netdev - Accelerated RFS steering the same flow repeatedly on sfc?

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>] [day] [month] [year] [list]

Date:	Thu, 20 Mar 2014 12:44:15 -0700
From:	Andy Lutomirski <luto@...capital.net>
To:	Network Development <netdev@...r.kernel.org>,
	Paul Emberson <pemberson@...arflare.com>,
	Tom Herbert <therbert@...gle.com>,
	Solarflare linux maintainers <linux-net-drivers@...arflare.com>
Subject: Accelerated RFS steering the same flow repeatedly on sfc?

This is a snipped from my kernel log:

[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
159.95.26.109:23042:208.92.147.76:44811 to queue 28 [flow 21 filter
5781] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 130 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
199.29.245.125:11097:10.253.8.22:47709 to queue 28 [flow 477 filter
1497] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
199.29.245.125:11097:10.253.8.22:47709 to queue 30 [flow 272 filter
1497] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
199.29.245.125:11097:10.253.8.22:47709 to queue 30 [flow 272 filter
1497] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 400 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1
[info] sfc 0000:05:00.1 p3p2: steering TCP
169.33.100.144:15986:208.92.147.76:60517 to queue 28 [flow 399 filter
3187] SUBSYSTEM=pci DEVICE=+pci:0000:05:00.1

Sorry about the formatting.  This snippet is less than one second long.

It seems rather odd:

1. The same flow is being steered over and over.  This means that
set_rps_cpu is being called repeatedly, I think.

2. The same flow doesn't have the same flow_id (i.e. rxhash) every time.

The rxhash setting on my interface doesn't seem to matter.

This is Linux 3.13 running the latest OpenOnload driver.  I can try
the stock driver if needed next week.

The data is coming in on a vlan, in case that matters.

The OpenOnload driver is missing this fix:

commit c7cb38af79c1e71a72d398f39296f1e27779b778
Author: Tom Herbert <therbert@...gle.com>
Date:   Tue Dec 17 23:31:50 2013 -0800

    net: sfc calls skb_set_hash


-- 
Andy Lutomirski
AMA Capital Management, LLC
--
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