Skip to contentSkip to conversation

xtables-eb for the compatibility layer

PATCH In review Giuseppe Longo 2 participants 6 messages Feb 5, 2014 – Feb 6, 2014

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover xtables-eb for the compatibility layer

Hi,
this patchset implements ebtables for the nftables compatibility layer. I would explain some points:

- The code is not built with the gcc version 4.7.3 (Gentoo 4.7.3-r1 p1.3, pie-0.5.5):
xtables_multi-nft-bridge.o:/home/giuseppe/git/iptables-nftables/iptables/nft-bridge.c:53: first defined here xtables_multi-nft-bridge.o: In function `nft_rule_to_xtables_ebt_entry':
/home/giuseppe/git/iptables-nftables/iptables/nft-bridge.c:226: multiple definition of `nft_rule_to_xtables_ebt_entry' xtables_multi-nft-bridge.o:/home/giuseppe/git/iptables-nftables/iptables/nft-bridge.c:226: first defined here xtables_multi-nft-bridge.o:(.data+0x0): multiple definition of `nft_family_ops_bridge' xtables_multi-nft-bridge.o:(.data+0x0): first defined here
collect2: error: ld returned 1 exit status

but, with the gcc version 4.8.2 (Debian 4.8.2-1) (on my vm) it works fine. I didn't figure out the reason, honestly.

- I didn't figure out how to replace some parts of codes, for istance:
from xtables-eb.c, line 928:
t = (struct ebt_u_target *)new_entry->t;

  if ((t->parse(c - t->option_offset, argv, argc, new_entry, &t->flags, &t->t))) {
	  if (ebt_errormsg[0] != '\0')
		  return -1;
	  goto check_extension;

}

/* Is it a match_option? */
for (m = ebt_matches; m; m = m->next)

	  if (m->parse(c - m->option_offset, argv, argc, new_entry, &m->flags, &m->m))
		  break;
  if (m != NULL) {
	  if (ebt_errormsg[0] != '\0')
		  return -1;
	  if (m->used == 0) {
		  ebt_add_match(new_entry, m);
		  m->used = 1;
	  }
	  goto check_extension;

}

Looking at extensions, I noticed that any xtables_match/xtables_target have parse functions. Maybe I'm wrong. Even the code from line 992 is not very clear, about final checks. I don't know if I can remove this part of code.

Thanks

Giuseppe Longo (2):
xtables: bootstrap xtables-eb for nftables
Operations for bridge family

 include/linux/netfilter_bridge.h            |   33 +
 include/linux/netfilter_bridge/ebtables.h   |  276 +++++++
 include/linux/netfilter_bridge/ethernetdb.h |   58 ++
 iptables/Makefile.am                        |    8 +-
 iptables/getethertype.c                     |  161 ++++
 iptables/nft-bridge.c                       |  384 +++++++++
 iptables/nft-shared.c                       |    3 +
 iptables/nft.c                              |   28 +
 iptables/nft.h                              |   12 +
 iptables/xtables-eb-standalone.c            |   87 +++
 iptables/xtables-eb.c                       | 1122 +++++++++++++++++++++++++++
 iptables/xtables-ebtables.h                 |   39 +
 iptables/xtables-multi.c                    |    1 +
 iptables/xtables-multi.h                    |    1 +
 14 files changed, 2210 insertions(+), 3 deletions(-)
 create mode 100644 include/linux/netfilter_bridge.h
 create mode 100644 include/linux/netfilter_bridge/ebtables.h
 create mode 100644 include/linux/netfilter_bridge/ethernetdb.h
 create mode 100644 iptables/getethertype.c
 create mode 100644 iptables/nft-bridge.c
 create mode 100644 iptables/xtables-eb-standalone.c
 create mode 100644 iptables/xtables-eb.c
 create mode 100644 iptables/xtables-ebtables.h
Signature · 1 line
1.8.1.5
#1 lore raw

