netdev - Re: [PATCH v5 net-next 09/14] ip6_tun: Add infrastructure for doing encapsulation

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:	Mon, 16 May 2016 12:24:37 -0700
From:	Alexander Duyck <alexander.duyck@...il.com>
To:	Tom Herbert <tom@...bertland.com>
Cc:	David Miller <davem@...emloft.net>,
	Netdev <netdev@...r.kernel.org>, Kernel Team <kernel-team@...com>
Subject: Re: [PATCH v5 net-next 09/14] ip6_tun: Add infrastructure for doing encapsulation

On Sun, May 15, 2016 at 4:42 PM, Tom Herbert <tom@...bertland.com> wrote:
> Add encap_hlen and ip_tunnel_encap structure to ip6_tnl. Add functions
> for getting encap hlen, setting up encap on a tunnel, performing
> encapsulation operation.
>
> Signed-off-by: Tom Herbert <tom@...bertland.com>
> ---
>  include/net/ip6_tunnel.h  | 58 ++++++++++++++++++++++++++++++
>  net/ipv4/ip_tunnel_core.c |  5 +++
>  net/ipv6/ip6_tunnel.c     | 89 +++++++++++++++++++++++++++++++++++++++++------
>  3 files changed, 141 insertions(+), 11 deletions(-)

So a bisect is pointing to this patch as causing a regression in IPv6
GRE throughput from 20 Gb/s to .04 Mb/s

<...>

> diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
> index e79330f..9f0ea85 100644
> --- a/net/ipv6/ip6_tunnel.c
> +++ b/net/ipv6/ip6_tunnel.c
> @@ -1010,7 +1010,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
>         struct dst_entry *dst = NULL, *ndst = NULL;
>         struct net_device *tdev;
>         int mtu;
> -       unsigned int max_headroom = sizeof(struct ipv6hdr);
> +       unsigned int max_headroom = sizeof(struct ipv6hdr) + t->hlen;
>         int err = -1;
>
>         /* NBMA tunnel */
> @@ -1063,7 +1063,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
>                                      t->parms.name);
>                 goto tx_err_dst_release;
>         }
> -       mtu = dst_mtu(dst) - sizeof(*ipv6h);
> +       mtu = dst_mtu(dst) - sizeof(*ipv6h) - t->hlen;
>         if (encap_limit >= 0) {
>                 max_headroom += 8;
>                 mtu -= 8;

So I am pretty sure this bit here is causing the regression.  Your skb
already has a GRE header added and it is included in skb->len.  In the
tests just below here you are comparing skb->len to mtu, but you now
have the GRE header included twice so it is going to fail.  Odds are
this should be t->encap_hlen, and not t->hlen.

- Alex