X-Git-Url: http://git.pld-linux.org/?a=blobdiff_plain;f=kernel-imq.patch;h=87ee89ac682277f7f3ceef3ab2c724fcfc049e2f;hb=968cc686e4754b540581e741e5ee450a69101639;hp=240ddd0c527369a55d54559feefe33aa62f8df20;hpb=deb242c8a7b0a57f97a43cd1d7edd5fc501e6839;p=packages%2Fkernel.git diff --git a/kernel-imq.patch b/kernel-imq.patch index 240ddd0c..87ee89ac 100644 --- a/kernel-imq.patch +++ b/kernel-imq.patch @@ -1,151 +1,7 @@ -diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig -index d18eb60..739a98a 100644 ---- a/drivers/net/Kconfig -+++ b/drivers/net/Kconfig -@@ -234,6 +234,125 @@ config RIONET_RX_SIZE - depends on RIONET - default "128" - -+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: https://github.com/imq/linuximq -+ -+ 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 setting 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: https://github.com/imq/linuximq -+ -+ If not sure leave the default settings alone. -+ -+config IMQ_BEHAVIOR_AA -+ bool "IMQ AA" -+ help -+ This setting 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: https://github.com/imq/linuximq -+ -+ If not sure leave the default settings alone. -+ -+config IMQ_BEHAVIOR_AB -+ bool "IMQ AB" -+ help -+ This setting 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: https://github.com/imq/linuximq -+ -+ If not sure leave the default settings alone. -+ -+config IMQ_BEHAVIOR_BA -+ bool "IMQ BA" -+ help -+ This setting 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: https://github.com/imq/linuximq -+ -+ If not sure leave the default settings alone. -+ -+config IMQ_BEHAVIOR_BB -+ bool "IMQ BB" -+ help -+ This setting 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: https://github.com/imq/linuximq -+ -+ 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 "16" -+ help -+ This setting defines how many IMQ devices will be created. -+ -+ The default value is 16. -+ -+ More information can be found at: https://github.com/imq/linuximq -+ -+ If not sure leave the default settings alone. -+ - config TUN - tristate "Universal TUN/TAP device driver support" - depends on INET -diff --git a/drivers/net/Makefile b/drivers/net/Makefile -index 900b0c5..e093402 100644 ---- a/drivers/net/Makefile -+++ b/drivers/net/Makefile -@@ -10,6 +10,7 @@ obj-$(CONFIG_IPVLAN) += ipvlan/ - obj-$(CONFIG_DUMMY) += dummy.o - obj-$(CONFIG_EQUALIZER) += eql.o - obj-$(CONFIG_IFB) += ifb.o -+obj-$(CONFIG_IMQ) += imq.o - obj-$(CONFIG_MACVLAN) += macvlan.o - obj-$(CONFIG_MACVTAP) += macvtap.o - obj-$(CONFIG_MII) += mii.o -diff --git a/drivers/net/imq.c b/drivers/net/imq.c -new file mode 100644 -index 0000000..c60929b ---- /dev/null -+++ b/drivers/net/imq.c -@@ -0,0 +1,903 @@ +diff -Naupr linux-4.10_orig/drivers/net/imq.c linux-4.10/drivers/net/imq.c +--- linux-4.10_orig/drivers/net/imq.c 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/drivers/net/imq.c 2017-02-28 19:03:58.883221583 +0700 +@@ -0,0 +1,945 @@ +/* + * Pseudo-driver for the intermediate queue device. + * @@ -158,7 +14,7 @@ index 0000000..c60929b + * + * The first version was written by Martin Devera, + * -+ * See Creditis.txt ++ * See Credits.txt + */ + +#include @@ -464,8 +320,10 @@ index 0000000..c60929b +{ + struct nf_queue_entry *entry = skb->nf_queue_entry; + ++ rcu_read_lock(); ++ + skb->nf_queue_entry = NULL; -+ dev->trans_start = jiffies; ++ netif_trans_update(dev); + + dev->stats.tx_bytes += skb->len; + dev->stats.tx_packets++; @@ -491,6 +349,7 @@ index 0000000..c60929b + dev->stats.tx_dropped++; + dev_kfree_skb(skb); + ++ rcu_read_unlock(); + return NETDEV_TX_OK; + } + @@ -503,6 +362,7 @@ index 0000000..c60929b + + nf_reinject(entry, NF_ACCEPT); + ++ rcu_read_unlock(); + return NETDEV_TX_OK; +} + @@ -697,6 +557,7 @@ index 0000000..c60929b +{ + struct sk_buff *skb_orig, *skb, *skb_shared, *skb_popd; + struct Qdisc *q; ++ struct sk_buff *to_free = NULL; + struct netdev_queue *txq; + spinlock_t *root_lock; + int users; @@ -755,7 +616,7 @@ index 0000000..c60929b + + /* backup skb->cb, as qdisc layer will overwrite it */ + skb_save_cb(skb_shared); -+ qdisc_enqueue_root(skb_shared, q); /* might kfree_skb */ ++ qdisc_enqueue_root(skb_shared, q, &to_free); /* might kfree_skb */ + if (likely(atomic_read(&skb_shared->users) == users + 1)) { + bool validate; + @@ -779,13 +640,13 @@ index 0000000..c60929b + /* Note that we validate skb (GSO, checksum, ...) outside of locks */ + if (validate) + skb_popd = validate_xmit_skb_list(skb_popd, dev); -+ ++ + if (skb_popd) { + int dummy_ret; + int cpu = smp_processor_id(); /* ok because BHs are off */ + + txq = skb_get_tx_queue(dev, skb_popd); -+ /* ++ /* + IMQ device will not be frozen or stoped, and it always be successful. + So we need not check its status and return value to accelerate. + */ @@ -832,6 +693,9 @@ index 0000000..c60929b + } + retval = -1; +out: ++ if (unlikely(to_free)) { ++ kfree_skb_list(to_free); ++ } + return retval; +} +static unsigned int imq_nf_hook(void *priv, @@ -853,6 +717,10 @@ index 0000000..c60929b + return 0; +} + ++static struct device_type imq_device_type = { ++ .name = "imq", ++}; ++ +static const struct net_device_ops imq_netdev_ops = { + .ndo_open = imq_open, + .ndo_stop = imq_close, @@ -918,6 +786,33 @@ index 0000000..c60929b + return ret; +} + ++#ifdef CONFIG_LOCKDEP ++ static struct lock_class_key imq_netdev_addr_lock_key; ++ ++ static void __init imq_dev_set_lockdep_one(struct net_device *dev, ++ struct netdev_queue *txq, void *arg) ++ { ++ /* ++ * the IMQ transmit locks can be taken recursively, ++ * for example with one IMQ rule for input- and one for ++ * output network devices in iptables! ++ * until we find a better solution ignore them. ++ */ ++ lockdep_set_novalidate_class(&txq->_xmit_lock); ++ } ++ ++ static void imq_dev_set_lockdep_class(struct net_device *dev) ++ { ++ lockdep_set_class_and_name(&dev->addr_list_lock, ++ &imq_netdev_addr_lock_key, "_xmit_addr_IMQ"); ++ netdev_for_each_tx_queue(dev, imq_dev_set_lockdep_one, NULL); ++} ++#else ++ static inline void imq_dev_set_lockdep_class(struct net_device *dev) ++ { ++ } ++#endif ++ +static int __init imq_init_one(int index) +{ + struct net_device *dev; @@ -932,10 +827,13 @@ index 0000000..c60929b + goto fail; + + dev->rtnl_link_ops = &imq_link_ops; ++ SET_NETDEV_DEVTYPE(dev, &imq_device_type); + ret = register_netdevice(dev); + if (ret < 0) + goto fail; + ++ imq_dev_set_lockdep_class(dev); ++ + return 0; +fail: + free_netdev(dev); @@ -1045,15 +943,153 @@ index 0000000..c60929b +MODULE_PARM_DESC(numqueues, "number of queues per IMQ device"); +MODULE_PARM_DESC(imq_dev_accurate_stats, "Notify if need the accurate imq device stats"); + -+MODULE_AUTHOR("http://https://github.com/imq/linuximq"); ++MODULE_AUTHOR("https://github.com/imq/linuximq"); +MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See https://github.com/imq/linuximq/wiki for more information."); +MODULE_LICENSE("GPL"); +MODULE_ALIAS_RTNL_LINK("imq"); -diff --git a/include/linux/imq.h b/include/linux/imq.h -new file mode 100644 -index 0000000..1babb09 ---- /dev/null -+++ b/include/linux/imq.h +diff -Naupr linux-4.10_orig/drivers/net/Kconfig linux-4.10/drivers/net/Kconfig +--- linux-4.10_orig/drivers/net/Kconfig 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/drivers/net/Kconfig 2017-02-28 18:44:55.978280593 +0700 +@@ -260,6 +260,125 @@ config RIONET_RX_SIZE + depends on RIONET + default "128" + ++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: https://github.com/imq/linuximq ++ ++ 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 setting 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: https://github.com/imq/linuximq ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_AA ++ bool "IMQ AA" ++ help ++ This setting 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: https://github.com/imq/linuximq ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_AB ++ bool "IMQ AB" ++ help ++ This setting 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: https://github.com/imq/linuximq ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_BA ++ bool "IMQ BA" ++ help ++ This setting 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: https://github.com/imq/linuximq ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_BB ++ bool "IMQ BB" ++ help ++ This setting 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: https://github.com/imq/linuximq ++ ++ 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 "16" ++ help ++ This setting defines how many IMQ devices will be created. ++ ++ The default value is 16. ++ ++ More information can be found at: https://github.com/imq/linuximq ++ ++ If not sure leave the default settings alone. ++ + config TUN + tristate "Universal TUN/TAP device driver support" + depends on INET +diff -Naupr linux-4.10_orig/drivers/net/Makefile linux-4.10/drivers/net/Makefile +--- linux-4.10_orig/drivers/net/Makefile 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/drivers/net/Makefile 2017-02-28 18:44:55.978280593 +0700 +@@ -11,6 +11,7 @@ obj-$(CONFIG_DUMMY) += dummy.o + obj-$(CONFIG_EQUALIZER) += eql.o + obj-$(CONFIG_IFB) += ifb.o + obj-$(CONFIG_MACSEC) += macsec.o ++obj-$(CONFIG_IMQ) += imq.o + obj-$(CONFIG_MACVLAN) += macvlan.o + obj-$(CONFIG_MACVTAP) += macvtap.o + obj-$(CONFIG_MII) += mii.o +diff -Naupr linux-4.10_orig/include/linux/imq.h linux-4.10/include/linux/imq.h +--- linux-4.10_orig/include/linux/imq.h 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/include/linux/imq.h 2017-02-28 18:44:55.978280593 +0700 @@ -0,0 +1,13 @@ +#ifndef _IMQ_H +#define _IMQ_H @@ -1068,11 +1104,10 @@ index 0000000..1babb09 + +#endif /* _IMQ_H */ + -diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h -index 2d15e38..ebbecce 100644 ---- a/include/linux/netdevice.h -+++ b/include/linux/netdevice.h -@@ -3330,6 +3330,19 @@ static inline void netif_tx_unlock_bh(struct net_device *dev) +diff -Naupr linux-4.10_orig/include/linux/netdevice.h linux-4.10/include/linux/netdevice.h +--- linux-4.10_orig/include/linux/netdevice.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/linux/netdevice.h 2017-02-28 18:44:55.978280593 +0700 +@@ -3604,6 +3604,19 @@ static inline void netif_tx_unlock_bh(st } \ } @@ -1092,11 +1127,9 @@ index 2d15e38..ebbecce 100644 static inline void netif_tx_disable(struct net_device *dev) { unsigned int i; -diff --git a/include/linux/netfilter/xt_IMQ.h b/include/linux/netfilter/xt_IMQ.h -new file mode 100644 -index 0000000..9b07230 ---- /dev/null -+++ b/include/linux/netfilter/xt_IMQ.h +diff -Naupr linux-4.10_orig/include/linux/netfilter/xt_IMQ.h linux-4.10/include/linux/netfilter/xt_IMQ.h +--- linux-4.10_orig/include/linux/netfilter/xt_IMQ.h 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/include/linux/netfilter/xt_IMQ.h 2017-02-28 18:44:55.981613941 +0700 @@ -0,0 +1,9 @@ +#ifndef _XT_IMQ_H +#define _XT_IMQ_H @@ -1107,11 +1140,9 @@ index 0000000..9b07230 + +#endif /* _XT_IMQ_H */ + -diff --git a/include/linux/netfilter_ipv4/ipt_IMQ.h b/include/linux/netfilter_ipv4/ipt_IMQ.h -new file mode 100644 -index 0000000..7af320f ---- /dev/null -+++ b/include/linux/netfilter_ipv4/ipt_IMQ.h +diff -Naupr linux-4.10_orig/include/linux/netfilter_ipv4/ipt_IMQ.h linux-4.10/include/linux/netfilter_ipv4/ipt_IMQ.h +--- linux-4.10_orig/include/linux/netfilter_ipv4/ipt_IMQ.h 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/include/linux/netfilter_ipv4/ipt_IMQ.h 2017-02-28 18:44:55.981613941 +0700 @@ -0,0 +1,10 @@ +#ifndef _IPT_IMQ_H +#define _IPT_IMQ_H @@ -1123,11 +1154,9 @@ index 0000000..7af320f + +#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 -index 0000000..198ac01 ---- /dev/null -+++ b/include/linux/netfilter_ipv6/ip6t_IMQ.h +diff -Naupr