DPDK usage discussions
 help / color / mirror / Atom feed
From: "Abdullah Ömer Yamaç" <aomeryamac@gmail.com>
To: Thomas Monjalon <thomas@monjalon.net>
Cc: users@dpdk.org
Subject: Re: Direct Mem Pool vs Indirect mem pool creation
Date: Sat, 2 Sep 2023 00:30:20 +0300	[thread overview]
Message-ID: <CA+sj1i-3q1RKx+kY=qfXE=NReHZq2+4sEt6viuFzQYDsMP=zUw@mail.gmail.com> (raw)
In-Reply-To: <CA+sj1i-J8P59EHUFjf0S4T9hxenwFoiN0L_P571OF4wEfRng6w@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 2253 bytes --]

I found the answer in the comments: "/* Creates a shallow copy of mbuf */"
for the rte_pktmbuf_clone. Thanks

On Fri, Sep 1, 2023 at 4:31 PM omer yamac <aomeryamac@gmail.com> wrote:

> Thank you for the clarification. Now, I understand more clearly. May I ask
> one more question? I used the rte_pktmbuf_clone function to duplicate the
> packet. Then, the new packet is cloned to the indirect memory. After the
> clone, I used the RTE_MBUF_DIRECT macro to check this, which returns false.
> So, to get a direct buffer, I need to use the alloc function and copy the
> data. Am I right?
>
> I just want to use the "static inline int rte_vlan_insert(struct rte_mbuf
> **m)" function; a direct buffer is necessary for this function.
>
> On Fri, Sep 1, 2023 at 4:08 PM Thomas Monjalon <thomas@monjalon.net>
> wrote:
>
>> 01/09/2023 14:47, omer yamac:
>> > Hello,
>> >
>> > I need clarification while creating direct/indirect buffers for mbuf. I
>> > couldn't find exact documentation, and I just looked over the
>> fragmentation
>> > test case and saw that two pools were created. One is a
>> > direct pool, and the other is an indirect pool. Here are the methods to
>> > create pools:
>> > direct_pool = rte_pktmbuf_pool_create("FRAG_D_MBUF_POOL",
>> >                           NUM_MBUFS, BURST, 0,
>> >                           RTE_MBUF_DEFAULT_BUF_SIZE,
>> >                           SOCKET_ID_ANY);
>> > indirect_pool = rte_pktmbuf_pool_create("FRAG_I_MBUF_POOL",
>> >                         NUM_MBUFS, BURST, 0,
>> >                         0, SOCKET_ID_ANY)
>> >
>> > I couldn't see the exact difference. Just the "data_room_size"
>> parameter is
>> > different. If this parameter is 0, then is the pool indirect?
>>
>> A pool is neither direct or indirect, it is just a pool of buffers
>> with a defined size for all buffers of a pool.
>> You are free to create any pool for your needs.
>>
>> Now if you create a pool of buffers with size 0,
>> we can expect you will save some data elsewhere,
>> using rte_pktmbuf_attach_extbuf() for instance.
>>
>> More explanations can be found in the doc:
>>
>> https://doc.dpdk.org/guides/prog_guide/mbuf_lib.html#direct-and-indirect-buffers
>>
>>
>>

[-- Attachment #2: Type: text/html, Size: 3167 bytes --]

      reply	other threads:[~2023-09-01 21:30 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-01 12:47 omer yamac
2023-09-01 13:08 ` Thomas Monjalon
2023-09-01 13:31   ` omer yamac
2023-09-01 21:30     ` Abdullah Ömer Yamaç [this message]

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='CA+sj1i-3q1RKx+kY=qfXE=NReHZq2+4sEt6viuFzQYDsMP=zUw@mail.gmail.com' \
    --to=aomeryamac@gmail.com \
    --cc=thomas@monjalon.net \
    --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).