DPDK usage discussions
 help / color / mirror / Atom feed
From: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
To: Kumaraparameshwaran Rathnavel <krath@cloudsimple.com>, users@dpdk.org
Subject: Re: [dpdk-users] DPDK Memory Allocation
Date: Mon, 25 Sep 2017 10:52:57 +0100	[thread overview]
Message-ID: <e057837d-22ad-f6da-e8e9-7a342b5401d6@intel.com> (raw)
In-Reply-To: <FB9452AE-E2A5-47C0-8FFC-AFF4143AA325@cloudsimple.com>

Hi,

On 22/09/2017 18:09, Kumaraparameshwaran Rathnavel wrote:
> Hi All,
>
> We have been using DPDK for quite sometime. First let me tell my assumption on how the DPDK works. For the DPDK memory allocation we give size in terms of MB. When this size is given the rte_eal_memory probes for the sysctl files and figures the starting Physical address of the huge pages. Depending on the size given some number of huge pages should be available. So whenever the memory is not contiguous different segments are formed and each segment is contiguous. The Physical addresses are memory mapped and heap regions are formed and rte_malloc returns the memory from this Heap.

Correct.

> We give 2GB to our application and the rte_eal initialisation takes place successfully. But in the application the rte_malloc fails. This happens randomly When we reboot the system we don’t see this issue. So my doubt is why does this happen, DPDK memory initialisation is successful but the memory allocation fails. We are sure that application has not consumed 2GB of memory.  Is there any condition that we should reboot the system for every run of application that uses DPDK.Should we do anything when we restart the application say some cleanup after the first run.

You give 2GB to the application, is it 2GB total of 2MB size huge pages?
Are there other applications using huge pages?
Do you set cgroup or quota limit on the number of pages to be allocated 
by the application?

Thanks,
Sergio

> Thanking You,
> Param.

      reply	other threads:[~2017-09-25  9:53 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-09-22 17:09 Kumaraparameshwaran Rathnavel
2017-09-25  9:52 ` Sergio Gonzalez Monroy [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=e057837d-22ad-f6da-e8e9-7a342b5401d6@intel.com \
    --to=sergio.gonzalez.monroy@intel.com \
    --cc=krath@cloudsimple.com \
    --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).