linux-4.10_orig/include/linux/netfilter_ipv6/ip6t_IMQ.h linux-4.10/include/linux/netfilter_ipv6/ip6t_IMQ.h +--- linux-4.10_orig/include/linux/netfilter_ipv6/ip6t_IMQ.h 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/include/linux/netfilter_ipv6/ip6t_IMQ.h 2017-02-28 18:44:55.981613941 +0700 @@ -0,0 +1,10 @@ +#ifndef _IP6T_IMQ_H +#define _IP6T_IMQ_H @@ -1139,22 +1168,21 @@ index 0000000..198ac01 + +#endif /* _IP6T_IMQ_H */ + -diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h -index 2b0a30a..f8e727c 100644 ---- a/include/linux/skbuff.h -+++ b/include/linux/skbuff.h -@@ -38,6 +38,10 @@ - #include +diff -Naupr linux-4.10_orig/include/linux/skbuff.h linux-4.10/include/linux/skbuff.h +--- linux-4.10_orig/include/linux/skbuff.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/linux/skbuff.h 2017-02-28 18:44:55.981613941 +0700 +@@ -39,6 +39,10 @@ #include + #include #include +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) +#include +#endif + - /* A. Checksumming of received packets by device. - * -@@ -557,6 +561,9 @@ struct sk_buff { + /* The interface for checksum offload between the stack and networking drivers + * is as follows... +@@ -661,6 +665,9 @@ struct sk_buff { * first. This is owned by whoever has the skb queued ATM. */ char cb[48] __aligned(8); @@ -1164,7 +1192,7 @@ index 2b0a30a..f8e727c 100644 unsigned long _skb_refdst; void (*destructor)(struct sk_buff *skb); -@@ -566,6 +573,9 @@ struct sk_buff { +@@ -670,6 +677,9 @@ struct sk_buff { #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) struct nf_conntrack *nfct; #endif @@ -1174,17 +1202,17 @@ index 2b0a30a..f8e727c 100644 #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) struct nf_bridge_info *nf_bridge; #endif -@@ -633,6 +643,9 @@ struct sk_buff { - __u8 inner_protocol_type:1; - __u8 remcsum_offload:1; - /* 3 or 5 bit hole */ -+ #if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) -+ __u8 imq_flags:IMQ_F_BITS; -+ #endif +@@ -750,6 +760,9 @@ struct sk_buff { + __u8 offload_fwd_mark:1; + #endif + /* 2, 4 or 5 bit hole */ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ __u8 imq_flags:IMQ_F_BITS; ++#endif #ifdef CONFIG_NET_SCHED __u16 tc_index; /* traffic control index */ -@@ -789,6 +802,12 @@ void kfree_skb_list(struct sk_buff *segs); +@@ -910,6 +923,12 @@ void kfree_skb_list(struct sk_buff *segs void skb_tx_error(struct sk_buff *skb); void consume_skb(struct sk_buff *skb); void __kfree_skb(struct sk_buff *skb); @@ -1197,7 +1225,7 @@ index 2b0a30a..f8e727c 100644 extern struct kmem_cache *skbuff_head_cache; void kfree_skb_partial(struct sk_buff *skb, bool head_stolen); -@@ -3335,6 +3354,10 @@ static inline void __nf_copy(struct sk_buff *dst, const struct sk_buff *src, +@@ -3607,6 +3626,10 @@ static inline void __nf_copy(struct sk_b if (copy) dst->nfctinfo = src->nfctinfo; #endif @@ -1208,13 +1236,12 @@ index 2b0a30a..f8e727c 100644 #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) dst->nf_bridge = src->nf_bridge; nf_bridge_get(src->nf_bridge); -diff --git a/include/net/netfilter/nf_queue.h b/include/net/netfilter/nf_queue.h -index e863585..40904cb 100644 ---- a/include/net/netfilter/nf_queue.h -+++ b/include/net/netfilter/nf_queue.h +diff -Naupr linux-4.10_orig/include/net/netfilter/nf_queue.h linux-4.10/include/net/netfilter/nf_queue.h +--- linux-4.10_orig/include/net/netfilter/nf_queue.