Brendan Cully
2009-Nov-13 01:13 UTC
[Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
# HG changeset patch # User Brendan Cully <brendan@cs.ubc.ca> # Date 1258048720 28800 # Node ID 54a559a767055a765602f9b525bb5b43c7aae67e # Parent c317c41b1ac0d2241b7f40816f34b77bf3b1b168 IMQ allows the network buffer queuing discipline to block inbound packets from the guest (normal queuing disciplines can only block outbound packets).>From http://www.linuximq.net/patchs/linux-2.6.18-imq.diffdiff --git a/drivers/net/Kconfig b/drivers/net/Kconfig --- a/drivers/net/Kconfig +++ b/drivers/net/Kconfig @@ -93,6 +93,129 @@ To compile this driver as a module, choose M here: the module will be called eql. If unsure, say N. +config IMQ + tristate "IMQ (intermediate queueing device) support" + depends on NETDEVICES && NETFILTER + ---help--- + The IMQ device(s) is used as placeholder for QoS queueing disciplines. + Every packet entering/leaving the IP stack can be directed through + the IMQ device where it''s enqueued/dequeued to the attached qdisc. + This allows you to treat network devices as classes and distribute + bandwidth among them. Iptables is used to specify through which IMQ + device, if any, packets travel. + + More information at: http://www.linuximq.net/ + + To compile this driver as a module, choose M here: the module + will be called imq. If unsure, say N. + +choice + prompt "IMQ behavior (PRE/POSTROUTING)" + depends on IMQ + default IMQ_BEHAVIOR_AB + help + + This settings defines how IMQ behaves in respect to its + hooking in PREROUTING and POSTROUTING. + + IMQ can work in any of the following ways: + + PREROUTING | POSTROUTING + -----------------|------------------- + #1 After NAT | After NAT + #2 After NAT | Before NAT + #3 Before NAT | After NAT + #4 Before NAT | Before NAT + + The default behavior is to hook before NAT on PREROUTING + and after NAT on POSTROUTING (#3). + + This settings are specially usefull when trying to use IMQ + to shape NATed clients. + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + +config IMQ_BEHAVIOR_AA + bool "IMQ AA" + help + This settings defines how IMQ behaves in respect to its + hooking in PREROUTING and POSTROUTING. + + Choosing this option will make IMQ hook like this: + + PREROUTING: After NAT + POSTROUTING: After NAT + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + +config IMQ_BEHAVIOR_AB + bool "IMQ AB" + help + This settings defines how IMQ behaves in respect to its + hooking in PREROUTING and POSTROUTING. + + Choosing this option will make IMQ hook like this: + + PREROUTING: After NAT + POSTROUTING: Before NAT + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + +config IMQ_BEHAVIOR_BA + bool "IMQ BA" + help + This settings defines how IMQ behaves in respect to its + hooking in PREROUTING and POSTROUTING. + + Choosing this option will make IMQ hook like this: + + PREROUTING: Before NAT + POSTROUTING: After NAT + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + +config IMQ_BEHAVIOR_BB + bool "IMQ BB" + help + This settings defines how IMQ behaves in respect to its + hooking in PREROUTING and POSTROUTING. + + Choosing this option will make IMQ hook like this: + + PREROUTING: Before NAT + POSTROUTING: Before NAT + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + +endchoice + +config IMQ_NUM_DEVS + + int "Number of IMQ devices" + range 2 16 + depends on IMQ + default "2" + help + + This settings defines how many IMQ devices will be + created. + + The default value is 2. + + More information can be found at: www.linuximq.net + + If not sure leave the default settings alone. + config TUN tristate "Universal TUN/TAP device driver support" select CRC32 diff --git a/drivers/net/Makefile b/drivers/net/Makefile --- a/drivers/net/Makefile +++ b/drivers/net/Makefile @@ -131,6 +131,7 @@ endif obj-$(CONFIG_DUMMY) += dummy.o +obj-$(CONFIG_IMQ) += imq.o obj-$(CONFIG_IFB) += ifb.o obj-$(CONFIG_DE600) += de600.o obj-$(CONFIG_DE620) += de620.o diff --git a/drivers/net/imq.c b/drivers/net/imq.c new file mode 100644 --- /dev/null +++ b/drivers/net/imq.c @@ -0,0 +1,405 @@ +/* + * Pseudo-driver for the intermediate queue device. