DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Mattias Rönnblom" <mattias.ronnblom@ericsson.com>
To: bugzilla@dpdk.org, dev@dpdk.org
Cc: vipin.varghese@intel.com
Subject: Re: [dpdk-dev] [Bug 106] running dpdk-procinfo or any secondary leading to system memory exhaustion
Date: Mon, 12 Nov 2018 20:31:43 +0100	[thread overview]
Message-ID: <9239c635-8d5c-f5cf-8a2e-c4df7e09bc2b@ericsson.com> (raw)
In-Reply-To: <bug-106-3@http.bugs.dpdk.org/>

On 2018-11-12 11:44, bugzilla@dpdk.org wrote:
> https://bugs.dpdk.org/show_bug.cgi?id=106
> 
>              Bug ID: 106
>             Summary: running dpdk-procinfo or any secondary leading to
>                      system memory exhaustion
>             Product: DPDK
>             Version: 18.05
>            Hardware: x86
>                  OS: Linux
>              Status: CONFIRMED
>            Severity: critical
>            Priority: Normal
>           Component: core
>            Assignee: dev@dpdk.org
>            Reporter: vipin.varghese@intel.com
>    Target Milestone: ---
> 
> Issue: Running dpdk secondary instance in multiple iterations leads to memory
> exhaustion
> 

Exhaustion, meaning the kernel eventually - given enough iterations of 
the test program - invoke the OOM handler (i.e. crash)?

> 
> Run script:
> #!/bin/sh
> 
> ./build/app/test
> ret=$?
> 
> while [ $ret -eq 0 ]
> do
>     sync; echo 3 > /proc/sys/vm/drop_caches
>     ./build/app/test
>     ret=$?
>     echo "ret ---------------- $ret"
> done
> 

An application causing the kernel to cache more pages wouldn't normally 
be considered to be leaking memory, if those pages are not dirty. If 
"echo 3 > /proc/sys/vm/drop_caches" frees the memory "leaked", it wasn't 
leaked. That holds true even for system with swapping enabled.

You can't use MemFree (in /proc/meminfo) or similar as an indication how 
much memory is actually available to applications, would they need it. 
Pages that "echo 3 > /proc/sys/vm/drop_caches" frees, would also be free 
would the system come under memory pressure. A drop in MemAvailable is a 
better indication of applications' using more memory.

find $HOME -type f | xargs cat >/dev/null
For example, find+cat above aren't leaking memory, even if the pipeline 
likely causes MemFree to go down.

Anyway. If the answer to my first question is "the kernel crashes", you 
can ignore all this.

  reply	other threads:[~2018-11-12 19:31 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-11-12 10:44 bugzilla
2018-11-12 19:31 ` Mattias Rönnblom [this message]
2018-11-13 17:03 ` bugzilla

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=9239c635-8d5c-f5cf-8a2e-c4df7e09bc2b@ericsson.com \
    --to=mattias.ronnblom@ericsson.com \
    --cc=bugzilla@dpdk.org \
    --cc=dev@dpdk.org \
    --cc=vipin.varghese@intel.com \
    /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).