PATCH 1/2 xtables: bootstrap xtables-eb for nftables

This patch bootstraps xtables-eb for the nftables compatibility layer

---
 include/linux/netfilter_bridge.h            |   33 +
 include/linux/netfilter_bridge/ebtables.h   |  276 +++++++
 include/linux/netfilter_bridge/ethernetdb.h |   58 ++
 iptables/Makefile.am                        |    6 +-
 iptables/getethertype.c                     |  161 ++++
 iptables/nft.c                              |   28 +
 iptables/nft.h                              |    9 +
 iptables/xtables-eb-standalone.c            |   87 +++
 iptables/xtables-eb.c                       | 1122 +++++++++++++++++++++++++++
 iptables/xtables-ebtables.h                 |   39 +
 iptables/xtables-multi.c                    |    1 +
 iptables/xtables-multi.h                    |    1 +
 12 files changed, 1819 insertions(+), 2 deletions(-)
 create mode 100644 include/linux/netfilter_bridge.h
 create mode 100644 include/linux/netfilter_bridge/ebtables.h
 create mode 100644 include/linux/netfilter_bridge/ethernetdb.h
 create mode 100644 iptables/getethertype.c
 create mode 100644 iptables/xtables-eb-standalone.c
 create mode 100644 iptables/xtables-eb.c
 create mode 100644 iptables/xtables-ebtables.h
include/linux/netfilter_bridge.h new file +33 -0
new file mode 100644
index 0000000..71e9299
@@ -0,0 +1,33 @@
+#ifndef __LINUX_BRIDGE_NETFILTER_H
+#define __LINUX_BRIDGE_NETFILTER_H
+
+/* bridge-specific defines for netfilter. 
+ */
+#include <limits.h>
+
+/* Bridge Hooks */
+/* After promisc drops, checksum checks. */
+#define NF_BR_PRE_ROUTING	0
+/* If the packet is destined for this box. */
+#define NF_BR_LOCAL_IN		1
+/* If the packet is destined for another interface. */
+#define NF_BR_FORWARD		2
+/* Packets coming from a local process. */
+#define NF_BR_LOCAL_OUT		3
+/* Packets about to hit the wire. */
+#define NF_BR_POST_ROUTING	4
+/* Not really a hook, but used for the ebtables broute table */
+#define NF_BR_BROUTING		5
+#define NF_BR_NUMHOOKS		6
+
+enum nf_br_hook_priorities {
+	NF_BR_PRI_FIRST = INT_MIN,
+	NF_BR_PRI_FILTER_BRIDGED = -200,
+	NF_BR_PRI_FILTER_OTHER = 200,
+	NF_BR_PRI_NAT_DST_BRIDGED = -300,
+	NF_BR_PRI_NAT_DST_OTHER = 100,
+	NF_BR_PRI_NAT_SRC = 300,
+	NF_BR_PRI_LAST = INT_MAX,
+};
+
+#endif
include/linux/netfilter_bridge/ebtables.h new file +276 -0

276 lines not on this page · raw message

include/linux/netfilter_bridge/ethernetdb.h new file +58 -0

58 lines not on this page · raw message

iptables/Makefile.am +4 -2
index 5a8c52f..be31d0e 100644
@@ -32,7 +32,9 @@ xtables_multi_SOURCES += xtables-save.c xtables-restore.c \
 			 xtables-standalone.c xtables.c nft.c \
 			 nft-shared.c nft-ipv4.c nft-ipv6.c nft-arp.c \
 			 xtables-config.c xtables-events.c \
-			 xtables-arp-standalone.c xtables-arp.c
+			 xtables-arp-standalone.c xtables-arp.c \
+			 getethertype.c nft-bridge.c \
+			 xtables-eb-standalone.c xtables-eb.c
 xtables_multi_LDADD   += -lmnl -lnftables ${libmnl_LIBS} ${libnftables_LIBS}
 xtables_multi_CFLAGS  += -DENABLE_NFTABLES
 # yacc and lex generate dirty code
@@ -57,7 +59,7 @@ v6_sbin_links  = ip6tables ip6tables-restore ip6tables-save
 endif
 if ENABLE_NFTABLES
 x_sbin_links  = xtables xtables-restore xtables-save xtables-config \
-		xtables-events xtables-arp
+		xtables-events xtables-arp xtables-eb
 endif
 
 iptables-extensions.8: ${srcdir}/iptables-extensions.8.tmpl ../extensions/matches.man ../extensions/targets.man
iptables/getethertype.c new file +161 -0

161 lines not on this page · raw message

iptables/nft.c +28 -0
index 01e0264..8c8c026 100644
@@ -427,6 +427,34 @@ struct builtin_table xtables_arp[TABLES_MAX] = {
 	},
 };
 
