net/sched: Retire ipt action
authorJamal Hadi Salim <jhs@mojatatu.com>
Thu, 21 Dec 2023 21:31:03 +0000 (16:31 -0500)
committerDavid S. Miller <davem@davemloft.net>
Tue, 2 Jan 2024 12:41:16 +0000 (12:41 +0000)
The tc ipt action was intended to run all netfilter/iptables target.
Unfortunately it has not benefitted over the years from proper updates when
netfilter changes, and for that reason it has remained rudimentary.
Pinging a bunch of people that i was aware were using this indicates that
removing it wont affect them.
Retire it to reduce maintenance efforts. Buh-bye.

Reviewed-by: Victor Noguiera <victor@mojatatu.com>
Reviewed-by: Pedro Tammela <pctammela@mojatatu.com>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/net/tc_act/tc_ipt.h [deleted file]
include/net/tc_wrapper.h
include/uapi/linux/pkt_cls.h
include/uapi/linux/tc_act/tc_ipt.h [deleted file]
net/sched/Makefile
net/sched/act_ipt.c [deleted file]
tools/testing/selftests/tc-testing/config
tools/testing/selftests/tc-testing/tdc.sh

diff --git a/include/net/tc_act/tc_ipt.h b/include/net/tc_act/tc_ipt.h
deleted file mode 100644 (file)
index 4225fcb..0000000
+++ /dev/null
@@ -1,17 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef __NET_TC_IPT_H
-#define __NET_TC_IPT_H
-
-#include <net/act_api.h>
-
-struct xt_entry_target;
-
-struct tcf_ipt {
-       struct tc_action        common;
-       u32                     tcfi_hook;
-       char                    *tcfi_tname;
-       struct xt_entry_target  *tcfi_t;
-};
-#define to_ipt(a) ((struct tcf_ipt *)a)
-
-#endif /* __NET_TC_IPT_H */
index a6d481b5bcbcf308cc3dc27ff5a96ec3dc9b35c4..a608546bcefcf81fa9056dfc567f985c8fccb9b9 100644 (file)
@@ -117,10 +117,6 @@ static inline int tc_act(struct sk_buff *skb, const struct tc_action *a,
        if (a->ops->act == tcf_ife_act)
                return tcf_ife_act(skb, a, res);
 #endif
-#if IS_BUILTIN(CONFIG_NET_ACT_IPT)
-       if (a->ops->act == tcf_ipt_act)
-               return tcf_ipt_act(skb, a, res);
-#endif
 #if IS_BUILTIN(CONFIG_NET_ACT_SIMP)
        if (a->ops->act == tcf_simp_act)
                return tcf_simp_act(skb, a, res);
index c7082cc60d21672409a885fd01eeb62b36c10a1f..2fec9b51d28d66fdbb7e8c7d2b02632b330cfec2 100644 (file)
@@ -99,7 +99,7 @@ enum {
  * versions.