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/net/netfilter/nf_queue.h 2017-02-28 18:44:55.981613941 +0700 @@ -31,6 +31,12 @@ struct nf_queue_handler { - void nf_register_queue_handler(const struct nf_queue_handler *qh); - void nf_unregister_queue_handler(void); + void nf_register_queue_handler(struct net *net, const struct nf_queue_handler *qh); + void nf_unregister_queue_handler(struct net *net); void nf_reinject(struct nf_queue_entry *entry, unsigned int verdict); +void nf_queue_entry_release_refs(struct nf_queue_entry *entry); + @@ -1223,13 +1250,12 @@ index e863585..40904cb 100644 +void nf_unregister_queue_imq_handler(void); +#endif - bool nf_queue_entry_get_refs(struct nf_queue_entry *entry); + void nf_queue_entry_get_refs(struct nf_queue_entry *entry); void nf_queue_entry_release_refs(struct nf_queue_entry *entry); -diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h -index 401038d..4668849 100644 ---- a/include/net/pkt_sched.h -+++ b/include/net/pkt_sched.h -@@ -104,6 +104,8 @@ int sch_direct_xmit(struct sk_buff *skb, struct Qdisc *q, +diff -Naupr linux-4.10_orig/include/net/pkt_sched.h linux-4.10/include/net/pkt_sched.h +--- linux-4.10_orig/include/net/pkt_sched.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/net/pkt_sched.h 2017-02-28 18:44:55.981613941 +0700 +@@ -107,6 +107,8 @@ int sch_direct_xmit(struct sk_buff *skb, void __qdisc_run(struct Qdisc *q); @@ -1238,52 +1264,50 @@ index 401038d..4668849 100644 static inline void qdisc_run(struct Qdisc *q) { if (qdisc_run_begin(q)) -diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h -index 444faa8..e652d8d 100644 ---- a/include/net/sch_generic.h -+++ b/include/net/sch_generic.h -@@ -502,6 +502,12 @@ static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch) - return sch->enqueue(skb, sch); +diff -Naupr linux-4.10_orig/include/net/sch_generic.h linux-4.10/include/net/sch_generic.h +--- linux-4.10_orig/include/net/sch_generic.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/net/sch_generic.h 2017-02-28 18:44:55.981613941 +0700 +@@ -518,6 +518,13 @@ static inline int qdisc_enqueue(struct s + return sch->enqueue(skb, sch, to_free); } -+static inline int qdisc_enqueue_root(struct sk_buff *skb, struct Qdisc *sch) ++static inline int qdisc_enqueue_root(struct sk_buff *skb, struct Qdisc *sch, ++ struct sk_buff **to_free) +{ + qdisc_skb_cb(skb)->pkt_len = skb->len; -+ return qdisc_enqueue(skb, sch) & NET_XMIT_MASK; ++ return qdisc_enqueue(skb, sch, to_free) & NET_XMIT_MASK; +} + static inline bool qdisc_is_percpu_stats(const struct Qdisc *q) { return q->flags & TCQ_F_CPUSTATS; -diff --git a/include/uapi/linux/netfilter.h b/include/uapi/linux/netfilter.h -index d93f949..23fb6d1 100644 ---- a/include/uapi/linux/netfilter.h -+++ b/include/uapi/linux/netfilter.h +diff -Naupr linux-4.10_orig/include/uapi/linux/netfilter.h linux-4.10/include/uapi/linux/netfilter.h +--- linux-4.10_orig/include/uapi/linux/netfilter.h 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/include/uapi/linux/netfilter.h 2017-02-28 18:44:55.981613941 +0700 @@ -14,7 +14,8 @@ #define NF_QUEUE 3 #define NF_REPEAT 4 - #define NF_STOP 5 + #define NF_STOP 5 /* Deprecated, for userspace nf_queue compatibility. */ -#define NF_MAX_VERDICT NF_STOP +#define NF_IMQ_QUEUE 6 +#define NF_MAX_VERDICT NF_IMQ_QUEUE /* we overload the higher bits for encoding auxiliary data such as the queue * number or errno values. Not nice, but better than additional function -diff --git a/net/core/dev.c b/net/core/dev.c -index 6bb6470..13cda63 100644 ---- a/net/core/dev.c -+++ b/net/core/dev.c -@@ -136,6 +136,9 @@ - #include +diff -Naupr linux-4.10_orig/net/core/dev.c linux-4.10/net/core/dev.c +--- linux-4.10_orig/net/core/dev.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/core/dev.c 2017-02-28 18:44:55.984947288 +0700 +@@ -140,6 +140,9 @@ #include #include + #include +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) +#include +#endif #include "net-sysfs.h" -@@ -2675,7 +2678,12 @@ static int xmit_one(struct sk_buff *skb, struct net_device *dev, +@@ -2881,7 +2884,12 @@ static int xmit_one(struct sk_buff *skb, unsigned int len; int rc; @@ -1296,30 +1320,22 @@ index 6bb6470..13cda63 100644 dev_queue_xmit_nit(skb, dev); len = skb->len; -@@ -2713,6 +2721,7 @@ out: - *ret = rc; +@@ -2920,6 +2928,8 @@ out: return skb; } -+EXPORT_SYMBOL(dev_hard_start_xmit); ++EXPORT_SYMBOL_GPL(dev_hard_start_xmit); ++ static struct sk_buff *validate_xmit_vlan(struct sk_buff *skb, netdev_features_t features) -@@ -2801,6 +2810,7 @@ struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *d - } - return head; - } -+EXPORT_SYMBOL(validate_xmit_skb_list); - - static void qdisc_pkt_len_init(struct sk_buff *skb) { -diff --git a/net/core/skbuff.c b/net/core/skbuff.c -index dad4dd3..9c71959 100644 ---- a/net/core/skbuff.c -+++ b/net/core/skbuff.c -@@ -79,6 +79,87 @@ - - struct kmem_cache *skbuff_head_cache __read_mostly; +diff -Naupr linux-4.10_orig/net/core/skbuff.c linux-4.10/net/core/skbuff.c +--- linux-4.10_orig/net/core/skbuff.