+#include <linux/netfilter_bridge.h>
+
+struct builtin_table xtables_bridge[TABLES_MAX] = {
+	[FILTER] = {
+	.name = "filter",
+	.chains = {
+			{
+				.name   = "INPUT",
+				.type   = "filter",
+				.prio   = NF_BR_PRI_FILTER_BRIDGED,
+				.hook   = NF_BR_LOCAL_IN,
+			},
+			{
+				.name   = "FORWARD",
+				.type   = "filter",
+				.prio   = NF_BR_PRI_FILTER_BRIDGED,
+				.hook   = NF_BR_FORWARD,
+			},
+			{
+				.name   = "OUTPUT",
+				.type   = "filter",
+				.prio   = NF_BR_PRI_FILTER_BRIDGED,
+				.hook   = NF_BR_LOCAL_OUT,
+			},
+		},
+	},
+};
+
 int
 nft_table_builtin_add(struct nft_handle *h, struct builtin_table *_t,
 			bool dormant)
iptables/nft.h +9 -0
index 26b60b9..2a8831e 100644
@@ -38,6 +38,7 @@ struct nft_handle {
 
 extern struct builtin_table xtables_ipv4[TABLES_MAX];
 extern struct builtin_table xtables_arp[TABLES_MAX];
+extern struct builtin_table xtables_bridge[TABLES_MAX];
 
 int mnl_talk(struct nft_handle *h, struct nlmsghdr *nlh,
 	     int (*cb)(const struct nlmsghdr *nlh, void *data),
@@ -138,6 +139,8 @@ const char *nft_strerror(int err);
 int do_commandx(struct nft_handle *h, int argc, char *argv[], char **table);
 /* For xtables-arptables.c */
 int do_commandarp(struct nft_handle *h, int argc, char *argv[], char **table);
+/* For xtables-eb.c */
+int do_commandeb(struct nft_handle *h, int argc, char *argv[], char **table);
 
 /*
  * Parse config for tables and chain helper functions
@@ -170,4 +173,10 @@ int nft_arp_rule_insert(struct nft_handle *h, const char *chain,
 
 void nft_rule_to_arpt_entry(struct nft_rule *r, struct arpt_entry *fw);
 
+/*
+ * BRIDGE
+ */
+
+struct ebt_entry;
+
 #endif
iptables/xtables-eb-standalone.c new file +87 -0

87 lines not on this page · raw message

iptables/xtables-eb.c new file +1122 -0

1,122 lines not on this page · raw message

iptables/xtables-ebtables.h new file +39 -0
new file mode 100644
index 0000000..a734574
@@ -0,0 +1,39 @@
+#ifndef _XTABLES_EBTABLES_H_
+#define _XTABLES_EBTABLES_H_
+
+#include <netinet/in.h>
+#include <linux/netfilter_bridge/ebtables.h>
+#include <linux/netfilter/x_tables.h>
+
+/* We use replace->flags, so we can't use the following values:
+ * 0x01 == OPT_COMMAND, 0x02 == OPT_TABLE, 0x100 == OPT_ZERO */
+#define LIST_N    0x04
+#define LIST_C    0x08
+#define LIST_X    0x10
+#define LIST_MAC2 0x20
+
+/* Be backwards compatible, so don't use '+' in kernel */
+#define IF_WILDCARD 1
+
+int ebt_get_mac_and_mask(const char *from, unsigned char *to, unsigned char *mask);
+
+struct xtables_ebt_entry
+{
+	unsigned int bitmask;
+	unsigned int invflags;
+	unsigned int flags;
+	uint16_t ethproto;
+	char in[IFNAMSIZ];
+	char logical_in[IFNAMSIZ];
+	char out[IFNAMSIZ];
+	char logical_out[IFNAMSIZ];
+	unsigned char sourcemac[ETH_ALEN];
+	unsigned char sourcemsk[ETH_ALEN];
+	unsigned char destmac[ETH_ALEN];
+	unsigned char destmsk[ETH_ALEN];
+	struct xtables_rule_match *matches;
+	struct xtables_target *target;
+	struct xt_counters counters;
+	const char *jumpto;
+};
+#endif
iptables/xtables-multi.c +1 -0
index 5f48735..47e16f2 100644
@@ -43,6 +43,7 @@ static const struct subcommand multi_subcommands[] = {
 	{"xtables-config",      xtables_config_main},
 	{"xtables-events",      xtables_events_main},
 	{"xtables-arp",		xtables_arp_main},
+	{"xtables-eb",		xtables_eb_main},
 #endif
 	{NULL},
 };
iptables/xtables-multi.h +1 -0
index 759e24f..66ccbba 100644
@@ -8,5 +8,6 @@ extern int xtables_restore_main(int, char **);
 extern int xtables_config_main(int, char **);
 extern int xtables_events_main(int, char **);
 extern int xtables_arp_main(int, char **);
+extern int xtables_eb_main(int, char **);
 