  */
 #define TCA_ACT_GACT 5
-#define TCA_ACT_IPT 6
+#define TCA_ACT_IPT 6 /* obsoleted, can be reused */
 #define TCA_ACT_PEDIT 7
 #define TCA_ACT_MIRRED 8
 #define TCA_ACT_NAT 9
@@ -120,7 +120,7 @@ enum tca_id {
        TCA_ID_UNSPEC = 0,
        TCA_ID_POLICE = 1,
        TCA_ID_GACT = TCA_ACT_GACT,
-       TCA_ID_IPT = TCA_ACT_IPT,
+       TCA_ID_IPT = TCA_ACT_IPT, /* Obsoleted, can be reused */
        TCA_ID_PEDIT = TCA_ACT_PEDIT,
        TCA_ID_MIRRED = TCA_ACT_MIRRED,
        TCA_ID_NAT = TCA_ACT_NAT,
diff --git a/include/uapi/linux/tc_act/tc_ipt.h b/include/uapi/linux/tc_act/tc_ipt.h
deleted file mode 100644 (file)
index c48d7da..0000000
+++ /dev/null
@@ -1,20 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#ifndef __LINUX_TC_IPT_H
-#define __LINUX_TC_IPT_H
-
-#include <linux/pkt_cls.h>
-
-enum {
-       TCA_IPT_UNSPEC,
-       TCA_IPT_TABLE,
-       TCA_IPT_HOOK,
-       TCA_IPT_INDEX,
-       TCA_IPT_CNT,
-       TCA_IPT_TM,
-       TCA_IPT_TARG,
-       TCA_IPT_PAD,
-       __TCA_IPT_MAX
-};
-#define TCA_IPT_MAX (__TCA_IPT_MAX - 1)
-                                                                                
-#endif
index b5fd49641d9103b14242585990b98b6665ba13f6..82c3f78ca486ee700b1ce0127ddda40229212977 100644 (file)
@@ -13,7 +13,6 @@ obj-$(CONFIG_NET_ACT_POLICE)  += act_police.o
 obj-$(CONFIG_NET_ACT_GACT)     += act_gact.o
 obj-$(CONFIG_NET_ACT_MIRRED)   += act_mirred.o
 obj-$(CONFIG_NET_ACT_SAMPLE)   += act_sample.o
-obj-$(CONFIG_NET_ACT_IPT)      += act_ipt.o
 obj-$(CONFIG_NET_ACT_NAT)      += act_nat.o
 obj-$(CONFIG_NET_ACT_PEDIT)    += act_pedit.o
 obj-$(CONFIG_NET_ACT_SIMP)     += act_simple.o
diff --git a/net/sched/act_ipt.c b/net/sched/act_ipt.c
deleted file mode 100644 (file)
index 598d6e2..0000000
+++ /dev/null
@@ -1,464 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/*
- * net/sched/act_ipt.c         iptables target interface
- *
- *TODO: Add other tables. For now we only support the ipv4 table targets
- *
- * Copyright:  Jamal Hadi Salim (2002-13)
- */
-
-#include <linux/types.h>
-#include <linux/kernel.h>
-#include <linux/string.h>
-#include <linux/errno.h>
-#include <linux/skbuff.h>
-#include <linux/rtnetlink.h>
-#include <linux/module.h>
-#include <linux/init.h>
-#include <linux/slab.h>
-#include <net/netlink.h>
-#include <net/pkt_sched.h>
-#include <linux/tc_act/tc_ipt.h>
-#include <net/tc_act/tc_ipt.h>
-#include <net/tc_wrapper.h>
-#include <net/ip.h>
-
-#include <linux/netfilter_ipv4/ip_tables.h>
-
-
-static struct tc_action_ops act_ipt_ops;
-static struct tc_action_ops act_xt_ops;
-
-static int ipt_init_target(struct net *net, struct xt_entry_target *t,
-                          char *table, unsigned int hook)
-{
-       struct xt_tgchk_param par;
-       struct xt_target *target;
-       struct ipt_entry e = {};
-       int ret = 0;
-
-       target = xt_request_find_target(AF_INET, t->u.user.name,
-                                       t->u.user.revision);
-       if (IS_ERR(target))
-               return PTR_ERR(target);
-
-       t->u.kernel.target = target;
-       memset(&par, 0, sizeof(par));
-       par.net       = net;
-       par.table     = table;
-       par.entryinfo = &e;
-       par.target    = target;
-       par.targinfo  = t->data;
-       par.hook_mask = 1 << hook;
-       par.family    = NFPROTO_IPV4;
