netdev - Re: [Pv-drivers] rps and pvdrivers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sat, 4 Sep 2010 09:38:30 -0400
From:	Chetan Loke <chetanloke@...il.com>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	Bhavesh Davda <bhavesh@...are.com>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"pv-drivers@...are.com" <pv-drivers@...are.com>,
	"therbert@...gle.com" <therbert@...gle.com>
Subject: Re: [Pv-drivers] rps and pvdrivers

Hi Eric,

On Sat, Sep 4, 2010 at 2:51 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> Reproduce what exactly ?
>
> I dont understand what the problem is, reading your description.

With 2.6.35 the rx-path performance on a 10G vNIC is way too low. If
you setup a simple 'recvfrom' loop on a promiscuous interface you can
see this easily.
My use-case is to achieve high speed pkt capturing.

> RPS is not automatically switched on, you have to configure it.
>
> echo ffff >/sys/class/net/eth0/queues/rx-0/rps_cpus
>
> Same for RFS if you prefer to use RFS
>
> echo 16384 >/sys/class/net/eth0/queues/rx-0/rps_flow_cnt
>

Ok, thanks for sharing this. I tried this and still doesn't help.

> If you receive a flood, your cpu stay in NAPI mode, and no hardware
> interrupt is received while you process xxx.xxx packets per second.
>

I see. Ok then that's whats happening. I guess I will have to look at
the napi/ksoftirq/rps block in detail to understand this. But I would
think that even w/o the rps settings I should still get the same
numbers as compared to non-rps case, correct?

On a VM(virtual machine) using a 1G vNIC I can capture ~250K
pkts/sec(even higher in some cases). But I can't go beyond 100K
pkts/sec on a 10G vNIC because ksoftirqd consumes 1-cpu 100% of the
time. That's why I thought of switching to the 2.6.35 kernel to see if
I could scale on 10G.

It's possible that a VM cannot handle that much load. So I tried
sending only 10% of line-rate(10G) which is 1G. It still doesn't work.
I still can't capture that many pkts.

Chetan Loke
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html