 #endif /* _XTABLES_MULTI_H */
Signature · 1 line
1.8.1.5
#2 lore raw

Re: [PATCH 1/2] xtables: bootstrap xtables-eb for nftables

I made these changes, if I am not wrong now is possible to parse target and match using libxtables:

iptables/xtables-eb.c +0 -0
index 90bfc65..c43e10c 100644
@@ -459,6 +459,8 @@ int do_commandeb(struct nft_handle *h, int argc,

char *argv[], char **table)

        int rulenum_end = 0;
        int ret = 0;
        struct xtables_target *t;

+ struct xtables_rule_match *matchp = NULL;
+ struct xtables_match *m = NULL;

        memset(&fw, 0, sizeof(fw));

@@ -925,19 +927,23 @@ big_iface_length:

                        continue;
                default:
                        /* Is it a target option? */

- /*t = (struct ebt_u_target *)new_entry->t;
- if ((t->parse(c - t->option_offset, argv,
argc, new_entry, &t->flags, &t->t))) {
- if (ebt_errormsg[0] != '\0')
- return -1;
+ if (fw.target) {
+ xtables_option_tpcall(c, argv,

fw.invflags, fw.target, &fw);
                                goto check_extension;

- }*/
+ }

                        /* Is it a match_option? */

- /*for (m = ebt_matches; m; m = m->next)
- if (m->parse(c - m->option_offset,
argv, argc, new_entry, &m->flags, &m->m))
- break;
+ for (matchp = fw.matches; matchp; matchp =
matchp->next) {
+ m = matchp->match;
+
+ if (matchp->completed ||
+ (m->x6_parse == NULL && m->parse == NULL))
+ continue;
+
+ xtables_option_mpcall(c, argv,
fw.invflags, m, &fw);
+ }

- if (m != NULL) {

+                       /*if (m != NULL) {
                                if (ebt_errormsg[0] != '\0')
                                        return -1;
                                if (m->used == 0) {
#3 lore raw

PATCH 2/2 Operations for bridge family

In reply to Giuseppe Longo

The following patch implements the operations for bridge family.

---
 iptables/Makefile.am  |   2 +-
 iptables/nft-bridge.c | 384 ++++++++++++++++++++++++++++++++++++++++++++++++++
 iptables/nft-shared.c |   3 +
 iptables/nft.h        |   5 +-
 4 files changed, 392 insertions(+), 2 deletions(-)
 create mode 100644 iptables/nft-bridge.c
iptables/Makefile.am +1 -1
index be31d0e..e322afb 100644
@@ -31,7 +31,7 @@ xtables_multi_SOURCES += xtables-config-parser.y xtables-config-syntax.l
 xtables_multi_SOURCES += xtables-save.c xtables-restore.c \
 			 xtables-standalone.c xtables.c nft.c \
 			 nft-shared.c nft-ipv4.c nft-ipv6.c nft-arp.c \
-			 xtables-config.c xtables-events.c \
+			 nft-bridge.c xtables-config.c xtables-events.c \
 			 xtables-arp-standalone.c xtables-arp.c \
 			 getethertype.c nft-bridge.c \
 			 xtables-eb-standalone.c xtables-eb.c
iptables/nft-bridge.c new file +384 -0

384 lines not on this page · raw message

iptables/nft-shared.c +3 -0
index e0eaa17..233011c 100644
@@ -32,6 +32,7 @@
 extern struct nft_family_ops nft_family_ops_ipv4;
 extern struct nft_family_ops nft_family_ops_ipv6;
 extern struct nft_family_ops nft_family_ops_arp;
+extern struct nft_family_ops nft_family_ops_bridge;
 
 void add_meta(struct nft_rule *r, uint32_t key)
 {
@@ -649,6 +650,8 @@ struct nft_family_ops *nft_family_ops_lookup(int family)
 		return &nft_family_ops_ipv6;
 	case NFPROTO_ARP:
 		return &nft_family_ops_arp;
+	case NFPROTO_BRIDGE:
+		return &nft_family_ops_bridge;
 	default:
 		break;
 	}
iptables/nft.h +4 -1
index 2a8831e..68f674e 100644
@@ -177,6 +177,9 @@ void nft_rule_to_arpt_entry(struct nft_rule *r, struct arpt_entry *fw);
  * BRIDGE
  */
 