-
-       ret = xt_check_target(&par, t->u.target_size - sizeof(*t), 0, false);
-       if (ret < 0) {
-               module_put(t->u.kernel.target->me);
-               return ret;
-       }
-       return 0;
-}
-
-static void ipt_destroy_target(struct xt_entry_target *t, struct net *net)
-{
-       struct xt_tgdtor_param par = {
-               .target   = t->u.kernel.target,
-               .targinfo = t->data,
-               .family   = NFPROTO_IPV4,
-               .net      = net,
-       };
-       if (par.target->destroy != NULL)
-               par.target->destroy(&par);
-       module_put(par.target->me);
-}
-
-static void tcf_ipt_release(struct tc_action *a)
-{
-       struct tcf_ipt *ipt = to_ipt(a);
-
-       if (ipt->tcfi_t) {
-               ipt_destroy_target(ipt->tcfi_t, a->idrinfo->net);
-               kfree(ipt->tcfi_t);
-       }
-       kfree(ipt->tcfi_tname);
-}
-
-static const struct nla_policy ipt_policy[TCA_IPT_MAX + 1] = {
-       [TCA_IPT_TABLE] = { .type = NLA_STRING, .len = IFNAMSIZ },
-       [TCA_IPT_HOOK]  = NLA_POLICY_RANGE(NLA_U32, NF_INET_PRE_ROUTING,
-                                          NF_INET_NUMHOOKS),
-       [TCA_IPT_INDEX] = { .type = NLA_U32 },
-       [TCA_IPT_TARG]  = { .len = sizeof(struct xt_entry_target) },
-};
-
-static int __tcf_ipt_init(struct net *net, unsigned int id, struct nlattr *nla,
-                         struct nlattr *est, struct tc_action **a,
-                         const struct tc_action_ops *ops,
-                         struct tcf_proto *tp, u32 flags)
-{
-       struct tc_action_net *tn = net_generic(net, id);
-       bool bind = flags & TCA_ACT_FLAGS_BIND;
-       struct nlattr *tb[TCA_IPT_MAX + 1];
-       struct tcf_ipt *ipt;
-       struct xt_entry_target *td, *t;
-       char *tname;
-       bool exists = false;
-       int ret = 0, err;
-       u32 hook = 0;
-       u32 index = 0;
-
-       if (nla == NULL)
-               return -EINVAL;
-
-       err = nla_parse_nested_deprecated(tb, TCA_IPT_MAX, nla, ipt_policy,
-                                         NULL);
-       if (err < 0)
-               return err;
-
-       if (tb[TCA_IPT_INDEX] != NULL)
-               index = nla_get_u32(tb[TCA_IPT_INDEX]);
-
-       err = tcf_idr_check_alloc(tn, &index, a, bind);
-       if (err < 0)
-               return err;
-       exists = err;
-       if (exists && bind)
-               return 0;
-
-       if (tb[TCA_IPT_HOOK] == NULL || tb[TCA_IPT_TARG] == NULL) {
-               if (exists)
-                       tcf_idr_release(*a, bind);
-               else
-                       tcf_idr_cleanup(tn, index);
-               return -EINVAL;
-       }
-
-       td = (struct xt_entry_target *)nla_data(tb[TCA_IPT_TARG]);
-       if (nla_len(tb[TCA_IPT_TARG]) != td->u.target_size) {
-               if (exists)
-                       tcf_idr_release(*a, bind);
-               else
-                       tcf_idr_cleanup(tn, index);
-               return -EINVAL;
-       }
-
-       if (!exists) {
-               ret = tcf_idr_create(tn, index, est, a, ops, bind,
-                                    false, flags);
-               if (ret) {
-                       tcf_idr_cleanup(tn, index);
-                       return ret;
-               }
-               ret = ACT_P_CREATED;
-       } else {
