Get icmp ratelimit from sysctl in ipt_REJECT.c
Subsystems (3)
PATCH 1/1 Get icmp ratelimit from sysctl in ipt_REJECT.c
Hi,
Whilst working on a netfilter module, I noticed a comment in ipt_REJECT.c saying 'FIXME: Use sysctl number. --RR' before a call to xrlim_allow(). This is fixed in the attached patch (against 2.6.11-rc2-mm1, but applies cleanly to rc3-mm1 as well).
Dunk
diff -ur linux-2.6.11-rc2-mm1.orig/net/ipv4/icmp.c linux-2.6.11-rc2-mm1/net/ipv4/icmp.c
--- linux-2.6.11-rc2-mm1.orig/net/ipv4/icmp.c
2005-01-28 13:01:17.000000000 +1100
+++ linux-2.6.11-rc2-mm1/net/ipv4/icmp.c 2005-02-08
15:07:11.000000000 +1100
@@ -208,6 +208,8 @@
int sysctl_icmp_ratelimit = 1 * HZ;
int sysctl_icmp_ratemask = 0x1818;
+EXPORT_SYMBOL(sysctl_icmp_ratelimit);
+
/*
* ICMP control array. This specifies what to do with
each ICMP.
*/
diff -ur linux-2.6.11-rc2-mm1.orig/net/ipv4/netfilter/ipt_REJECT.c linux-2.6.11-rc2-mm1/net/ipv4/netfilter/ipt_REJECT.c
---linux-2.6.11-rc2-mm1.orig/net/ipv4/netfilter/ipt_REJECT.c
2005-01-28 13:01:18.000000000 +1100 +++ linux-2.6.11-rc2-mm1/net/ipv4/netfilter/ipt_REJECT.c 2005-02-08 15:02:42.000000000 +1100
@@ -234,8 +234,7 @@
if (!rt)
return;
- /* FIXME: Use sysctl number. --RR */
- if (!xrlim_allow(&rt->u.dst, 1*HZ))
+ if (!xrlim_allow(&rt->u.dst, sysctl_icmp_ratelimit))
return;
iph = skb_in->nh.iph;
diff -ur linux-2.6.11-rc2-mm1.orig/net/ipv4/sysctl_net_ipv4.c linux-2.6.11-rc2-mm1/net/ipv4/sysctl_net_ipv4.c
---linux-2.6.11-rc2-mm1.orig/net/ipv4/sysctl_net_ipv4.c
2004-12-25 08:35:23.000000000 +1100 +++ linux-2.6.11-rc2-mm1/net/ipv4/sysctl_net_ipv4.c 2005-02-08 15:07:54.000000000 +1100
@@ -15,6 +15,7 @@
#include <net/ip.h>
#include <net/route.h>
#include <net/tcp.h>
+#include <net/icmp.h>
/* From af_inet.c */
extern int sysctl_ip_nonlocal_bind;
@@ -34,7 +35,6 @@
extern int sysctl_ip_dynaddr;
/* From icmp.c */
-extern int sysctl_icmp_ratelimit;
extern int sysctl_icmp_ratemask;
/* From igmp.c */
diff -ur linux-2.6.11-rc2-mm1.orig/include/net/icmp.h linux-2.6.11-rc2-mm1/include/net/icmp.h
--- linux-2.6.11-rc2-mm1.orig/include/net/icmp.h
2005-01-28 13:01:11.000000000 +1100
+++ linux-2.6.11-rc2-mm1/include/net/icmp.h 2005-02-08
15:07:04.000000000 +1100
@@ -38,6 +38,8 @@
#define ICMP_INC_STATS_BH(field)SNMP_INC_STATS_BH(icmp_statistics, field)
#define ICMP_INC_STATS_USER(field)SNMP_INC_STATS_USER(icmp_statistics, field)
+extern int sysctl_icmp_ratelimit;
+
extern void icmp_send(struct sk_buff *skb_in, int
type, int code, u32 info);
extern int icmp_rcv(struct sk_buff *skb);
extern int icmp_ioctl(struct sock *sk, int cmd,
unsigned long arg);
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
Quoted 2 lines · --- Duncan Palmer
fixed in the attached patch (against 2.6.11-rc2-mm1, but applies cleanly to rc3-mm1 as well).
Bugger - yahoo has mangled the patch - sorry. I've attached it instead...
