipv4.s_addr);
>         tlen                = pkt->pktSize -
>                 (pkt->ether_hdr_size + sizeof(ipHdr_t));
> 
>         tip->ip.len         = htons(tlen);
>         tip->ip.proto       = pkt->ipProto;
> 
>         tip->tcp.sport      = htons(pkt->sport);
>         tip->tcp.dport      = htons(pkt->dport);
> 
> *tip->tcp.seq        = htonl(DEFAULT_PKT_NUMBER);*
> 
>         tip->tcp.ack        = htonl(DEFAULT_ACK_NUMBER);
>         tip->tcp.offset     = ((sizeof(tcpHdr_t) / sizeof(uint32_t)) << 4);     /* Offset in words */
>         tip->tcp.flags      = ACK_FLAG;                                         /* ACK */
>         tip->tcp.window     = htons(DEFAULT_WND_SIZE);
>         tip->tcp.urgent     = 0;
> 
>         tlen                = pkt->pktSize - pkt->ether_hdr_size;
> 
>         tip->tcp.cksum      = cksum(tip, tlen, 0);
> }
> 
> -- 
> 
> Thanks,
> Shailja
> 

Regards,
Keith

  reply	other threads:[~2017-10-09 17:12 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-09 13:42 Shailja Pandey
2017-10-09 17:12 ` Wiles, Keith [this message]
2017-10-10 23:00 ` Stephen Hemminger
2017-10-11  0:10   ` Wiles, Keith

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=99AABA1F-637E-45FA-8F38-3688749B048D@intel.com \
    --to=keith.wiles@intel.com \
    --cc=csz168117@iitd.ac.in \
    --cc=dev@dpdk.org \
    /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).