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + * + * Authors: Patrick McHardy, <kaber@trash.net> + * + * The first version was written by Martin Devera, <devik@cdi.cz> + * + * Credits: Jan Rafaj <imq2t@cedric.vabo.cz> + * - Update patch to 2.4.21 + * Sebastian Strollo <sstrollo@nortelnetworks.com> + * - Fix "Dead-loop on netdevice imq"-issue + * Marcel Sebek <sebek64@post.cz> + * - Update to 2.6.2-rc1 + * + * After some time of inactivity there is a group taking care + * of IMQ again: http://www.linuximq.net + * + * + * 2004/06/30 - New version of IMQ patch to kernels <=2.6.7 including + * the following changes: + * + * - Correction of ipv6 support "+"s issue (Hasso Tepper) + * - Correction of imq_init_devs() issue that resulted in + * kernel OOPS unloading IMQ as module (Norbert Buchmuller) + * - Addition of functionality to choose number of IMQ devices + * during kernel config (Andre Correa) + * - Addition of functionality to choose how IMQ hooks on + * PRE and POSTROUTING (after or before NAT) (Andre Correa) + * - Cosmetic corrections (Norbert Buchmuller) (Andre Correa) + * + * + * 2005/12/16 - IMQ versions between 2.6.7 and 2.6.13 were + * released with almost no problems. 2.6.14-x was released + * with some important changes: nfcache was removed; After + * some weeks of trouble we figured out that some IMQ fields + * in skb were missing in skbuff.c - skb_clone and copy_skb_header. + * These functions are correctly patched by this new patch version. + * + * Thanks for all who helped to figure out all the problems with + * 2.6.14.x: Patrick McHardy, Rune Kock, VeNoMouS, Max CtRiX, + * Kevin Shanahan, Richard Lucassen, Valery Dachev (hopefully + * I didn''t forget anybody). I apologize again for my lack of time. + * + * More info at: http://www.linuximq.net/ (Andre Correa) + */ + +#include <linux/config.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/moduleparam.h> +#include <linux/skbuff.h> +#include <linux/netdevice.h> +#include <linux/rtnetlink.h> +#include <linux/if_arp.h> +#include <linux/netfilter.h> +#include <linux/netfilter_ipv4.h> +#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) + #include <linux/netfilter_ipv6.h> +#endif +#include <linux/imq.h> +#include <net/pkt_sched.h> + +extern int qdisc_restart1(struct net_device *dev); + +static nf_hookfn imq_nf_hook; + +static struct nf_hook_ops imq_ingress_ipv4 = { + .hook = imq_nf_hook, + .owner = THIS_MODULE, + .pf = PF_INET, + .hooknum = NF_IP_PRE_ROUTING, +#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) + .priority = NF_IP_PRI_MANGLE + 1 +#else + .priority = NF_IP_PRI_NAT_DST + 1 +#endif +}; + +static struct nf_hook_ops imq_egress_ipv4 = { + .hook = imq_nf_hook, + .owner = THIS_MODULE, + .pf = PF_INET, + .hooknum = NF_IP_POST_ROUTING, +#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA) + .priority = NF_IP_PRI_LAST +#else + .priority = NF_IP_PRI_NAT_SRC - 1 +#endif +}; + +#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) +static struct nf_hook_ops imq_ingress_ipv6 = { + .hook = imq_nf_hook, + .owner = THIS_MODULE, + .pf = PF_INET6, + .hooknum = NF_IP6_PRE_ROUTING, +#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) + .priority = NF_IP6_PRI_MANGLE + 1 +#else + .priority = NF_IP6_PRI_NAT_DST + 1 +#endif +}; + +static struct nf_hook_ops imq_egress_ipv6 = { + .hook = imq_nf_hook, + .owner = THIS_MODULE, + .pf = PF_INET6, + .hooknum = NF_IP6_POST_ROUTING, +#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA) + .priority = NF_IP6_PRI_LAST +#else + .priority = NF_IP6_PRI_NAT_SRC - 1 +#endif +}; +#endif + +#if defined(CONFIG_IMQ_NUM_DEVS) +static unsigned int numdevs = CONFIG_IMQ_NUM_DEVS; +#else +static unsigned int numdevs = 2; +#endif + +static struct net_device *imq_devs; + +static struct net_device_stats *imq_get_stats(struct net_device *dev) +{ + return (struct net_device_stats *)dev->priv; +} + +/* called for packets kfree''d in qdiscs at places other than enqueue */ +static void imq_skb_destructor(struct sk_buff *skb) +{ + struct nf_info *info = skb->nf_info; + + if (info) { + if (info->indev) + dev_put(info->indev); + if (info->outdev) + dev_put(info->outdev); + kfree(info); + } +} + +static int imq_dev_xmit(struct sk_buff *skb, struct net_device *dev) +{ + struct net_device_stats *stats = (struct net_device_stats*) dev->priv; + + stats->tx_bytes += skb->len; + stats->tx_packets++; + + skb->imq_flags = 0; + skb->destructor = NULL; + + dev->trans_start = jiffies; + nf_reinject(skb, skb->nf_info, NF_ACCEPT); + return 0; +} + +static int imq_nf_queue(struct sk_buff *skb, struct nf_info *info, unsigned queue_num, void *data) +{ + struct net_device *dev; + struct net_device_stats *stats; + struct sk_buff *skb2 = NULL; + struct Qdisc *q; + unsigned int index = skb->imq_flags&IMQ_F_IFMASK; + int ret = -1; + + if (index > numdevs) + return -1; + + dev = imq_devs + index; + if (!(dev->flags & IFF_UP)) { + skb->imq_flags = 0; + nf_reinject(skb, info, NF_ACCEPT); + return 0; + } + dev->last_rx = jiffies; + + if (skb->destructor) { + skb2 = skb; + skb = skb_clone(skb, GFP_ATOMIC); + if (!skb) + return -1; + } + skb->nf_info = info; + + stats = (struct net_device_stats *)dev->priv; + stats->rx_bytes+= skb->len; + stats->rx_packets++; + + spin_lock_bh(&dev->queue_lock); + q = dev->qdisc; + if (q->enqueue) { + q->enqueue(skb_get(skb), q); + if (skb_shared(skb)) { + skb->destructor = imq_skb_destructor; + kfree_skb(skb); + ret = 0; + } + } + if (spin_is_locked(&dev->_xmit_lock)) + netif_schedule(dev); + else + + while (!netif_queue_stopped(dev) && + qdisc_restart1(dev)<0) + /* NOTHING */; + + spin_unlock_bh(&dev->queue_lock); + + if (skb2) + kfree_skb(ret ? skb : skb2); + + return ret; +} + +static struct nf_queue_handler nfqh = { + .name = "imq", + .outfn = imq_nf_queue, +}; + +static unsigned int imq_nf_hook(unsigned int hook, struct sk_buff **pskb, + const struct net_device *indev, + const struct net_device *outdev, + int (*okfn)(struct sk_buff *)) +{ + if ((*pskb)->imq_flags & IMQ_F_ENQUEUE) + return NF_QUEUE; + + return NF_ACCEPT; +} + + +static int __init imq_init_hooks(void) +{ + int err; + + err = nf_register_queue_handler(PF_INET, &nfqh); + if (err > 0) + goto err1; + if ((err = nf_register_hook(&imq_ingress_ipv4))) + goto err2; + if ((err = nf_register_hook(&imq_egress_ipv4))) + goto err3; +#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) + if ((err = nf_register_queue_handler(PF_INET6, &nfqh))) + goto err4; + if ((err = nf_register_hook(&imq_ingress_ipv6))) + goto err5; + if ((err = nf_register_hook(&imq_egress_ipv6))) + goto err6; +#endif + + return 0; + +#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) +err6: + nf_unregister_hook(&imq_ingress_ipv6); +err5: + nf_unregister_queue_handler(PF_INET6); +err4: + nf_unregister_hook(&imq_egress_ipv6); +#endif +err3: + nf_unregister_hook(&imq_ingress_ipv4); +err2: + nf_unregister_queue_handler(PF_INET); +err1: + return err; +} + +static void __exit imq_unhook(void) +{ +#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) + nf_unregister_hook(&imq_ingress_ipv6); + nf_unregister_hook(&imq_egress_ipv6); + nf_unregister_queue_handler(PF_INET6); +#endif + nf_unregister_hook(&imq_ingress_ipv4); + nf_unregister_hook(&imq_egress_ipv4); + nf_unregister_queue_handler(PF_INET); +} + +static int __init imq_dev_init(struct net_device *dev) +{ + dev->hard_start_xmit = imq_dev_xmit; + dev->type = ARPHRD_VOID; + dev->mtu = 16000; + dev->tx_queue_len = 11000; + dev->flags = IFF_NOARP; + dev->priv = kmalloc(sizeof(struct net_device_stats), GFP_KERNEL); + if (dev->priv == NULL) + return -ENOMEM; + memset(dev->priv, 0, sizeof(struct net_device_stats)); + dev->get_stats = imq_get_stats; + + return 0; +} + +static void imq_dev_uninit(struct net_device *dev) +{ + kfree(dev->priv); +} + +static int __init imq_init_devs(void) +{ + struct net_device *dev; + int i,j; + j = numdevs; + + if (!numdevs || numdevs > IMQ_MAX_DEVS) { + printk(KERN_ERR "IMQ: numdevs has to be betweed 1 and %u\n", + IMQ_MAX_DEVS); + return -EINVAL; + } + + imq_devs = kmalloc(sizeof(struct net_device) * numdevs, GFP_KERNEL); + if (!imq_devs) + return -ENOMEM; + memset(imq_devs, 0, sizeof(struct net_device) * numdevs); + + /* we start counting at zero */ + numdevs--; + + for (i = 0, dev = imq_devs; i <= numdevs; i++, dev++) { + SET_MODULE_OWNER(dev); + strcpy(dev->name, "imq%d"); + dev->init = imq_dev_init; + dev->uninit = imq_dev_uninit; + + if (register_netdev(dev) < 0) + goto err_register; + } + printk(KERN_INFO "IMQ starting with %u devices...\n", j); + return 0; + +err_register: + for (; i; i--) + unregister_netdev(--dev); + kfree(imq_devs); + return -EIO; +} + +static void imq_cleanup_devs(void) +{ + int i; + struct net_device *dev = imq_devs; + + for (i = 0; i <= numdevs; i++) + unregister_netdev(dev++); + + kfree(imq_devs); +} + +static int __init imq_init_module(void) +{ + int err; + + if ((err = imq_init_devs())) { + printk(KERN_ERR "IMQ: Error trying imq_init_devs()\n"); + return err; + } + if ((err = imq_init_hooks())) { + printk(KERN_ERR "IMQ: Error trying imq_init_hooks()\n"); + imq_cleanup_devs(); + return err; + } + + printk(KERN_INFO "IMQ driver loaded successfully.\n"); + +#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) + printk(KERN_INFO "\tHooking IMQ before NAT on PREROUTING.\n"); +#else + printk(KERN_INFO "\tHooking IMQ after NAT on PREROUTING.\n"); +#endif +#if defined(CONFIG_IMQ_BEHAVIOR_AB) || defined(CONFIG_IMQ_BEHAVIOR_BB) + printk(KERN_INFO "\tHooking IMQ before NAT on POSTROUTING.\n"); +#else + printk(KERN_INFO "\tHooking IMQ after NAT on POSTROUTING.\n"); +#endif + + return 0; +} + +static void __exit imq_cleanup_module(void) +{ + imq_unhook(); + imq_cleanup_devs(); + printk(KERN_INFO "IMQ driver unloaded successfully.\n"); +} + + +module_init(imq_init_module); +module_exit(imq_cleanup_module); + +module_param(numdevs, int, 0); +MODULE_PARM_DESC(numdevs, "number of IMQ devices (how many imq* devices will be created)"); +MODULE_AUTHOR("http://www.linuximq.net"); +MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See http://www.linuximq.net/ for more information."); +MODULE_LICENSE("GPL"); diff --git a/include/linux/imq.h b/include/linux/imq.h new file mode 100644 --- /dev/null +++ b/include/linux/imq.h @@ -0,0 +1,9 @@ +#ifndef _IMQ_H +#define _IMQ_H + +#define IMQ_MAX_DEVS 16 + +#define IMQ_F_IFMASK 0x7f +#define IMQ_F_ENQUEUE 0x80 + +#endif /* _IMQ_H */ diff --git a/include/linux/netfilter_ipv4/ipt_IMQ.h b/include/linux/netfilter_ipv4/ipt_IMQ.h