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/core/skbuff.c 2017-02-28 18:44:55.984947288 +0700 +@@ -82,6 +82,87 @@ struct kmem_cache *skbuff_head_cache __r static struct kmem_cache *skbuff_fclone_cache __read_mostly; + int sysctl_max_skb_frags __read_mostly = MAX_SKB_FRAGS; + EXPORT_SYMBOL(sysctl_max_skb_frags); +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) +static struct kmem_cache *skbuff_cb_store_cache __read_mostly; +#endif @@ -1404,7 +1420,7 @@ index dad4dd3..9c71959 100644 /** * skb_panic - private function for out-of-line support -@@ -643,6 +724,28 @@ static void skb_release_head_state(struct sk_buff *skb) +@@ -654,6 +735,28 @@ static void skb_release_head_state(struc WARN_ON(in_irq()); skb->destructor(skb); } @@ -1433,7 +1449,7 @@ index dad4dd3..9c71959 100644 #if IS_ENABLED(CONFIG_NF_CONNTRACK) nf_conntrack_put(skb->nfct); #endif -@@ -765,6 +868,10 @@ static void __copy_skb_header(struct sk_buff *new, const struct sk_buff *old) +@@ -843,6 +946,10 @@ static void __copy_skb_header(struct sk_ new->sp = secpath_get(old->sp); #endif __nf_copy(new, old, false); @@ -1444,7 +1460,7 @@ index dad4dd3..9c71959 100644 /* Note : this field could be in headers_start/headers_end section * It is not yet because we do not want to have a 16 bit hole -@@ -3324,6 +3431,13 @@ void __init skb_init(void) +@@ -3465,6 +3572,13 @@ void __init skb_init(void) 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); @@ -1458,11 +1474,10 @@ index dad4dd3..9c71959 100644 } /** -diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c -index 92b1aa3..e5a5dd5 100644 ---- a/net/ipv6/ip6_output.c -+++ b/net/ipv6/ip6_output.c -@@ -64,9 +64,6 @@ static int ip6_finish_output2(struct sock *sk, struct sk_buff *skb) +diff -Naupr linux-4.10_orig/net/ipv6/ip6_output.c linux-4.10/net/ipv6/ip6_output.c +--- linux-4.10_orig/net/ipv6/ip6_output.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/ipv6/ip6_output.c 2017-02-28 18:44:55.988280636 +0700 +@@ -67,9 +67,6 @@ static int ip6_finish_output2(struct net struct in6_addr *nexthop; int ret; @@ -1472,7 +1487,7 @@ index 92b1aa3..e5a5dd5 100644 if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr)) { struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb)); -@@ -143,6 +140,13 @@ int ip6_output(struct sock *sk, struct sk_buff *skb) +@@ -159,6 +156,13 @@ int ip6_output(struct net *net, struct s return 0; } @@ -1486,11 +1501,25 @@ index 92b1aa3..e5a5dd5 100644 return NF_HOOK_COND(NFPROTO_IPV6, NF_INET_POST_ROUTING, net, sk, skb, NULL, dev, ip6_finish_output, -diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig -index 3e1b4ab..8721e48 100644 ---- a/net/netfilter/Kconfig -+++ b/net/netfilter/Kconfig -@@ -784,6 +784,18 @@ config NETFILTER_XT_TARGET_LOG +diff -Naupr linux-4.10_orig/net/netfilter/core.c linux-4.10/net/netfilter/core.c +--- linux-4.10_orig/net/netfilter/core.