From: Jakub Kicinski <kuba@kernel.org>
To: Breno Leitao <leitao@debian.org>
Cc: horms@kernel.org, davem@davemloft.net, edumazet@google.com,
pabeni@redhat.com, thepacketgeek@gmail.com,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
davej@codemonkey.org.uk, vlad.wing@gmail.com, max@kutsevol.com,
kernel-team@meta.com, jiri@resnulli.us, jv@jvosburgh.net,
andy@greyhouse.net, aehkn@xenhub.one,
Rik van Riel <riel@surriel.com>,
Al Viro <viro@zeniv.linux.org.uk>
Subject: Re: [PATCH net-next 1/3] net: netpoll: Defer skb_pool population until setup success
Date: Fri, 1 Nov 2024 19:01:01 -0700 [thread overview]
Message-ID: <20241101190101.4a2b765f@kernel.org> (raw)
In-Reply-To: <20241101-prompt-carrot-hare-ff2aaa@leitao>
On Fri, 1 Nov 2024 11:18:29 -0700 Breno Leitao wrote:
> > I think that a best mechanism might be something like:
> >
> > * If find_skb() needs to consume from the pool (which is rare, only
> > when alloc_skb() fails), raise workthread that tries to repopulate the
> > pool in the background.
> >
> > * Eventually avoid alloc_skb() first, and getting directly from the
> > pool first, if the pool is depleted, try to alloc_skb(GPF_ATOMIC).
> > This might make the code faster, but, I don't have data yet.
>
> I've hacked this case (getting the skb from the pool first and refilling
> it on a workqueue) today, and the performance is expressive.
>
> I've tested sending 2k messages, and meassured the time it takes to
> run `netpoll_send_udp`, which is the critical function in netpoll.
The purpose of the pool is to have a reserve in case of OOM, AFAIU.
We may speed things up by taking the allocations out of line but
we risk the pool being empty when we really need it.
next prev parent reply other threads:[~2024-11-02 2:01 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-25 14:20 [PATCH net-next 0/3] net: netpoll: Improve SKB pool management Breno Leitao
2024-10-25 14:20 ` [PATCH net-next 1/3] net: netpoll: Defer skb_pool population until setup success Breno Leitao
2024-11-01 1:26 ` Jakub Kicinski
2024-11-01 10:51 ` Breno Leitao
2024-11-01 18:18 ` Breno Leitao
2024-11-02 2:01 ` Jakub Kicinski [this message]
2024-11-04 20:40 ` Breno Leitao
2024-11-06 1:00 ` Jakub Kicinski
2024-11-06 15:06 ` Breno Leitao
2024-11-06 23:43 ` Jakub Kicinski
2024-11-07 11:50 ` Breno Leitao
2024-10-25 14:20 ` [PATCH net-next 2/3] net: netpoll: Individualize the skb pool Breno Leitao
2024-11-01 1:28 ` Jakub Kicinski
2024-11-01 11:56 ` Breno Leitao
2024-10-25 14:20 ` [PATCH net-next 3/3] net: netpoll: flush skb pool during cleanup Breno Leitao
2024-11-01 1:29 ` Jakub Kicinski
2024-11-01 11:57 ` Breno Leitao
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=20241101190101.4a2b765f@kernel.org \
--to=kuba@kernel.org \
--cc=aehkn@xenhub.one \
--cc=andy@greyhouse.net \
--cc=davej@codemonkey.org.uk \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=horms@kernel.org \
--cc=jiri@resnulli.us \
--cc=jv@jvosburgh.net \
--cc=kernel-team@meta.com \
--cc=leitao@debian.org \
--cc=linux-kernel@vger.kernel.org \
--cc=max@kutsevol.com \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=riel@surriel.com \
--cc=thepacketgeek@gmail.com \
--cc=viro@zeniv.linux.org.uk \
--cc=vlad.wing@gmail.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).