DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Wiles, Keith" <keith.wiles@intel.com>
To: Alexandru Ciobotaru <alex.ciobotaru@gmail.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] Question about adding a new EAL
Date: Tue, 22 May 2018 19:53:30 +0000	[thread overview]
Message-ID: <231457C6-BC47-42E3-8B3B-9B8AA121A580@intel.com> (raw)
In-Reply-To: <CANKOshyKZACv4DSmTLcUTGxcrFNfCs3i3__HwXbjutCsZDcp+g@mail.gmail.com>



On May 22, 2018, at 2:23 PM, Alexandru Ciobotaru <alex.ciobotaru@gmail.com<mailto:alex.ciobotaru@gmail.com>> wrote:

Greetings,

I can not answer all of your questions, but maybe a few.


I'm currently checking on how to add new EAL "app" to the DPDK so I started
by adding a new "xyzapp" next to "linuxapp" and "bsdapp".
Now, I would probably expect to make this new EAL mainline compliant on
day, and to avoid future headaches, the plan is to avoid modifying things
outside of my "xyzapp".
And thus, the first adaptation issue has arrived:

EAL is designed so that I can re-implement the "common" source files into
my "xyzapp" but how would I override the "include/rte_xyz.h" headers from
the "common" part of the EAL library (e.g.
librte_eal/common/include/rte_eal.h)? Some of the includes in these headers
are currently N/A to my toolchain (via CROSS=) or to my executive
environment.

Normally, these types of differences are just empty variables or something that allows the system build even when you do not support it.

For example I do not have <sched.h> or <sys/queue.h> which are currently
included in some EAL API headers.

Maybe you should make simple empty headers, to work around those headers. I am sure it would be easier, but maybe someone has a better idea.

Should more work into making the "common" part truly generic be put into
this?

We always like to see patches to make the EAL simpler to maintain, so please introduce a patch.

Is the current EAL API frozen for compatibility reasons? Any "next" branch
where such modifications are accepted?

Changing a EAL API is really discouraged as it effect a lot of places. But we do have a method to change exposed APIs. It will take two releases to get your change into DPDK the first release gives notice and the second one allows the API to exist in the code. You can read more abort the process here: http://dpdk.org/dev make sure you read it all :-)

Or, is there a straight method of enforcing my "xyzapp" symlinked EAL
headers in the build directory without modifying the "common" Makefile?

You should avoid symlinks in the main repo directories as they are difficult to maintain and some systems do not support them.


Thank you,
Alex

Let us know if you need anything else.

Regards,
Keith

  reply	other threads:[~2018-05-22 19:53 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-05-22 19:23 Alexandru Ciobotaru
2018-05-22 19:53 ` Wiles, Keith [this message]
2018-05-22 19:59 ` Wiles, Keith
2018-05-22 20:18   ` Thomas Monjalon
2018-05-22 20:56     ` Alex Ciobotaru
2018-05-22 21:04       ` Thomas Monjalon

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=231457C6-BC47-42E3-8B3B-9B8AA121A580@intel.com \
    --to=keith.wiles@intel.com \
    --cc=alex.ciobotaru@gmail.com \
    --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).