netdev - Re: [PATCH] net: Check skb->rxhash in gro

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 for Android: free password hash cracker in your pocket

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

Date:	Fri, 10 Jan 2014 09:45:35 -0800
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Tom Herbert <therbert@...gle.com>
Cc:	David Miller <davem@...emloft.net>,
	Linux Netdev List <netdev@...r.kernel.org>
Subject: Re: [PATCH] net: Check skb->rxhash in gro_receive

On Fri, 2014-01-10 at 08:27 -0800, Tom Herbert wrote:
> On Thu, Jan 9, 2014 at 9:38 PM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> > On Thu, 2014-01-09 at 20:54 -0800, Tom Herbert wrote:
> >> When initializing a gro_list for a packet, first check the rxhash of
> >> the incoming skb against that of the skb's in the list. This should be
> >> a very strong inidicator of whether the flow is going to be matched,
> >> and potentially allows a lot of other checks to be short circuited.
> >>
> >
> > Hmm... this idea was discussed in the past. I used it when attempting to
> > use a hash table instead of a gro_list last year.
> >
> > Unfortunately this added lot of cycles when rxhash is not provided by
> > hardware, and my tests found it was not a win.
> >
> > Remember : in most cases, gro_list contains one flow, so this test does
> > nothing special but adds overhead.
> 
> I don't understand what your basis is that gro_list in most cases
> contains one flow, but assuming that were true, maybe we should make
> the it only contain one flow eliminating the complexity of multiple
> flows (same_flow logic is convoluted and layers of encapsulation is
> not going to simplify things).
> 

The complexity of GRO is the aggregation itself. You wont avoid it.

> If we are doing RPS or RFS, rxhash will be computed anyway, so the
> case your optimizing is pretty narrow: no RPS, no RFS, no hardware
> hash, and a single flow in gro_list. Nevertheless, if this is really
> an important concern, we can make the check directly against
> skb->rxhash so to be opportunistic and avoid the possibility of
> computation.


We'll compute rxhash once per GRO packet, containing up to 40 MSS
packets.

Thats a big difference.

If your patch was doing this, I would have no complain.

(No new conditional branch, and skb->rxhash, if provided by the NIC,
can give a hint.)

diff --git a/net/core/dev.c b/net/core/dev.c
index ce01847793c0..c9f7a26d7ce7 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3798,7 +3798,8 @@ static void gro_list_prepare(struct napi_struct *napi, struct sk_buff *skb)
        for (p = napi->gro_list; p; p = p->next) {
                unsigned long diffs;
 
-               diffs = (unsigned long)p->dev ^ (unsigned long)skb->dev;
+               diffs = p->rxhash ^ skb->rxhash;
+               diffs |= (unsigned long)p->dev ^ (unsigned long)skb->dev;
                diffs |= p->vlan_tci ^ skb->vlan_tci;
                if (maclen == ETH_HLEN)
                        diffs |= compare_ether_header(skb_mac_header(p),



--
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