-struct ebt_entry;
+#include "xtables-ebtables.h"
 
+struct xtables_ebt_entry;
+
+void nft_rule_to_xtables_ebt_entry(struct nft_rule *r, struct xtables_ebt_entry *fw);
 #endif
Signature · 1 line
1.8.1.5
#4 lore raw
In reply to Giuseppe Longo

Giuseppe Longo schreef op 5/02/2014 19:17:

Quoted 27 lines

Hi,
this patchset implements ebtables for the nftables compatibility layer. I would explain some points:
- I didn't figure out how to replace some parts of codes, for istance:

   from xtables-eb.c, line 928:
   t = (struct ebt_u_target *)new_entry->t;
   if ((t->parse(c - t->option_offset, argv, argc, new_entry, &t->flags, &t->t))) {
	  if (ebt_errormsg[0] != '\0')
		  return -1;
	  goto check_extension;
   }

   /* Is it a match_option? */
   for (m = ebt_matches; m; m = m->next)
	  if (m->parse(c - m->option_offset, argv, argc, new_entry, &m->flags, &m->m))
		  break;

   if (m != NULL) {
	  if (ebt_errormsg[0] != '\0')
		  return -1;
	  if (m->used == 0) {
		  ebt_add_match(new_entry, m);
		  m->used = 1;
	  }
	  goto check_extension;
   }
Quoted 1 line

Looking at extensions, I noticed that any xtables_match/xtables_target have parse functions. Maybe I'm wrong.

The parse functions of the targets/matches/watchers will return 0 if the command-line argument isn't one of their options. If the parse function returns a non-zero value, the command line option belongs to that target/match/watcher. So, unless you don't want support for target/watcher/match command-line arguments, you will need that code.

Quoted 1 line

Even the code from line 992 is not very clear, about final checks. I don't know if I can remove this part of code.

The code you commented out checks for loops in the call chain and runs the final_check functions of the target/matches/watchers. As an example, the ebt_ip module's final_check function makes sure the Ethernet protocol is specified to be IPv4.
I see you comment out more stuff below that. All I can tell you is that it sure is required for correct behavior in the oldschool ebtables program. Please take the time to figure this out (read the code comments, step through the code with a debugger) because I doubt this will be a smooth and stable integration process otherwise.

cheers,
Bart

#5 lore raw

Hi Bart,
thank you for your explanation, I have to figure out how to port the extensions to xtables.
It's more clearer now.

Cheers,
Giuseppe

2014-02-05 Bart De Schuymer <[email protected]>:

Quoted 60 lines

Giuseppe Longo schreef op 5/02/2014 19:17:

Quoted 30 lines

Hi,
this patchset implements ebtables for the nftables compatibility layer. I would explain some points:
- I didn't figure out how to replace some parts of codes, for istance:

   from xtables-eb.c, line 928:
   t = (struct ebt_u_target *)new_entry->t;
   if ((t->parse(c - t->option_offset, argv, argc, new_entry, &t->flags,
&t->t))) {
          if (ebt_errormsg[0] != '\0')
                  return -1;
          goto check_extension;

}

   /* Is it a match_option? */
   for (m = ebt_matches; m; m = m->next)
          if (m->parse(c - m->option_offset, argv, argc, new_entry,
&m->flags, &m->m))
                  break;
   if (m != NULL) {
          if (ebt_errormsg[0] != '\0')
                  return -1;
          if (m->used == 0) {
                  ebt_add_match(new_entry, m);
                  m->used = 1;
          }
          goto check_extension;
   }
Quoted 2 lines

Looking at extensions, I noticed that any xtables_match/xtables_target have parse functions. Maybe I'm wrong.

The parse functions of the targets/matches/watchers will return 0 if the command-line argument isn't one of their options. If the parse function returns a non-zero value, the command line option belongs to that target/match/watcher. So, unless you don't want support for target/watcher/match command-line arguments, you will need that code.

Quoted 2 lines

Even the code from line 992 is not very clear, about final checks. I don't know if I can remove this part of code.

The code you commented out checks for loops in the call chain and runs the final_check functions of the target/matches/watchers. As an example, the ebt_ip module's final_check function makes sure the Ethernet protocol is specified to be IPv4.
I see you comment out more stuff below that. All I can tell you is that it sure is required for correct behavior in the oldschool ebtables program. Please take the time to figure this out (read the code comments, step through the code with a debugger) because I doubt this will be a smooth and stable integration process otherwise.

cheers,
Bart