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/netfilter/core.c 2017-02-28 18:44:55.988280636 +0700 +@@ -318,6 +318,11 @@ int nf_hook_slow(struct sk_buff *skb, st + if (ret == 0) + ret = -EPERM; + return ret; ++ case NF_IMQ_QUEUE: ++ ret = nf_queue(skb, state, &entry, verdict); ++ if (ret == -ECANCELED) ++ continue; ++ return ret; + case NF_QUEUE: + ret = nf_queue(skb, state, &entry, verdict); + if (ret == 1 && entry) +diff -Naupr linux-4.10_orig/net/netfilter/Kconfig linux-4.10/net/netfilter/Kconfig +--- linux-4.10_orig/net/netfilter/Kconfig 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/netfilter/Kconfig 2017-02-28 18:44:55.988280636 +0700 +@@ -852,6 +852,18 @@ config NETFILTER_XT_TARGET_LOG To compile it as a module, choose M here. If unsure, say N. @@ -1509,11 +1538,10 @@ index 3e1b4ab..8721e48 100644 config NETFILTER_XT_TARGET_MARK tristate '"MARK" target support' depends on NETFILTER_ADVANCED -diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile -index 70d026d..5469b14 100644 ---- a/net/netfilter/Makefile -+++ b/net/netfilter/Makefile -@@ -110,6 +110,7 @@ obj-$(CONFIG_NETFILTER_XT_TARGET_CT) += xt_CT.o +diff -Naupr linux-4.10_orig/net/netfilter/Makefile linux-4.10/net/netfilter/Makefile +--- linux-4.10_orig/net/netfilter/Makefile 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/netfilter/Makefile 2017-02-28 18:44:55.988280636 +0700 +@@ -125,6 +125,7 @@ obj-$(CONFIG_NETFILTER_XT_TARGET_CT) += obj-$(CONFIG_NETFILTER_XT_TARGET_DSCP) += xt_DSCP.o obj-$(CONFIG_NETFILTER_XT_TARGET_HL) += xt_HL.o obj-$(CONFIG_NETFILTER_XT_TARGET_HMARK) += xt_HMARK.o @@ -1521,44 +1549,12 @@ index 70d026d..5469b14 100644 obj-$(CONFIG_NETFILTER_XT_TARGET_LED) += xt_LED.o obj-$(CONFIG_NETFILTER_XT_TARGET_LOG) += xt_LOG.o obj-$(CONFIG_NETFILTER_XT_TARGET_NETMAP) += xt_NETMAP.o -diff --git a/net/netfilter/core.c b/net/netfilter/core.c -index 8e47f81..fa4ddd3 100644 ---- a/net/netfilter/core.c -+++ b/net/netfilter/core.c -@@ -309,9 +309,11 @@ next_hook: - ret = NF_DROP_GETERR(verdict); - if (ret == 0) - ret = -EPERM; -- } else if ((verdict & NF_VERDICT_MASK) == NF_QUEUE) { -+ } else if ((verdict & NF_VERDICT_MASK) == NF_QUEUE || -+ (verdict & NF_VERDICT_MASK) == NF_IMQ_QUEUE) { - int err = nf_queue(skb, elem, state, -- verdict >> NF_VERDICT_QBITS); -+ verdict >> NF_VERDICT_QBITS, -+ verdict & NF_VERDICT_MASK); - if (err < 0) { - if (err == -ECANCELED) - goto next_hook; -diff --git a/net/netfilter/nf_internals.h b/net/netfilter/nf_internals.h -index 0655225..25d4141 100644 ---- a/net/netfilter/nf_internals.h -+++ b/net/netfilter/nf_internals.h -@@ -18,7 +18,7 @@ unsigned int nf_iterate(struct list_head *head, struct sk_buff *skb, - - /* nf_queue.c */ - int nf_queue(struct sk_buff *skb, struct nf_hook_ops *elem, -- struct nf_hook_state *state, unsigned int queuenum); -+ struct nf_hook_state *state, unsigned int queuenum, unsigned int queuetype); - void nf_queue_nf_hook_drop(struct net *net, struct nf_hook_ops *ops); - int __init netfilter_queue_init(void); - -diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c -index 96777f9..c42d671 100644 ---- a/net/netfilter/nf_queue.c -+++ b/net/netfilter/nf_queue.c -@@ -28,6 +28,23 @@ +diff -Naupr linux-4.10_orig/net/netfilter/nf_queue.c linux-4.10/net/netfilter/nf_queue.c +--- linux-4.10_orig/net/netfilter/nf_queue.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/netfilter/nf_queue.c 2017-02-28 18:44:55.988280636 +0700 +@@ -27,6 +27,23 @@ + * receives, no matter what. */ - static const struct nf_queue_handler __rcu *queue_handler __read_mostly; +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) +static const struct nf_queue_handler __rcu *queue_imq_handler __read_mostly; @@ -1579,53 +1575,68 @@ index 96777f9..c42d671 100644 + /* return EBUSY when somebody else is registered, return EEXIST if the * same handler is registered, return 0 in case of success. */ - void nf_register_queue_handler(const struct nf_queue_handler *qh) -@@ -123,7 +140,8 @@ void nf_queue_nf_hook_drop(struct net *net, struct nf_hook_ops *ops) - int nf_queue(struct sk_buff *skb, - struct nf_hook_ops *elem, - struct nf_hook_state *state, -- unsigned int queuenum) -+ unsigned int queuenum, -+ unsigned int queuetype) + void nf_register_queue_handler(struct net *net, const struct nf_queue_handler *qh) +@@ -108,16 +125,28 @@ void nf_queue_nf_hook_drop(struct net *n + } + + static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, +- struct nf_hook_entry *hook_entry, unsigned int queuenum) ++ struct nf_hook_entry *hook_entry, unsigned int verdict) { int status = -ENOENT; struct nf_queue_entry *entry = NULL; -@@ -133,7 +151,17 @@ int nf_queue(struct sk_buff *skb, + const struct nf_afinfo *afinfo; const struct nf_queue_handler *qh; + struct net *net = state->net; ++ unsigned int queuetype = verdict & NF_VERDICT_MASK; ++ unsigned int queuenum = verdict >> NF_VERDICT_QBITS; /* QUEUE == DROP if no one is waiting, to be safe. */ -- qh = rcu_dereference(queue_handler); +- qh = rcu_dereference(net->nf.queue_handler); + if (queuetype == NF_IMQ_QUEUE) { +#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) -+ qh = rcu_dereference(queue_imq_handler); ++ qh = rcu_dereference(queue_imq_handler); +#else -+ BUG(); -+ goto err_unlock; ++ BUG(); ++ goto err_unlock; +#endif + } else { -+ qh = rcu_dereference(queue_handler); ++ qh = rcu_dereference(net->nf.queue_handler); + } + if (!qh) { status = -ESRCH; - goto err_unlock; -@@ -219,8 +247,10 @@ void nf_reinject(struct nf_queue_entry *entry, unsigned int verdict) + goto err; +@@ -164,8 +193,17 @@ int nf_queue(struct sk_buff *skb, struct + struct nf_hook_entry *entry = *entryp; + int ret; + +- ret = __nf_queue(skb, state, entry, verdict >> NF_VERDICT_QBITS); ++ ret = __nf_queue(skb, state, entry, verdict); + if (ret < 0) { ++ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ /* IMQ Bypass */ ++ if (ret == -ECANCELED && skb->imq_flags == 0) { ++ *entryp = rcu_dereference(entry->next); ++ return 1; ++ } ++#endif ++ + if (ret == -ESRCH && + (verdict & NF_VERDICT_FLAG_QUEUE_BYPASS)) { + *entryp = rcu_dereference(entry->next); +@@ -232,6 +270,7 @@ okfn: local_bh_enable(); break; case NF_QUEUE: + case NF_IMQ_QUEUE: - err = nf_queue(skb, elem, &entry->state, -- verdict >> NF_VERDICT_QBITS); -+ verdict >> NF_VERDICT_QBITS, -+ verdict & NF_VERDICT_MASK); - if (err < 0) { - if (err == -ECANCELED) - goto next_hook; -diff --git a/net/netfilter/xt_IMQ.c b/net/netfilter/xt_IMQ.c -new file mode 100644 -index 0000000..86d7b84 ---- /dev/null -+++ b/net/netfilter/xt_IMQ.c + err = nf_queue(skb, &entry->state, &hook_entry, verdict); + if (err == 1) { + if (hook_entry) +diff -Naupr linux-4.10_orig/net/netfilter/xt_IMQ.c linux-4.10/net/netfilter/xt_IMQ.c +--- linux-4.10_orig/net/netfilter/xt_IMQ.c 1970-01-01 07:00:00.000000000 +0700 ++++ linux-4.10/net/netfilter/xt_IMQ.c 2017-02-28 18:44:55.988280636 +0700 @@ -0,0 +1,72 @@ +/* + * This target marks packets to be enqueued to an imq device @@ -1693,17 +1704,16 @@ index 0000000..86d7b84 +module_init(imq_init); +module_exit(imq_fini); + -+MODULE_AUTHOR("http://https://github.com/imq/linuximq"); ++MODULE_AUTHOR("https://github.com/imq/linuximq"); +MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See https://github.com/imq/linuximq/wiki for more information."); +MODULE_LICENSE("GPL"); +MODULE_ALIAS("ipt_IMQ"); +MODULE_ALIAS("ip6t_IMQ"); + -diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c -index cb5d4ad..0d818be 100644 ---- a/net/sched/sch_generic.c -+++ b/net/sched/sch_generic.c -@@ -108,6 +108,14 @@ static struct sk_buff *dequeue_skb(struct Qdisc *q, bool *validate, +diff -Naupr linux-4.10_orig/net/sched/sch_generic.c linux-4.10/net/sched/sch_generic.c +--- linux-4.10_orig/net/sched/sch_generic.c 2017-02-20 05:34:00.000000000 +0700 ++++ linux-4.10/net/sched/sch_generic.c 2017-02-28 18:44:55.988280636 +0700 +@@ -154,6 +154,14 @@ bulk: return skb; } @@ -1715,6 +1725,6 @@ index cb5d4ad..0d818be 100644 +} +EXPORT_SYMBOL(qdisc_dequeue_skb); + - static inline int handle_dev_cpu_collision(struct sk_buff *skb, - struct netdev_queue *dev_queue, - struct Qdisc *q) + /* + * Transmit possibly several skbs, and handle the return status as + * required. Owning running seqcount bit guarantees that