From: dpdk <dpdk-H32Fclmsjq1BDgjK7y7TUQ@public.gmane.org>
To: dev <dev-VfR2kkLFssw@public.gmane.org>
Subject: zerocopy between userspace vhost virqueue and vmdq eth nic queu
Date: Sun, 2 Mar 2014 10:55:37 +0800 [thread overview]
Message-ID: <tencent_25C93A91764026FD2028C210@qq.com> (raw)
From ver 1.6, dpdk support userspace vhost backend, which is very useful for high performance & low lagency soft vswitching.
But as I see the sample in git tree (dpdk.org/browse/dpdk/tree/examples/vhost/main.c), packets tx/rx between peth nic and virtio vnic queues are both copied once.
Since gust vm are created on hugepage, and if vmdq exists for hw based L2 dispatch, may we directly use buffer from virtqueue for hw DMA, so a totally zerocopy is obtained?
Several years before, intel submit a patch of similar idea to KVM, but not being accepted(lwn.net/Articles/413239/), I think the reason is too much change on kernel. Now with userspace vhost, we have a new chance to implement it.
reply other threads:[~2014-03-02 2:55 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=tencent_25C93A91764026FD2028C210@qq.com \
--to=dpdk-h32fclmsjq1bdgjk7y7tuq@public.gmane.org \
--cc=dev-VfR2kkLFssw@public.gmane.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).