| Message ID | 201001190058.44495.opurdila@ixiacom.com |
|---|---|
| State | Accepted, archived |
| Delegated to: | David Miller |
| Headers | show |
Hi Octavian, On Tue, 19 Jan 2010 00:58:44 +0200 Octavian Purdila <opurdila@ixiacom.com> wrote: > If the per device ARP_ACCEPT option is enable, currently we only allow > creating new ARP cache entries for response type gratuitous ARP. > > Allowing gratuitous ARP to create new ARP entries (not only to update > existing ones) is useful when we want to avoid unnecessary delays for > the first packet of a stream. > > This patch allows request type gratuitous ARP to create new ARP cache > entries as well. This is useful when we want to populate the ARP cache > entries for a large number of hosts on the same LAN. > I don't think this complies with the ARP RFC (RFC826). Walking through the validation steps, Gratuitous ARP seems to only pass if there is an existing entry in the table, otherwise they'd fail, as the target address in the GA packet doesn't match that of the receiving host. I can guess your motivation for this, however I wonder if going outside the spec for relatively rare testing cases (in comparison to all the ARP usage on all IP LANs) is wise. While ARP is vulnerable to DoS attacks, I think allowing GAs to generate new ARP entries would increase its vulnerability. There may also be some other unexpected consequences that haven't been identified. Possibly for your test environments, maybe static ARP entries might be a feasible alternative that's still compliant with RFC826? Regards, Mark. > Signed-off-by: Octavian Purdila <opurdila@ixiacom.com> > --- > Documentation/networking/ip-sysctl.txt | 15 ++++++++++++--- > net/ipv4/arp.c | 3 ++- > 2 files changed, 14 insertions(+), 4 deletions(-) > > diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt > index c532884..2dc7a1d 100644 > --- a/Documentation/networking/ip-sysctl.txt > +++ b/Documentation/networking/ip-sysctl.txt > @@ -852,9 +852,18 @@ arp_notify - BOOLEAN > or hardware address changes. > > arp_accept - BOOLEAN > - Define behavior when gratuitous arp replies are received: > - 0 - drop gratuitous arp frames > - 1 - accept gratuitous arp frames > + Define behavior for gratuitous ARP frames who's IP is not > + already present in the ARP table: > + 0 - don't create new entries in the ARP table > + 1 - create new entries in the ARP table > + > + Both replies and requests type gratuitous arp will trigger the > + ARP table to be updated, if this setting is on. > + > + If the ARP table already contains the IP address of the > + gratuitous arp frame, the arp table will be updated regardless > + if this setting is on or off. > + > > app_solicit - INTEGER > The maximum number of probes to send to the user space ARP daemon > diff --git a/net/ipv4/arp.c b/net/ipv4/arp.c > index 0787092..1940b4d 100644 > --- a/net/ipv4/arp.c > +++ b/net/ipv4/arp.c > @@ -907,7 +907,8 @@ static int arp_process(struct sk_buff *skb) > devices (strip is candidate) > */ > if (n == NULL && > - arp->ar_op == htons(ARPOP_REPLY) && > + (arp->ar_op == htons(ARPOP_REPLY) || > + (arp->ar_op == htons(ARPOP_REQUEST) && tip == sip)) && > inet_addr_type(net, sip) == RTN_UNICAST) > n = __neigh_lookup(&arp_tbl, &sip, dev, 1); > } > -- > 1.5.6.5 > > -- > To unsubscribe from this list: send the line "unsubscribe netdev" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Mark Smith <lk-netdev@lk-netdev.nosense.org> Date: Tue, 19 Jan 2010 10:09:13 +1030 > I don't think this complies with the ARP RFC (RFC826). Walking through > the validation steps, Gratuitous ARP seems to only pass if there is an > existing entry in the table, otherwise they'd fail, as the target > address in the GA packet doesn't match that of the receiving host. It's fine if it's guarded by the per-device ARP_ACCEPT sysctl which defaults to off, and that's what he's doing. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Octavian Purdila <opurdila@ixiacom.com> Date: Tue, 19 Jan 2010 00:58:44 +0200 > If the per device ARP_ACCEPT option is enable, currently we only allow > creating new ARP cache entries for response type gratuitous ARP. > > Allowing gratuitous ARP to create new ARP entries (not only to update > existing ones) is useful when we want to avoid unnecessary delays for > the first packet of a stream. > > This patch allows request type gratuitous ARP to create new ARP cache > entries as well. This is useful when we want to populate the ARP cache > entries for a large number of hosts on the same LAN. > > Signed-off-by: Octavian Purdila <opurdila@ixiacom.com> Applied, thanks. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Dave, On Tue, 19 Jan 2010 02:11:38 -0800 (PST) David Miller <davem@davemloft.net> wrote: > From: Mark Smith <lk-netdev@lk-netdev.nosense.org> > Date: Tue, 19 Jan 2010 10:09:13 +1030 > > > I don't think this complies with the ARP RFC (RFC826). Walking through > > the validation steps, Gratuitous ARP seems to only pass if there is an > > existing entry in the table, otherwise they'd fail, as the target > > address in the GA packet doesn't match that of the receiving host. > > It's fine if it's guarded by the per-device ARP_ACCEPT sysctl which > defaults to off, and that's what he's doing. Fair enough. I wasn't familiar with the ARP_ACCEPT sysctl. It seems that this patch is not only adding the feature of creating new ARP table entries in response to GARPs, but is also fixing the ARP implementation because it looks like the default wasn't to accept GARPs, despite the validation steps in RFC826 allowing them. ARP_ACCEPT / the /proc "arp_accept" file is really confusingly named - unless you knew better I doubt anybody would guess it's only referring to GARPs. Is there any chance of renaming it or symlinking a new name of something like "garp_accept" to the old one? Is this a significant enough behaviour change that it creates that opportunity? Regards, Mark. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Mark Smith <lk-netdev@lk-netdev.nosense.org> Date: Tue, 19 Jan 2010 21:18:18 +1030 > ARP_ACCEPT / the /proc "arp_accept" file is really confusingly named - > unless you knew better I doubt anybody would guess it's only referring > to GARPs. Is there any chance of renaming it or symlinking a new name > of something like "garp_accept" to the old one? Is this a significant > enough behaviour change that it creates that opportunity? It's been there forever, it's well documented, and creating symlinks is messy. So no. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt index c532884..2dc7a1d 100644 --- a/Documentation/networking/ip-sysctl.txt +++ b/Documentation/networking/ip-sysctl.txt @@ -852,9 +852,18 @@ arp_notify - BOOLEAN or hardware address changes. arp_accept - BOOLEAN - Define behavior when gratuitous arp replies are received: - 0 - drop gratuitous arp frames - 1 - accept gratuitous arp frames + Define behavior for gratuitous ARP frames who's IP is not + already present in the ARP table: + 0 - don't create new entries in the ARP table + 1 - create new entries in the ARP table + + Both replies and requests type gratuitous arp will trigger the + ARP table to be updated, if this setting is on. + + If the ARP table already contains the IP address of the + gratuitous arp frame, the arp table will be updated regardless + if this setting is on or off. + app_solicit - INTEGER The maximum number of probes to send to the user space ARP daemon diff --git a/net/ipv4/arp.c b/net/ipv4/arp.c index 0787092..1940b4d 100644 --- a/net/ipv4/arp.c +++ b/net/ipv4/arp.c @@ -907,7 +907,8 @@ static int arp_process(struct sk_buff *skb) devices (strip is candidate) */ if (n == NULL && - arp->ar_op == htons(ARPOP_REPLY) && + (arp->ar_op == htons(ARPOP_REPLY) || + (arp->ar_op == htons(ARPOP_REQUEST) && tip == sip)) && inet_addr_type(net, sip) == RTN_UNICAST) n = __neigh_lookup(&arp_tbl, &sip, dev, 1); }
If the per device ARP_ACCEPT option is enable, currently we only allow creating new ARP cache entries for response type gratuitous ARP. Allowing gratuitous ARP to create new ARP entries (not only to update existing ones) is useful when we want to avoid unnecessary delays for the first packet of a stream. This patch allows request type gratuitous ARP to create new ARP cache entries as well. This is useful when we want to populate the ARP cache entries for a large number of hosts on the same LAN. Signed-off-by: Octavian Purdila <opurdila@ixiacom.com> --- Documentation/networking/ip-sysctl.txt | 15 ++++++++++++--- net/ipv4/arp.c | 3 ++- 2 files changed, 14 insertions(+), 4 deletions(-)