new file mode 100644 --- /dev/null +++ b/include/linux/netfilter_ipv4/ipt_IMQ.h @@ -0,0 +1,8 @@ +#ifndef _IPT_IMQ_H +#define _IPT_IMQ_H + +struct ipt_imq_info { + unsigned int todev; /* target imq device */ +}; + +#endif /* _IPT_IMQ_H */ diff --git a/include/linux/netfilter_ipv6/ip6t_IMQ.h b/include/linux/netfilter_ipv6/ip6t_IMQ.h new file mode 100644 --- /dev/null +++ b/include/linux/netfilter_ipv6/ip6t_IMQ.h @@ -0,0 +1,8 @@ +#ifndef _IP6T_IMQ_H +#define _IP6T_IMQ_H + +struct ip6t_imq_info { + unsigned int todev; /* target imq device */ +}; + +#endif /* _IP6T_IMQ_H */ diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -299,6 +299,10 @@ #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) struct sk_buff *nfct_reasm; #endif +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) + unsigned char imq_flags; + struct nf_info *nf_info; +#endif #ifdef CONFIG_BRIDGE_NETFILTER struct nf_bridge_info *nf_bridge; #endif diff --git a/net/core/dev.c b/net/core/dev.c --- a/net/core/dev.c +++ b/net/core/dev.c @@ -94,6 +94,9 @@ #include <linux/skbuff.h> #include <net/sock.h> #include <linux/rtnetlink.h> +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) +#include <linux/imq.h> +#endif #include <linux/proc_fs.h> #include <linux/seq_file.h> #include <linux/stat.h> @@ -1349,7 +1352,11 @@ int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev) { if (likely(!skb->next)) { - if (netdev_nit) + if (netdev_nit +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) + && !(skb->imq_flags & IMQ_F_ENQUEUE) +#endif + ) dev_queue_xmit_nit(skb, dev); if (netif_needs_gso(dev, skb)) { diff --git a/net/core/skbuff.c b/net/core/skbuff.c --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -486,6 +486,10 @@ C(nfct_reasm); nf_conntrack_get_reasm(skb->nfct_reasm); #endif +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) + C(imq_flags); + C(nf_info); +#endif /*CONFIG_IMQ*/ #ifdef CONFIG_BRIDGE_NETFILTER C(nf_bridge); nf_bridge_get(skb->nf_bridge); @@ -550,6 +554,10 @@ #if defined(CONFIG_IP_VS) || defined(CONFIG_IP_VS_MODULE) new->ipvs_property = old->ipvs_property; #endif +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) + new->imq_flags = old->imq_flags; + new->nf_info = old->nf_info; +#endif /*CONFIG_IMQ*/ #ifdef CONFIG_BRIDGE_NETFILTER new->nf_bridge = old->nf_bridge; nf_bridge_get(old->nf_bridge); diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig --- a/net/ipv4/netfilter/Kconfig +++ b/net/ipv4/netfilter/Kconfig @@ -544,6 +544,17 @@ To compile it as a module, choose M here. If unsure, say N. +config IP_NF_TARGET_IMQ + tristate "IMQ target support" + depends on IP_NF_MANGLE + help + This option adds a `IMQ'' target which is used to specify if and + to which IMQ device packets should get enqueued/dequeued. + + For more information visit: http://www.linuximq.net/ + + To compile it as a module, choose M here. If unsure, say N. + config IP_NF_TARGET_TOS tristate "TOS target support" depends on IP_NF_MANGLE diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile --- a/net/ipv4/netfilter/Makefile +++ b/net/ipv4/netfilter/Makefile @@ -69,6 +69,7 @@ obj-$(CONFIG_IP_NF_TARGET_TOS) += ipt_TOS.o obj-$(CONFIG_IP_NF_TARGET_ECN) += ipt_ECN.o obj-$(CONFIG_IP_NF_TARGET_DSCP) += ipt_DSCP.o +obj-$(CONFIG_IP_NF_TARGET_IMQ) += ipt_IMQ.o obj-$(CONFIG_IP_NF_TARGET_MASQUERADE) += ipt_MASQUERADE.o obj-$(CONFIG_IP_NF_TARGET_REDIRECT) += ipt_REDIRECT.o obj-$(CONFIG_IP_NF_TARGET_NETMAP) += ipt_NETMAP.o diff --git a/net/ipv4/netfilter/ipt_IMQ.c b/net/ipv4/netfilter/ipt_IMQ.c new file mode 100644 --- /dev/null +++ b/net/ipv4/netfilter/ipt_IMQ.c @@ -0,0 +1,83 @@ +/* + * This target marks packets to be enqueued