| Message ID | 1279811939.2467.79.camel@edumazet-laptop |
|---|---|
| State | Not Applicable, archived |
| Delegated to: | David Miller |
| Headers | show |
On Thursday 2010-07-22 17:18, Eric Dumazet wrote: >Le jeudi 22 juillet 2010 à 16:19 +0200, Jan Engelhardt a écrit : >> On Thursday 2010-07-22 16:03, Eric Dumazet wrote: >> >> >This match is a bit strange, being packet content agnostic... >> >+/* >> >+ * Yes, packet content is not interesting for us, we only take care >> >+ * of cpu handling this packet >> >+ */ >> >> That is not so strange after all, we have many packet agnostic matches: >> xt_time, xt_condition, xt_IDLETIMER, xt_iface. >> So this little comment looks a bit redundant. >> >> Or it seems that academia can't come up with enough new protocols in time that >> we have to resort to do -m coffeemaker :) >> >> >@@ -0,0 +1,8 @@ >> >+#ifndef _XT_CPU_H >> >+#define _XT_CPU_H >> >+ >> >+struct xt_cpu_info { >> >+ unsigned int cpu; >> >+ int invert; >> >+}; >> >+#endif /*_XT_MAC_H*/ >> >> Please take a read in "Writing Netfilter Modules" e-book :-) >> It will tell you that types other than fixed ones are a no-no. > >Ok, let's do that, but I doubt sizeof(int) can be different than 4 on a >Linux 2.6 host right now. Never say never. "long" already bit people in the past, and now we have that CONFIG_COMPAT stuff. If invert is the only flag, perhaps it makes sense to use __u8 for it. >I prefer not doing the !!info->invert, and do the check only once. >+static int cpu_mt_check(const struct xt_mtchk_param *par) >+{ >+ const struct xt_cpu_info *info = par->matchinfo; >+ >+ if (info->invert & ~1) >+ return -EINVAL; >+ return 0; >+} >+ >+static bool cpu_mt(const struct sk_buff *skb, struct xt_action_param *par) >+{ >+ const struct xt_cpu_info *info = par->matchinfo; >+ >+ return (info->cpu == smp_processor_id()) ^ info->invert; >+} That works nicely indeed. Do you anticipate any future flags? -- 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
Le jeudi 22 juillet 2010 à 17:39 +0200, Jan Engelhardt a écrit : > Never say never. "long" already bit people in the past, and now we > have that CONFIG_COMPAT stuff. > I know pretty well the "long" problem, I received one of the first alpha machine ever built in the world (DEC 3000 AXP, with a fast 133 MHz cpu ;) ), before I began to use Linux :) > If invert is the only flag, perhaps it makes sense to use __u8 > for it. > Quite frankly it brings more problems than plain u32 - Possible security problems (padding bytes). Not applicable to iptables. - Some arches have slow byte/short accesses (21064 for example :) ) "int" is the natural type, fast on all arches. - Given alignment requirements of iptables rules, using less than 32bits here saves no ram. But I dont care that much. I even see compiler doesnt want to use a XOR instruction : 00000018 <cpu_mt>: 18: 55 push %ebp 19: 8b 42 04 mov 0x4(%edx),%eax 1c: 64 8b 15 00 00 00 00 mov %fs:0x0,%edx 23: 89 e5 mov %esp,%ebp 25: 5d pop %ebp 26: 39 10 cmp %edx,(%eax) 28: 0f 94 c2 sete %dl 2b: 0f b6 d2 movzbl %dl,%edx 2e: 3b 50 04 cmp 0x4(%eax),%edx 31: 0f 95 c0 setne %al 34: c3 ret -- 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
Am 22.07.2010 17:18, schrieb Eric Dumazet: > [PATCH nf-next-2.6] netfilter: add xt_cpu match > > In some situations a CPU match permits a better spreading of > connections, or select targets only for a given cpu. > > With Remote Packet Steering or multiqueue NIC and appropriate IRQ > affinities, we can distribute trafic on available cpus, per session. > (all RX packets for a given flow is handled by a given cpu) > > Some legacy applications being not SMP friendly, one way to scale a > server is to run multiple copies of them. > > Instead of randomly choosing an instance, we can use the cpu number as a > key so that softirq handler for a whole instance is running on a single > cpu, maximizing cache effects in TCP/UDP stacks. > > Using NAT for example, a four ways machine might run four copies of > server application, using a separate listening port for each instance, > but still presenting an unique external port : > > iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 