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
net/ipv4/icmp.c +2 -0
@@ -208,6 +208,8 @@int sysctl_icmp_ratelimit =1 *HZ ;int sysctl_icmp_ratemask =0x1818 ; +EXPORT_SYMBOL (sysctl_icmp_ratelimit); +/* * ICMP control array. This specifies what to do with each ICMP. */
net/ipv4/netfilter/ipt_REJECT.c +1 -2
@@ -234,8 +234,7 @@if (!rt)return ; -/* FIXME: Use sysctl number. --RR */ -if (!xrlim_allow (&rt->u.dst,1 *HZ )) +if (!xrlim_allow (&rt->u.dst, sysctl_icmp_ratelimit))return ; iph = skb_in->nh.iph;
net/ipv4/sysctl_net_ipv4.c +1 -1
@@ -15,6 +15,7 @@#include <net/ip.h> #include <net/route.h> #include <net/tcp.h> +#include <net/icmp.h> /* From af_inet.c */ extern int sysctl_ip_nonlocal_bind;
@@ -34,7 +35,6 @@extern int sysctl_ip_dynaddr;/* From icmp.c */ -extern int sysctl_icmp_ratelimit;extern int sysctl_icmp_ratemask;/* From igmp.c */
include/net/icmp.h +2 -0
@@ -38,6 +38,8 @@#define ICMP_INC_STATS_BH (field) SNMP_INC_STATS_BH(icmp_statistics, field)#define ICMP_INC_STATS_USER (field) SNMP_INC_STATS_USER(icmp_statistics, field) +extern int sysctl_icmp_ratelimit; +extern void icmp_send (struct sk_buff *skb_in,int type,int code,u32 info);extern int icmp_rcv (struct sk_buff *skb);extern int icmp_ioctl (struct sock *sk,int cmd,unsigned long arg);
Quoted 8 lines · Duncan Palmer
Hi,
Whilst working on a netfilter module, I noticed a comment in
ipt_REJECT.csaying 'FIXME: Use sysctl number. --RR' before a call toxrlim_allow(). This is fixed in the attached patch (against 2.6.11-rc2-mm1, but applies cleanly to rc3-mm1 as well).
I was never convinced of using xrlim_allow in ipt_REJECT. We already have the limit match, which allows to set limits in the ruleset. The current of of xrlim_allow is actually broken, the max number of tokens in the dst_entry is limited by the timeout value, so both icmp.c and ipt_REJECT.c are limited by the smaller of both timeouts when ICMPs are sent by both. ipt_REJECT also doesn't care about icmp_ratemask and uses up token for types that shouldn't be limited. So we can either remove it entirely or use icmpv4_xrlim_allow. If there are no objections I would prefer to remove it.
Regards
Patrick
Quoted 5 lines · --- Patrick McHardy
I was never convinced of using
xrlim_allowinipt_REJECT. We
already have the limit match, which allows to set
limits in
the ruleset.
Is there anywhere that sends icmp pkts apart from icmp.c and ipt_REJECT.c? (a quick look suggests not). If not, then one problem I can see with removing xrlim_allow() from ipt_REJECT is that its behaviour will become inconsistent with that of icmp.c, in that it doesn't use sysctl for setting rlimit (not that it ever has...)
Quoted 4 lines
that shouldn't be limited. So we can either remove it entirely
or useicmpv4_xrlim_allow. If there are no objections I would prefer to remove it.
After reading the relevant bits of the RFC and a bit more code, I agree that xrlim_allow() is indeed buggered...
I'm far from being an expert on linux's networking internals, but it seems to me that many aspects of the operation of network stacks are configurable using sysctl variables. Not calling icmpv4_xrlim_allow() will make the icmp ratelimit parameter a bit of an odd one out as far as ipv4 is concerned, as I think there are other ipv4 sysctl parameters who's functionality could similarly be replaced by iptables...
I'll be happy do do up a patch on whatever is decided upon anyway...
Dunk
__________________________________
Do you Yahoo!?
The all-new My Yahoo! - Get yours free!
http://my.yahoo.com
Quoted 8 lines · Duncan Palmer
Is there anywhere that sends icmp pkts apart from
icmp.candipt_REJECT.c? (a quick look suggests not). If not, then one problem I can see with removingxrlim_allow()fromipt_REJECTis that its behaviour will become inconsistent with that oficmp.c, in that it doesn't use sysctl for setting rlimit (not that it ever has...)
It doesn't has to be consistent with icmp.c, but it should be consistent with the remainder of iptables, this means to do as the ruleset says.
Quoted 4 lines
After reading the relevant bits of the RFC and a bit more code, I agree that
xrlim_allow()is indeed buggered...
Quoted 9 lines
I'm far from being an expert on linux's networking internals, but it seems to me that many aspects of the operation of network stacks are configurable using sysctl variables. Not calling
icmpv4_xrlim_allow()will make the icmp ratelimit parameter a bit of an odd one out as far as ipv4 is concerned, as I think there are other ipv4 sysctl parameters who's functionality could similarly be replaced by iptables...
ipt_REJECT is different from icmp.c, it doesn't send ICMP messages in response to error conditions but because the admin said so in his ruleset. If he wants to limit it he can use the limit match.
Quoted 3 lines
I'll be happy do do up a patch on whatever is decided upon anyway...
I already removed it in my tree, but haven't committed it yet.
Regards
Patrick