-               if (bind)/* dont override defaults */
-                       return 0;
-
-               if (!(flags & TCA_ACT_FLAGS_REPLACE)) {
-                       tcf_idr_release(*a, bind);
-                       return -EEXIST;
-               }
-       }
-
-       err = -EINVAL;
-       hook = nla_get_u32(tb[TCA_IPT_HOOK]);
-       switch (hook) {
-       case NF_INET_PRE_ROUTING:
-               break;
-       case NF_INET_POST_ROUTING:
-               break;
-       default:
-               goto err1;
-       }
-
-       if (tb[TCA_IPT_TABLE]) {
-               /* mangle only for now */
-               if (nla_strcmp(tb[TCA_IPT_TABLE], "mangle"))
-                       goto err1;
-       }
-
-       tname = kstrdup("mangle", GFP_KERNEL);
-       if (unlikely(!tname))
-               goto err1;
-
-       t = kmemdup(td, td->u.target_size, GFP_KERNEL);
-       if (unlikely(!t))
-               goto err2;
-
-       err = ipt_init_target(net, t, tname, hook);
-       if (err < 0)
-               goto err3;
-
-       ipt = to_ipt(*a);
-
-       spin_lock_bh(&ipt->tcf_lock);
-       if (ret != ACT_P_CREATED) {
-               ipt_destroy_target(ipt->tcfi_t, net);
-               kfree(ipt->tcfi_tname);
-               kfree(ipt->tcfi_t);
-       }
-       ipt->tcfi_tname = tname;
-       ipt->tcfi_t     = t;
-       ipt->tcfi_hook  = hook;
-       spin_unlock_bh(&ipt->tcf_lock);
-       return ret;
-
-err3:
-       kfree(t);
-err2:
-       kfree(tname);
-err1:
-       tcf_idr_release(*a, bind);
-       return err;
-}
-
-static int tcf_ipt_init(struct net *net, struct nlattr *nla,
-                       struct nlattr *est, struct tc_action **a,
-                       struct tcf_proto *tp,
-                       u32 flags, struct netlink_ext_ack *extack)
-{
-       return __tcf_ipt_init(net, act_ipt_ops.net_id, nla, est,
-                             a, &act_ipt_ops, tp, flags);
-}
-
-static int tcf_xt_init(struct net *net, struct nlattr *nla,
-                      struct nlattr *est, struct tc_action **a,
-                      struct tcf_proto *tp,
-                      u32 flags, struct netlink_ext_ack *extack)
-{
-       return __tcf_ipt_init(net, act_xt_ops.net_id, nla, est,
-                             a, &act_xt_ops, tp, flags);
-}
-
-static bool tcf_ipt_act_check(struct sk_buff *skb)
-{
-       const struct iphdr *iph;
-       unsigned int nhoff, len;
-
-       if (!pskb_may_pull(skb, sizeof(struct iphdr)))
-               return false;
-
-       nhoff = skb_network_offset(skb);
-       iph = ip_hdr(skb);
-       if (iph->ihl < 5 || iph->version != 4)
-               return false;
-
-       len = skb_ip_totlen(skb);
-       if (skb->len < nhoff + len || len < (iph->ihl * 4u))
-               return false;
-
-       return pskb_may_pull(skb, iph->ihl * 4u);
-}
-
-TC_INDIRECT_SCOPE int tcf_ipt_act(struct sk_buff *skb,
-                                 const struct tc_action *a,
-                                 struct tcf_result *res)
-{
-       char saved_cb[sizeof_field(struct sk_buff, cb)];
-       int ret = 0, result = 0;
-       struct tcf_ipt *ipt = to_ipt(a);
-       struct xt_action_param par;
-       struct nf_hook_state state = {
-               .net    = dev_net(skb->dev),
-               .in     = skb->dev,
-               .hook   = ipt->tcfi_hook,
