DPDK usage discussions
 help / color / mirror / Atom feed
From: "Rosen, Rami" <rami.rosen@intel.com>
To: Sungho Hong <shong59@asu.edu>, "users@dpdk.org" <users@dpdk.org>
Subject: Re: [dpdk-users] locating the address of rte_mbuf payload
Date: Sat, 14 Jul 2018 16:24:17 +0000	[thread overview]
Message-ID: <9B0331B6EBBD0E4684FBFAEDA55776F967324849@HASMSX110.ger.corp.intel.com> (raw)
In-Reply-To: <CAKf1RwBCASS16CgmWwVo1dWM9uKgwEAFRYYwvjDaA4JVDrjT9g@mail.gmail.com>

Hi Sungho,

I would suggest two or three things which may shed more light on your issue:

1) How to you print the rte_mbuf packets ? is it with rte_pktmbuf_dump()? In case it is, please post here also the rte_mbuf metadata that you get  when calling rte_pktmbuf_dump() (for example, pkt_len, nb_segs, etc). You get this metadata as part of a regular rte_pktmbuf_dump() call. If you don’t use rte_pktmbuf_dump(), you can look in the following link for an example:
http://git.dpdk.org/dpdk/tree/test/test/test_mbuf.c

Regarding the eth header: you do print 
its length but not its DA/SA. can you also print the src address and the dest address and see if it makes sense ? 
You can use
print_ethaddr() for this; for example, see: http://git.dpdk.org/dpdk/tree/examples/l3fwd/main.c

Third thing is: if you will post here the snippet of code in which you craft the rte_mbuf it may clarify things. (if you do, please also let us know the size of PKT_SIZE that you use in your "message" struct.)

Regards,
Rami Rosen


-----Original Message-----
From: users [mailto:users-bounces@dpdk.org] On Behalf Of Sungho Hong
Sent: Saturday, July 14, 2018 08:46
To: users@dpdk.org
Subject: [dpdk-users] locating the address of rte_mbuf payload

Hello I have sent the packet with the data "hellohellohello*"
in a message struct using DPDK

    struct message{
    data[PKT_SIZE]
    }

I am sending this data from a client to server, and the server receives this data. the data received in the server can be viewed from the packet dump provided by DPDK. You can see the data I am sending "hello"s right at the bottom of the log

    00000000: A0 36 9F 83 AB BC 00 1B 21 A6 D4 D4 08 00 45 00 | .6......!.....E.
    00000010: 01 49 7C 60 00 00 80 11 C9 37 0A DA 6E 59 0A DA | .I|`.....7..nY..
    00000020: 6F FF 44 5C 44 5C 01 35 DF 27 7B 22 6E 61 6D 65 | o.D\D\.5.'{"name
    00000030: 73 70 61 63 65 73 22 3A 20 5B 31 33 37 30 34 36 | spaces":
[137046
    00000040: 36 31 39 33 2C 20 32 39 33 31 32 37 36 38 39 36 | 6193,
2931276896
    00000050: 2C 20 31 33 32 34 30 31 33 30 39 31 2C 20 31 33 | , 1324013091, 13
    00000060: 35 32 37 38 34 34 35 32 2C 20 31 34 33 33 32 37 | 52784452,
143327
    00000070: 32 31 36 36 2C 20 32 32 30 31 36 38 30 30 31 36 | 2166,
2201680016
    00000080: 2C 20 31 33 35 39 38 30 30 32 37 30 2C 20 31 34 | , 1359800270, 14
    00000090: 37 31 30 37 33 36 34 39 2C 20 31 39 37 34 38 36 | 71073649,
197486
    000000A0: 31 33 39 32 2C 20 32 32 33 38 33 34 32 39 32 38 | 1392,
2238342928
    000000B0: 2C 20 31 33 37 32 38 31 35 32 32 31 2C 20 33 30 | , 1372815221, 30
    000000C0: 37 33 32 34 31 37 36 30 2C 20 37 36 32 35 31 36 | 73241760,
762516
    000000D0: 39 35 35 2C 20 31 32 39 34 33 35 36 37 36 34 2C | 955, 1294356764,
    000000E0: 20 31 35 32 30 34 39 31 39 39 35 5D 2C 20 22 64 | 1520491995], "d
    000000F0: 69 73 70 6C 61 79 6E 61 6D 65 22 3A 20 22 22 2C |
isplayname": "",
    00000100: 20 22 76 65 72 73 69 6F 6E 22 3A 20 5B 32 2C 20 |  "version":
[2,
    00000110: 30 5D 2C 20 22 68 6F 73 74 5F 69 6E 74 22 3A 20 | 0],
"host_int":
    00000120: 32 30 39 37 36 33 35 34 38 32 33 39 38 34 38 32 |
2097635482398482
    00000130: 38 34 35 36 30 36 31 34 37 35 30 34 32 32 37 38 |
8456061475042278
    00000140: 33 35 34 37 35 37 32 2C 20 22 70 6F 72 74 22 3A | 3547572,
"port":
    00000150: 20 31 37 35 30 30 7D 68 65 6C 6C 6F 68 65 6C 6C | 17500}hellohell
    00000160: 6F 68 65 6C 6C 6F 68 65 6C 6C 6F 68 65 6C 6C 6F | ohellohellohello
    00000170: 68 65 6C 6C 6F 68 65 6C 6C 6F 68 65 6C 6C 6F 00 | hellohellohello.


The problem is I cannot locate the data offset by reading the rte_mbuf provided by dpdk. I need to extract the "hello"s using the correct method that is provided by DPDK, but I could not find a good way.

I have tried the size headers and the buf_address pointers to locate the location of the struct message, but I have no idea how to pin point the exact address of the data-offset using DPDK structures.


char *rtn = NULL;
rtn = rte_pktmbuf_mtod(m, char *); // points to the start of the data

struct ether_hdr *eth;
eth = rte_pktmbuf_mtod(m, struct ether_hdr *); // rtn+=sizeof(eth);

rtn = m->buf_addr;


if(rtn!=NULL) {
printf("length hdr: %ld\n", sizeof(struct ether_hdr)); printf("length eth: %ld\n", sizeof(eth)); printf("length m->buf_add: %ld\n", sizeof(m->buf_addr));

printf("length message: %ld\n", sizeof(struct message));

// 336
printf("rte_pktmbuf_mtod: %s\n", rtn);  // lenght of the offset: 456 rte_pktmbuf_dump(stdout, m, 1024); }


Would anyone familiar with DPDK, have a good answer for this?


Best

  reply	other threads:[~2018-07-14 16:24 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-07-14  5:45 Sungho Hong
2018-07-14 16:24 ` Rosen, Rami [this message]
2018-07-14 20:09   ` Sungho Hong

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=9B0331B6EBBD0E4684FBFAEDA55776F967324849@HASMSX110.ger.corp.intel.com \
    --to=rami.rosen@intel.com \
    --cc=shong59@asu.edu \
    --cc=users@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).