netdev - Re: [PATCH net-next 1/7] net: Add skb_get_hash

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, 29 Apr 2015 04:19:30 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Florian Westphal <fw@...len.de>
Cc:	Tom Herbert <tom@...bertland.com>, davem@...emloft.net,
	netdev@...r.kernel.org, jiri@...nulli.us
Subject: Re: [PATCH net-next 1/7] net: Add skb_get_hash_perturb

On Wed, 2015-04-29 at 09:59 +0200, Florian Westphal wrote:
> Tom Herbert <tom@...bertland.com> wrote:
> > This is used to get the skb->hash and then perturb it for a local use.
> > 
> > Signed-off-by: Tom Herbert <tom@...bertland.com>
> > ---
> >  include/linux/skbuff.h | 15 +++++++++++++++
> >  1 file changed, 15 insertions(+)
> > 
> > diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> > index 66e374d..b706889 100644
> > --- a/include/linux/skbuff.h
> > +++ b/include/linux/skbuff.h
> > @@ -17,6 +17,7 @@
> >  #include <linux/kernel.h>
> >  #include <linux/kmemcheck.h>
> >  #include <linux/compiler.h>
> > +#include <linux/jhash.h>
> >  #include <linux/time.h>
> >  #include <linux/bug.h>
> >  #include <linux/cache.h>
> > @@ -927,6 +928,20 @@ static inline __u32 skb_get_hash(struct sk_buff *skb)
> >  	return skb->hash;
> >  }
> >  
> > +static inline __u32 skb_get_hash_perturb(struct sk_buff *skb,
> > +					 u32 perturb)
> > +{
> > +	u32 hash = skb_get_hash(skb);
> > +
> > +	if (likely(hash)) {
> > +		hash = jhash_1word((__force __u32) hash, perturb);
> 
> Whats this perturb for?
> 
> perturb is supposed to make sure that if you have
> 
> flow1, flow2 where hash(flow1, perturb) == hash(flow2, perturb)
> the collision will be temporary and go away once perturb changes.
> 
> If you perturb after hashing, such collision is permanent.
> 
> So I think this either should flow_dissect + hash in software,
> or just use skb_get_hash without perturb.

Yes, I mentioned this several times to Tom in the various past attempts.

Also, this stuff means that sfq will call jhash twice instead of once,
unless skb->hash is already populated before sfq enqueue.

If the goal is to extend skb to include a flow_keys_digest instead of
u32 hash, lets state this right now ;)

I think I already suggested to add a 'perturb' parameter to
skb_get_hash() to use in place of hashrnd (found in
net/core/flow_dissector.c)

If hash is Toeplitz based and provided by a NIC, I doubt a jhash()
perturbation will get anything better.




--
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