netdev - Re: Get rxhash fixes and RFS support in tun

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:	Wed, 20 Nov 2013 18:42:48 -0800
From:	Tom Herbert <therbert@...gle.com>
To:	David Miller <davem@...emloft.net>
Cc:	Linux Netdev List <netdev@...r.kernel.org>,
	Eric Dumazet <edumazet@...gle.com>,
	Jerry Chu <hkchu@...gle.com>
Subject: Re: Get rxhash fixes and RFS support in tun

On Wed, Nov 20, 2013 at 6:23 PM, David Miller <davem@...emloft.net> wrote:
> From: Tom Herbert <therbert@...gle.com>
> Date: Wed, 20 Nov 2013 17:09:01 -0800
>
>> I think the sw_rxhash holds more useful information.  In
>> skb_get_rxhash() either we have an L4 hash or we try to get one by
>> doing the SW computation, this means we don't ever return a HW hash
>> which is not L4 so basically that case is treated as an invalid hash.
>> So after a call to skb_get_rxhash, at least one of l4_rxhash or
>> sw_rxhash is and we won't redo flow dissection on subsequent calls
>> unless the rxhash is cleared.
>
> Ok.... one interesting issue remains, which is that the SW flow
> dissector considers decapsulating protocols like AH and ESP as
> "L4".
>
Thinking about a little more, maybe the valid bit is better.  Using
rxhash for both populating the flow hash table for consumption at a
low level and using it for flow lookup at a higher level might be at
odds.  In the original patches Eric mentioned that tun should always
compute its own hash anyway.

> I think we'll need to do something about that at some point.

We need the rxhash to be the value seen at the point of RPS (to do RFS
correctly), which I think probably means we don't ever want to change
it after the first calculation! (clearing at tunnel decap wouldn't be
correct either)  For ESP or AH, I believe it's appropriate to use SPI
as a substitute for ports.

Thanks for the comments,
Tom
--
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