-               .pf     = NFPROTO_IPV4,
-       };
-
-       if (skb_protocol(skb, false) != htons(ETH_P_IP))
-               return TC_ACT_UNSPEC;
-
-       if (skb_unclone(skb, GFP_ATOMIC))
-               return TC_ACT_UNSPEC;
-
-       if (!tcf_ipt_act_check(skb))
-               return TC_ACT_UNSPEC;
-
-       if (state.hook == NF_INET_POST_ROUTING) {
-               if (!skb_dst(skb))
-                       return TC_ACT_UNSPEC;
-
-               state.out = skb->dev;
-       }
-
-       memcpy(saved_cb, skb->cb, sizeof(saved_cb));
-
-       spin_lock(&ipt->tcf_lock);
-
-       tcf_lastuse_update(&ipt->tcf_tm);
-       bstats_update(&ipt->tcf_bstats, skb);
-
-       /* yes, we have to worry about both in and out dev
-        * worry later - danger - this API seems to have changed
-        * from earlier kernels
-        */
-       par.state    = &state;
-       par.target   = ipt->tcfi_t->u.kernel.target;
-       par.targinfo = ipt->tcfi_t->data;
-
-       memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));
-
-       ret = par.target->target(skb, &par);
-
-       switch (ret) {
-       case NF_ACCEPT:
-               result = TC_ACT_OK;
-               break;
-       case NF_DROP:
-               result = TC_ACT_SHOT;
-               ipt->tcf_qstats.drops++;
-               break;
-       case XT_CONTINUE:
-               result = TC_ACT_PIPE;
-               break;
-       default:
-               net_notice_ratelimited("tc filter: Bogus netfilter code %d assume ACCEPT\n",
-                                      ret);
-               result = TC_ACT_OK;
-               break;
-       }
-       spin_unlock(&ipt->tcf_lock);
-
-       memcpy(skb->cb, saved_cb, sizeof(skb->cb));
-
-       return result;
-
-}
-
-static int tcf_ipt_dump(struct sk_buff *skb, struct tc_action *a, int bind,
-                       int ref)
-{
-       unsigned char *b = skb_tail_pointer(skb);
-       struct tcf_ipt *ipt = to_ipt(a);
-       struct xt_entry_target *t;
-       struct tcf_t tm;
-       struct tc_cnt c;
-
-       /* for simple targets kernel size == user size
-        * user name = target name
-        * for foolproof you need to not assume this
-        */
-
-       spin_lock_bh(&ipt->tcf_lock);
-       t = kmemdup(ipt->tcfi_t, ipt->tcfi_t->u.user.target_size, GFP_ATOMIC);
-       if (unlikely(!t))
-               goto nla_put_failure;
-
-       c.bindcnt = atomic_read(&ipt->tcf_bindcnt) - bind;
-       c.refcnt = refcount_read(&ipt->tcf_refcnt) - ref;
-       strcpy(t->u.user.name, ipt->tcfi_t->u.kernel.target->name);
-
-       if (nla_put(skb, TCA_IPT_TARG, ipt->tcfi_t->u.user.target_size, t) ||
-           nla_put_u32(skb, TCA_IPT_INDEX, ipt->tcf_index) ||
-           nla_put_u32(skb, TCA_IPT_HOOK, ipt->tcfi_hook) ||
-           nla_put(skb, TCA_IPT_CNT, sizeof(struct tc_cnt), &c) ||
-           nla_put_string(skb, TCA_IPT_TABLE, ipt->tcfi_tname))
-               goto nla_put_failure;
-
-       tcf_tm_dump(&tm, &ipt->tcf_tm);
-       if (nla_put_64bit(skb, TCA_IPT_TM, sizeof(tm), &tm, TCA_IPT_PAD))
-               goto nla_put_failure;
-
-       spin_unlock_bh(&ipt->tcf_lock);
-       kfree(t);
-       return skb->len;
-
-nla_put_failure:
-       spin_unlock_bh(&ipt->tcf_lock);
-       nlmsg_trim(skb, b);
-       kfree(t);
-       return -1;
-}
-