to an imq device + */ +#include <linux/module.h> +#include <linux/skbuff.h> +#include <linux/netfilter_ipv4/ip_tables.h> +#include <linux/netfilter_ipv4/ipt_IMQ.h> +#include <linux/imq.h> + +static unsigned int imq_target(struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, + const struct xt_target *target, + const void *targinfo, + void *userdata) +{ + struct ipt_imq_info *mr = (struct ipt_imq_info*)targinfo; + + (*pskb)->imq_flags = mr->todev | IMQ_F_ENQUEUE; + + return IPT_CONTINUE; +} + +static int imq_checkentry(const char *tablename, + const void *e, + const struct xt_target *target, + void *targinfo, + unsigned int targinfosize, + unsigned int hook_mask) +{ + struct ipt_imq_info *mr; + + if (targinfosize != IPT_ALIGN(sizeof(struct ipt_imq_info))) { + printk(KERN_WARNING "IMQ: invalid targinfosize\n"); + return 0; + } + mr = (struct ipt_imq_info*)targinfo; + + if (strcmp(tablename, "mangle") != 0) { + printk(KERN_WARNING + "IMQ: IMQ can only be called from \"mangle\" table, not \"%s\"\n", + tablename); + return 0; + } + + if (mr->todev > IMQ_MAX_DEVS) { + printk(KERN_WARNING + "IMQ: invalid device specified, highest is %u\n", + IMQ_MAX_DEVS); + return 0; + } + + return 1; +} + +static struct ipt_target ipt_imq_reg = { + .name = "IMQ", + .target = imq_target, + .checkentry = imq_checkentry, + .me = THIS_MODULE, + .targetsize = sizeof(struct ipt_imq_info), +}; + +static int __init init(void) +{ + if (ipt_register_target(&ipt_imq_reg)) + return -EINVAL; + + return 0; +} + +static void __exit fini(void) +{ + ipt_unregister_target(&ipt_imq_reg); +} + +module_init(init); +module_exit(fini); + +MODULE_AUTHOR("http://www.linuximq.net"); +MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See http://www.linuximq.net/ for more information."); +MODULE_LICENSE("GPL"); diff --git a/net/ipv6/netfilter/Kconfig b/net/ipv6/netfilter/Kconfig --- a/net/ipv6/netfilter/Kconfig +++ b/net/ipv6/netfilter/Kconfig @@ -163,6 +163,15 @@ To compile it as a module, choose M here. If unsure, say N. +config IP6_NF_TARGET_IMQ + tristate "IMQ target support" + depends on IP6_NF_MANGLE + help + This option adds a `IMQ'' target which is used to specify if and + to which imq device packets should get enqueued/dequeued. + + To compile it as a module, choose M here. If unsure, say N. + config IP6_NF_TARGET_HL tristate ''HL (hoplimit) target support'' depends on IP6_NF_MANGLE diff --git a/net/ipv6/netfilter/Makefile b/net/ipv6/netfilter/Makefile --- a/net/ipv6/netfilter/Makefile +++ b/net/ipv6/netfilter/Makefile @@ -13,6 +13,7 @@ obj-$(CONFIG_IP6_NF_MATCH_OWNER) += ip6t_owner.o obj-$(CONFIG_IP6_NF_FILTER) += ip6table_filter.o obj-$(CONFIG_IP6_NF_MANGLE) += ip6table_mangle.o +obj-$(CONFIG_IP6_NF_TARGET_IMQ) += ip6t_IMQ.o obj-$(CONFIG_IP6_NF_TARGET_HL) += ip6t_HL.o obj-$(CONFIG_IP6_NF_QUEUE) += ip6_queue.o obj-$(CONFIG_IP6_NF_TARGET_LOG) += ip6t_LOG.o diff --git a/net/ipv6/netfilter/ip6t_IMQ.c b/net/ipv6/netfilter/ip6t_IMQ.c new file mode 100644 --- /dev/null +++ b/net/ipv6/netfilter/ip6t_IMQ.c @@ -0,0 +1,80 @@ +/* + * This target marks packets to be enqueued to an imq device + */ +#include <linux/module.h> +#include <linux/skbuff.h> +#include <linux/netfilter_ipv6/ip6_tables.h> +#include <linux/netfilter_ipv6/ip6t_IMQ.h> +#include <linux/imq.h> + +static unsigned int imq_target(struct sk_buff **pskb, + unsigned int hooknum, + const struct net_device *in, + const struct net_device *out, + const void *targinfo, + void *userdata) +{ + struct ip6t_imq_info *mr = (struct ip6t_imq_info*)targinfo; + + (*pskb)->imq_flags = mr->todev | IMQ_F_ENQUEUE; + + return IP6T_CONTINUE; +} + +static int imq_checkentry(const char *tablename, + const struct ip6t_entry *e, + void *targinfo, + unsigned int targinfosize, + unsigned int hook_mask) +{ + struct ip6t_imq_info *mr; + + if (targinfosize != IP6T_ALIGN(sizeof(struct ip6t_imq_info))) { + printk(KERN_WARNING "IMQ: invalid targinfosize\n"); + return 0; + } + mr = (struct ip6t_imq_info*)targinfo; + + if (strcmp(tablename, "mangle") != 0) { + printk(KERN_WARNING + "IMQ: IMQ can only be called from \"mangle\" table, not \"%s\"\n", + tablename); + return 0; + } + + if (mr->todev > IMQ_MAX_DEVS) { + printk(KERN_WARNING + "IMQ: invalid device specified, highest is %u\n", + IMQ_MAX_DEVS); + return 0; + } + + return 1; +} + +static struct ip6t_target ip6t_imq_reg = { + .name = "IMQ", + .target = imq_target, + .checkentry = imq_checkentry, + .me = THIS_MODULE +}; + +static int __init init(void) +{ + if (ip6t_register_target(&ip6t_imq_reg)) + return -EINVAL; + + return 0; +} + +static void __exit fini(void) +{ + ip6t_unregister_target(&ip6t_imq_reg); +} + +module_init(init); +module_exit(fini); + +MODULE_AUTHOR("http://www.linuximq.net"); +MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See http://www.linuximq.net/ for more information."); +MODULE_LICENSE("GPL"); diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c --- a/net/sched/sch_generic.c +++ b/net/sched/sch_generic.c @@ -87,7 +87,6 @@ NOTE: Called under dev->queue_lock with locally disabled BH. */ - static inline int qdisc_restart(struct net_device *dev) { struct Qdisc *q = dev->qdisc; @@ -181,6 +180,11 @@ return q->q.qlen; } +int qdisc_restart1(struct net_device *dev) +{ + return qdisc_restart(dev); +} + void __qdisc_run(struct net_device *dev) { if (unlikely(dev->qdisc == &noop_qdisc)) @@ -617,3 +621,4 @@ EXPORT_SYMBOL(qdisc_reset); EXPORT_SYMBOL(qdisc_lock_tree); EXPORT_SYMBOL(qdisc_unlock_tree); +EXPORT_SYMBOL(qdisc_restart1); _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2009-Nov-13 15:50 UTC
Re: [Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
On 13/11/2009 01:13, "Brendan Cully" <brendan@cs.ubc.ca> wrote:> # HG changeset patch > # User Brendan Cully <brendan@cs.ubc.ca> > # Date 1258048720 28800 > # Node ID 54a559a767055a765602f9b525bb5b43c7aae67e > # Parent c317c41b1ac0d2241b7f40816f34b77bf3b1b168 > IMQ allows the network buffer queuing discipline to block inbound > packets from the guest (normal queuing disciplines can only block > outbound packets).Didn''t apply due to broken indentation in some places. Probably your mail client mangled it, so you might try re-sending as an attachment. All other patches applied okay, but you might want to check/test that none of the new files in those patches are similarly affected. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2009-Nov-13 17:11 UTC
Re: [Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
On 13/11/2009 15:50, "Keir Fraser" <keir.fraser@eu.citrix.com> wrote:> On 13/11/2009 01:13, "Brendan Cully" <brendan@cs.ubc.ca> wrote: > >> # HG changeset patch >> # User Brendan Cully <brendan@cs.ubc.ca> >> # Date 1258048720 28800 >> # Node ID 54a559a767055a765602f9b525bb5b43c7aae67e >> # Parent c317c41b1ac0d2241b7f40816f34b77bf3b1b168 >> IMQ allows the network buffer queuing discipline to block inbound >> packets from the guest (normal queuing disciplines can only block >> outbound packets). > > Didn''t apply due to broken indentation in some places. Probably your mail > client mangled it, so you might try re-sending as an attachment. All other > patches applied okay, but you might want to check/test that none of the new > files in those patches are similarly affected.The remus Makefiles were affected, which I''ve fixed. Looks like at least remus/imqebt is also affected -- seems your tabs got somehow turned into 7 spaces. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Brendan Cully