0 \ > -j REDIRECT --to-port 8080 > > iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 1 \ > -j REDIRECT --to-port 8081 > > iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 2 \ > -j REDIRECT --to-port 8082 > > iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 3 \ > -j REDIRECT --to-port 8083 > Applied, thanks Eric. -- 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/include/linux/netfilter/Kbuild b/include/linux/netfilter/Kbuild index bb103f4..1041a1d 100644 --- a/include/linux/netfilter/Kbuild +++ b/include/linux/netfilter/Kbuild @@ -19,12 +19,13 @@ header-y += xt_TCPMSS.h header-y += xt_TCPOPTSTRIP.h header-y += xt_TEE.h header-y += xt_TPROXY.h +header-y += xt_cluster.h header-y += xt_comment.h header-y += xt_connbytes.h header-y += xt_connlimit.h header-y += xt_connmark.h header-y += xt_conntrack.h -header-y += xt_cluster.h +header-y += xt_cpu.h header-y += xt_dccp.h header-y += xt_dscp.h header-y += xt_esp.h diff --git a/include/linux/netfilter/xt_cpu.h b/include/linux/netfilter/xt_cpu.h index e69de29..93c7f11 100644 --- a/include/linux/netfilter/xt_cpu.h +++ b/include/linux/netfilter/xt_cpu.h @@ -0,0 +1,11 @@ +#ifndef _XT_CPU_H +#define _XT_CPU_H + +#include <linux/types.h> + +struct xt_cpu_info { + __u32 cpu; + __u32 invert; +}; + +#endif /*_XT_CPU_H*/ diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig index aa2f106..523e8d0 100644 --- a/net/netfilter/Kconfig +++ b/net/netfilter/Kconfig @@ -647,6 +647,15 @@ config NETFILTER_XT_MATCH_CONNTRACK To compile it as a module, choose M here. If unsure, say N. +config NETFILTER_XT_MATCH_CPU + tristate '"cpu" match support' + depends on NETFILTER_ADVANCED + help + CPU matching allows you to match packets based on the CPU + currently handling the packet. + + To compile it as a module, choose M here. If unsure, say N. + config NETFILTER_XT_MATCH_DCCP tristate '"dccp" protocol match support' depends on NETFILTER_ADVANCED diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile index e28420a..6da84c3 100644 --- a/net/netfilter/Makefile +++ b/net/netfilter/Makefile @@ -69,6 +69,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_COMMENT) += xt_comment.o obj-$(CONFIG_NETFILTER_XT_MATCH_CONNBYTES) += xt_connbytes.o obj-$(CONFIG_NETFILTER_XT_MATCH_CONNLIMIT) += xt_connlimit.o obj-$(CONFIG_NETFILTER_XT_MATCH_CONNTRACK) += xt_conntrack.o +obj-$(CONFIG_NETFILTER_XT_MATCH_CPU) += xt_cpu.o obj-$(CONFIG_NETFILTER_XT_MATCH_DCCP) += xt_dccp.o obj-$(CONFIG_NETFILTER_XT_MATCH_DSCP) += xt_dscp.o obj-$(CONFIG_NETFILTER_XT_MATCH_ESP) += xt_esp.o diff --git a/net/netfilter/xt_cpu.c b/net/netfilter/xt_cpu.c index e69de29..b39db8a 100644 --- a/net/netfilter/xt_cpu.c +++ b/net/netfilter/xt_cpu.c @@ -0,0 +1,63 @@ +/* Kernel module to match running CPU */ + +/* + * Might be used to distribute connections on several daemons, if + * RPS (Remote Packet Steering) is enabled or NIC is multiqueue capable, + * each RX queue IRQ affined to one CPU (1:1 mapping) + * + */ + +/* (C) 2010 Eric Dumazet + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/module.h> +#include <linux/skbuff.h> +#include <linux/netfilter/xt_cpu.h> +#include <linux/netfilter/x_tables.h> + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Eric Dumazet <eric.dumazet@gmail.com>"); +MODULE_DESCRIPTION("Xtables: CPU match"); + +static int cpu_mt_check(const struct xt_mtchk_param *par) +{ + const struct xt_cpu_info *info = par->matchinfo; + + if (info->invert & ~1) + return -EINVAL; + return 0; +} + +static bool cpu_mt(const struct sk_buff *skb, struct xt_action_param *par) +{ + const struct xt_cpu_info *info = par->matchinfo; + + return (info->cpu == smp_processor_id()) ^ info->invert; +} + +static struct xt_match cpu_mt_reg __read_mostly = { + .name = "cpu", + .revision = 0, + .family = NFPROTO_UNSPEC, + .checkentry = cpu_mt_check, + .match = cpu_mt, + .matchsize = sizeof(struct xt_cpu_info), + .me = THIS_MODULE, +}; + +static int __init cpu_mt_init(void) +{ + return xt_register_match(&cpu_mt_reg); +} + +static void __exit cpu_mt_exit(void) +{ + xt_unregister_match(&cpu_mt_reg); +} + +module_init(cpu_mt_init); +module_exit(cpu_mt_exit);