-static struct tc_action_ops act_ipt_ops = {
-       .kind           =       "ipt",
-       .id             =       TCA_ID_IPT,
-       .owner          =       THIS_MODULE,
-       .act            =       tcf_ipt_act,
-       .dump           =       tcf_ipt_dump,
-       .cleanup        =       tcf_ipt_release,
-       .init           =       tcf_ipt_init,
-       .size           =       sizeof(struct tcf_ipt),
-};
-
-static __net_init int ipt_init_net(struct net *net)
-{
-       struct tc_action_net *tn = net_generic(net, act_ipt_ops.net_id);
-
-       return tc_action_net_init(net, tn, &act_ipt_ops);
-}
-
-static void __net_exit ipt_exit_net(struct list_head *net_list)
-{
-       tc_action_net_exit(net_list, act_ipt_ops.net_id);
-}
-
-static struct pernet_operations ipt_net_ops = {
-       .init = ipt_init_net,
-       .exit_batch = ipt_exit_net,
-       .id   = &act_ipt_ops.net_id,
-       .size = sizeof(struct tc_action_net),
-};
-
-static struct tc_action_ops act_xt_ops = {
-       .kind           =       "xt",
-       .id             =       TCA_ID_XT,
-       .owner          =       THIS_MODULE,
-       .act            =       tcf_ipt_act,
-       .dump           =       tcf_ipt_dump,
-       .cleanup        =       tcf_ipt_release,
-       .init           =       tcf_xt_init,
-       .size           =       sizeof(struct tcf_ipt),
-};
-
-static __net_init int xt_init_net(struct net *net)
-{
-       struct tc_action_net *tn = net_generic(net, act_xt_ops.net_id);
-
-       return tc_action_net_init(net, tn, &act_xt_ops);
-}
-
-static void __net_exit xt_exit_net(struct list_head *net_list)
-{
-       tc_action_net_exit(net_list, act_xt_ops.net_id);
-}
-
-static struct pernet_operations xt_net_ops = {
-       .init = xt_init_net,
-       .exit_batch = xt_exit_net,
-       .id   = &act_xt_ops.net_id,
-       .size = sizeof(struct tc_action_net),
-};
-
-MODULE_AUTHOR("Jamal Hadi Salim(2002-13)");
-MODULE_DESCRIPTION("Iptables target actions");
-MODULE_LICENSE("GPL");
-MODULE_ALIAS("act_xt");
-
-static int __init ipt_init_module(void)
-{
-       int ret1, ret2;
-
-       ret1 = tcf_register_action(&act_xt_ops, &xt_net_ops);
-       if (ret1 < 0)
-               pr_err("Failed to load xt action\n");
-
-       ret2 = tcf_register_action(&act_ipt_ops, &ipt_net_ops);
-       if (ret2 < 0)
-               pr_err("Failed to load ipt action\n");
-
-       if (ret1 < 0 && ret2 < 0) {
-               return ret1;
-       } else
-               return 0;
-}
-
-static void __exit ipt_cleanup_module(void)
-{
-       tcf_unregister_action(&act_ipt_ops, &ipt_net_ops);
-       tcf_unregister_action(&act_xt_ops, &xt_net_ops);
-}
-
-module_init(ipt_init_module);
-module_exit(ipt_cleanup_module);
index 012aa33b341b19f9abf71bddc977fa79704aeee6..c60acba951c2031b1444e169d306ebb0a75fc9a1 100644 (file)
@@ -82,7 +82,6 @@ CONFIG_NET_ACT_GACT=m
 CONFIG_GACT_PROB=y
 CONFIG_NET_ACT_MIRRED=m
 CONFIG_NET_ACT_SAMPLE=m
-CONFIG_NET_ACT_IPT=m
 CONFIG_NET_ACT_NAT=m
 CONFIG_NET_ACT_PEDIT=m
 CONFIG_NET_ACT_SIMP=m
index 407fa53822a068c5a611a0596a0be1ae22af750e..c53ede8b730df7a3bd6c772f4a1df7e9cadc7ad0 100755 (executable)
@@ -20,7 +20,6 @@ try_modprobe act_ct
 try_modprobe act_ctinfo
 try_modprobe act_gact
 try_modprobe act_gate
-try_modprobe act_ipt
 try_modprobe act_mirred
 try_modprobe act_mpls
 try_modprobe act_nat