2009-Nov-13 20:14 UTC
Re: [Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
On Friday, 13 November 2009 at 17:11, Keir Fraser wrote:> On 13/11/2009 15:50, "Keir Fraser" <keir.fraser@eu.citrix.com> wrote: > > > On 13/11/2009 01:13, "Brendan Cully" <brendan@cs.ubc.ca> wrote: > > > >> # HG changeset patch > >> # User Brendan Cully <brendan@cs.ubc.ca> > >> # Date 1258048720 28800 > >> # Node ID 54a559a767055a765602f9b525bb5b43c7aae67e > >> # Parent c317c41b1ac0d2241b7f40816f34b77bf3b1b168 > >> IMQ allows the network buffer queuing discipline to block inbound > >> packets from the guest (normal queuing disciplines can only block > >> outbound packets). > > > > Didn''t apply due to broken indentation in some places. Probably your mail > > client mangled it, so you might try re-sending as an attachment. All other > > patches applied okay, but you might want to check/test that none of the new > > files in those patches are similarly affected. > > The remus Makefiles were affected, which I''ve fixed. Looks like at least > remus/imqebt is also affected -- seems your tabs got somehow turned into 7 > spaces.I''m surprised, since I just used "hg email" to send the patches. I''ll investigate further and resend the IMQ patch. Thanks! _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Brendan Cully
2009-Nov-13 21:11 UTC
Re: [Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
On Friday, 13 November 2009 at 17:11, Keir Fraser wrote:> On 13/11/2009 15:50, "Keir Fraser" <keir.fraser@eu.citrix.com> wrote: > > > On 13/11/2009 01:13, "Brendan Cully" <brendan@cs.ubc.ca> wrote: > > > >> # HG changeset patch > >> # User Brendan Cully <brendan@cs.ubc.ca> > >> # Date 1258048720 28800 > >> # Node ID 54a559a767055a765602f9b525bb5b43c7aae67e > >> # Parent c317c41b1ac0d2241b7f40816f34b77bf3b1b168 > >> IMQ allows the network buffer queuing discipline to block inbound > >> packets from the guest (normal queuing disciplines can only block > >> outbound packets). > > > > Didn''t apply due to broken indentation in some places. Probably your mail > > client mangled it, so you might try re-sending as an attachment. All other > > patches applied okay, but you might want to check/test that none of the new > > files in those patches are similarly affected. > > The remus Makefiles were affected, which I''ve fixed. Looks like at least > remus/imqebt is also affected -- seems your tabs got somehow turned into 7 > spaces.Hmm. They looked ok when I downloaded them from the list and applied them with hg import --no-commit. Is it possible the mangling happened in your email client? Here''s IMQ for linux-2.6.18-xen.hg as an attachment. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser
2009-Nov-13 21:52 UTC
Re: [Xen-devel] [PATCH] Remus: IMQ patch for linux-2.6.18-xen.hg
On 13/11/2009 21:11, "Brendan Cully" <brendan@cs.ubc.ca> wrote:>> The remus Makefiles were affected, which I''ve fixed. Looks like at least >> remus/imqebt is also affected -- seems your tabs got somehow turned into 7 >> spaces. > > Hmm. They looked ok when I downloaded them from the list and applied > them with hg import --no-commit. Is it possible the mangling happened > in your email client? Here''s IMQ for linux-2.6.18-xen.hg as an > attachment.I use fetchmail and munpack, so not much mail client intelligence there. But still it could be on my side I suppose. Generally I prefer attachments, at least for bigger patches, as I''ve never had a problem with them. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel