From: Yuanhan Liu <yuanhan.liu@linux.intel.com>
To: Panu Matilainen <pmatilai@redhat.com>
Cc: dev@dpdk.org, huawei.xie@intel.com,
Thomas Monjalon <thomas.monjalon@6wind.com>,
Rich Lane <rich.lane@bigswitch.com>,
Tetsuya Mukawa <mukawa@igel.co.jp>
Subject: Re: [dpdk-dev] [PATCH v3 00/20] vhost ABI/API refactoring
Date: Thu, 30 Jun 2016 15:57:00 +0800 [thread overview]
Message-ID: <20160630075700.GH2831@yliu-dev.sh.intel.com> (raw)
In-Reply-To: <2add15af-5323-1f0c-b057-e29eeefb94c3@redhat.com>
On Thu, Jun 30, 2016 at 10:39:45AM +0300, Panu Matilainen wrote:
> On 06/07/2016 06:51 AM, Yuanhan Liu wrote:
> >v3: - adapted the new vhost ABI/API changes to tep_term example, to make
> > sure not break build at least.
> > - bumped the ABI version to 3
> >
> >NOTE: I created a branch at dpdk.org [0] for more conveinient testing:
> >
> > [0]: git://dpdk.org/next/dpdk-next-virtio for-testing
> >
> >
> >Every time we introduce a new feature to vhost, we are likely to break
> >ABI. Moreover, some cleanups (such as the one from Ilya to remove vec_buf
> >from vhost_virtqueue struct) also break ABI.
> >
> >This patch set is meant to resolve above issue ultimately, by hiding
> >virtio_net structure (as well as few others) internaly, and export the
> >virtio_net dev strut to applications by a number, vid, like the way
> >kernel exposes an fd to user space.
> >
> >Back to the patch set, the first part of this set makes some changes to
> >vhost example, vhost-pmd and vhost, bit by bit, to remove the dependence
> >to "virtio_net" struct. And then do the final change to make the current
> >APIs to adapt to using "vid".
> >
> >After that, "vrtio_net_device_ops" is the only left open struct that an
> >application can acces, therefore, it's the only place that might introduce
> >potential ABI breakage in future for extension. Hence, I made few more
> >(5) space reservation, to make sure we will not break ABI for a long time,
> >and hopefuly, forever.
>
> Been intending to say this for a while but seems I never actually got around
> to do so:
>
> This is a really fine example of how to refactor an API against constant ABI
> breakages, thank you Yuanhan!
Panu, thanks!
> Exported structs are one of the biggest
> obstacles in keeping a stable ABI while adding new features, and while its
> not always possible to hide everything to this extent, the damage (erm,
> exposure) can usually be considerably limited by careful API design.
Agreed.
> Since the first and the foremost objection against doing this in the DPDK
> context is always "but performance!", I'm curious as to what sort of numbers
> you're getting with the new API vs the old one? I'm really hoping other
> libraries would follow suit after seeing that its possible to provide a
> future-proof API/ABI without sacrificing performance :)
>From my (limited) test, nope, I see no performance drop at all, not even a
little.
--yliu
next prev parent reply other threads:[~2016-06-30 7:56 UTC|newest]
Thread overview: 76+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-05-02 22:25 [dpdk-dev] [PATCH 00/16] " Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 01/16] vhost: declare backend with int type Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 02/16] vhost: set/reset dev flags internally Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 03/16] vhost: declare device_fh as int Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 04/16] example/vhost: make a copy of virtio device id Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 05/16] vhost: rename device_fh to vid Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 06/16] vhost: get device by vid only Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 07/16] vhost: move vhost_device_ctx to cuse Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 08/16] vhost: query pmd internal by vid Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 09/16] vhost: add few more functions Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 10/16] vhost: export vid as the only interface to applications Yuanhan Liu
2016-05-10 16:17 ` Rich Lane
2016-05-10 16:39 ` Yuanhan Liu
2016-05-10 17:13 ` Rich Lane
2016-05-10 17:29 ` Yuanhan Liu
2016-05-10 19:22 ` Thomas Monjalon
2016-05-02 22:25 ` [dpdk-dev] [PATCH 11/16] vhost: hide internal structs/macros/functions Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 12/16] vhost: remove unnecessary fields Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 13/16] vhost: remove virtio-net.h Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 14/16] vhost: reserve few more space for future extension Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 15/16] vhost: per device vhost_hlen Yuanhan Liu
2016-05-02 22:25 ` [dpdk-dev] [PATCH 16/16] vhost: make buf vector for scatter Rx local Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 00/19] vhost ABI/API refactoring Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 01/19] vhost: declare backend with int type Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 02/19] vhost: set/reset dev flags internally Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 03/19] vhost: declare device fh as int Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 04/19] examples/vhost: make a copy of virtio device id Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 05/19] vhost: rename device fh to vid Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 06/19] vhost: get device by vid only Yuanhan Liu
2016-05-13 5:24 ` [dpdk-dev] [PATCH v2 07/19] vhost: move vhost device ctx to cuse Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 08/19] vhost: introduce new API to export numa node Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 09/19] vhost: introduce new API to export number of queues Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 10/19] vhost: introduce new API to export ifname Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 11/19] vhost: introduce new API to export queue free entries Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 12/19] vhost: remove dependency on priv field Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 13/19] vhost: export vid as the only interface to applications Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 14/19] vhost: hide internal structs/macros/functions Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 15/19] vhost: remove unnecessary fields Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 16/19] vhost: remove virtio-net.h Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 17/19] vhost: reserve few more space for future extension Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 18/19] vhost: per device virtio net header len Yuanhan Liu
2016-05-13 5:25 ` [dpdk-dev] [PATCH v2 19/19] vhost: make buf vector for scatter Rx local Yuanhan Liu
2016-05-26 17:04 ` [dpdk-dev] [PATCH v2 00/19] vhost ABI/API refactoring Rich Lane
2016-05-27 1:36 ` Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 00/20] " Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 01/20] vhost: declare backend with int type Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 02/20] vhost: set/reset dev flags internally Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 03/20] vhost: declare device fh as int Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 04/20] examples/vhost: make a copy of virtio device id Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 05/20] vhost: rename device fh to vid Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 06/20] vhost: get device by vid only Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 07/20] vhost: move vhost device ctx to cuse Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 08/20] vhost: introduce new API to export numa node Yuanhan Liu
2016-06-07 11:42 ` Panu Matilainen
2016-06-07 12:45 ` Yuanhan Liu
2016-06-08 21:51 ` Rich Lane
2016-06-09 4:45 ` Yuanhan Liu
2016-06-13 10:05 ` Yuanhan Liu
2016-06-07 3:51 ` [dpdk-dev] [PATCH v3 09/20] vhost: introduce new API to export number of queues Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 10/20] vhost: introduce new API to export ifname Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 11/20] vhost: introduce new API to export queue free entries Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 12/20] vhost: remove dependency on priv field Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 13/20] vhost: export vid as the only interface to applications Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 14/20] vhost: hide internal structs/macros/functions Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 15/20] vhost: remove unnecessary fields Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 16/20] vhost: remove virtio-net.h Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 17/20] vhost: reserve few more space for future extension Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 18/20] examples/tep_term: adapt to new vhost ABI/API changes Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 19/20] vhost: per device virtio net header len Yuanhan Liu
2016-06-07 3:52 ` [dpdk-dev] [PATCH v3 20/20] vhost: make buf vector for scatter Rx local Yuanhan Liu
2016-06-14 12:00 ` [dpdk-dev] [PATCH v3 00/20] vhost ABI/API refactoring Yuanhan Liu
2016-06-30 7:39 ` Panu Matilainen
2016-06-30 7:57 ` Yuanhan Liu [this message]
2016-06-30 9:05 ` Panu Matilainen
2016-06-30 11:15 ` Mcnamara, John
2016-06-30 11:40 ` 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=20160630075700.GH2831@yliu-dev.sh.intel.com \
--to=yuanhan.liu@linux.intel.com \
--cc=dev@dpdk.org \
--cc=huawei.xie@intel.com \
--cc=mukawa@igel.co.jp \
--cc=pmatilai@redhat.com \
--cc=rich.lane@bigswitch.com \
--cc=thomas.monjalon@6wind.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).