d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4f0c2fae8d
freebsd-skq/sys/netinet/ip_fw.h
Luigi Rizzo de240d1013 merge code from ipfw3-head to reduce contention on the ipfw lock 
and remove all O(N) sequences from kernel critical sections in ipfw.

In detail:

 1. introduce a IPFW_UH_LOCK to arbitrate requests from
     the upper half of the kernel. Some things, such as 'ipfw show',
     can be done holding this lock in read mode, whereas insert and
     delete require IPFW_UH_WLOCK.

  2. introduce a mapping structure to keep rules together. This replaces
     the 'next' chain currently used in ipfw rules. At the moment
     the map is a simple array (sorted by rule number and then rule_id),
     so we can find a rule quickly instead of having to scan the list.
     This reduces many expensive lookups from O(N) to O(log N).

  3. when an expensive operation (such as insert or delete) is done
     by userland, we grab IPFW_UH_WLOCK, create a new copy of the map
     without blocking the bottom half of the kernel, then acquire
     IPFW_WLOCK and quickly update pointers to the map and related info.
     After dropping IPFW_LOCK we can then continue the cleanup protected
     by IPFW_UH_LOCK. So userland still costs O(N) but the kernel side
     is only blocked for O(1).

  4. do not pass pointers to rules through dummynet, netgraph, divert etc,
     but rather pass a <slot, chain_id, rulenum, rule_id> tuple.
     We validate the slot index (in the array of #2) with chain_id,
     and if successful do a O(1) dereference; otherwise, we can find
     the rule in O(log N) through <rulenum, rule_id>

All the above does not change the userland/kernel ABI, though there
are some disgusting casts between pointers and uint32_t

Operation costs now are as follows:

  Function				Old	Now	  Planned
-------------------------------------------------------------------
  + skipto X, non cached		O(N)	O(log N)
  + skipto X, cached			O(1)	O(1)
XXX dynamic rule lookup			O(1)	O(log N)  O(1)
  + skipto tablearg			O(N)	O(1)
  + reinject, non cached		O(N)	O(log N)
  + reinject, cached			O(1)	O(1)
  + kernel blocked during setsockopt()	O(N)	O(1)
-------------------------------------------------------------------

The only (very small) regression is on dynamic rule lookup and this will
be fixed in a day or two, without changing the userland/kernel ABI

Supported by: Valeria Paoli
MFC after:	1 month
2009-12-22 19:01:47 +00:00

575 lines
17 KiB
C
Raw Blame History

/*-
* Copyright (c) 2002-2009 Luigi Rizzo, Universita` di Pisa
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
#ifndef _IPFW2_H
#define _IPFW2_H
/*
* The default rule number. By the design of ip_fw, the default rule
* is the last one, so its number can also serve as the highest number
* allowed for a rule. The ip_fw code relies on both meanings of this
* constant.
*/
#define IPFW_DEFAULT_RULE 65535
/*
* The number of ipfw tables. The maximum allowed table number is the
* (IPFW_TABLES_MAX - 1).
*/
#define IPFW_TABLES_MAX 128
/*
* Most commands (queue, pipe, tag, untag, limit...) can have a 16-bit
* argument between 1 and 65534. The value 0 is unused, the value
* 65535 (IP_FW_TABLEARG) is used to represent 'tablearg', i.e. the
* can be 1..65534, or 65535 to indicate the use of a 'tablearg'
* result of the most recent table() lookup.
* Note that 16bit is only a historical limit, resulting from
* the use of a 16-bit fields for that value. In reality, we can have
* 2^32 pipes, queues, tag values and so on, and use 0 as a tablearg.
*/
#define IPFW_ARG_MIN 1
#define IPFW_ARG_MAX 65534
#define IP_FW_TABLEARG 65535 /* XXX should use 0 */
/*
* The kernel representation of ipfw rules is made of a list of
* 'instructions' (for all practical purposes equivalent to BPF
* instructions), which specify which fields of the packet
* (or its metadata) should be analysed.
*
* Each instruction is stored in a structure which begins with
* "ipfw_insn", and can contain extra fields depending on the
* instruction type (listed below).
* Note that the code is written so that individual instructions
* have a size which is a multiple of 32 bits. This means that, if
* such structures contain pointers or other 64-bit entities,
* (there is just one instance now) they may end up unaligned on
* 64-bit architectures, so the must be handled with care.
*
* "enum ipfw_opcodes" are the opcodes supported. We can have up
* to 256 different opcodes. When adding new opcodes, they should
* be appended to the end of the opcode list before O_LAST_OPCODE,
* this will prevent the ABI from being broken, otherwise users
* will have to recompile ipfw(8) when they update the kernel.
*/
enum ipfw_opcodes { /* arguments (4 byte each) */
O_NOP,
O_IP_SRC, /* u32 = IP */
O_IP_SRC_MASK, /* ip = IP/mask */
O_IP_SRC_ME, /* none */
O_IP_SRC_SET, /* u32=base, arg1=len, bitmap */
O_IP_DST, /* u32 = IP */
O_IP_DST_MASK, /* ip = IP/mask */
O_IP_DST_ME, /* none */
O_IP_DST_SET, /* u32=base, arg1=len, bitmap */
O_IP_SRCPORT, /* (n)port list:mask 4 byte ea */
O_IP_DSTPORT, /* (n)port list:mask 4 byte ea */
O_PROTO, /* arg1=protocol */
O_MACADDR2, /* 2 mac addr:mask */
O_MAC_TYPE, /* same as srcport */
O_LAYER2, /* none */
O_IN, /* none */
O_FRAG, /* none */
O_RECV, /* none */
O_XMIT, /* none */
O_VIA, /* none */
O_IPOPT, /* arg1 = 2*u8 bitmap */
O_IPLEN, /* arg1 = len */
O_IPID, /* arg1 = id */
O_IPTOS, /* arg1 = id */
O_IPPRECEDENCE, /* arg1 = precedence << 5 */
O_IPTTL, /* arg1 = TTL */
O_IPVER, /* arg1 = version */
O_UID, /* u32 = id */
O_GID, /* u32 = id */
O_ESTAB, /* none (tcp established) */
O_TCPFLAGS, /* arg1 = 2*u8 bitmap */
O_TCPWIN, /* arg1 = desired win */
O_TCPSEQ, /* u32 = desired seq. */
O_TCPACK, /* u32 = desired seq. */
O_ICMPTYPE, /* u32 = icmp bitmap */
O_TCPOPTS, /* arg1 = 2*u8 bitmap */
O_VERREVPATH, /* none */
O_VERSRCREACH, /* none */
O_PROBE_STATE, /* none */
O_KEEP_STATE, /* none */
O_LIMIT, /* ipfw_insn_limit */
O_LIMIT_PARENT, /* dyn_type, not an opcode. */
/*
* These are really 'actions'.
*/
O_LOG, /* ipfw_insn_log */
O_PROB, /* u32 = match probability */
O_CHECK_STATE, /* none */
O_ACCEPT, /* none */
O_DENY, /* none */
O_REJECT, /* arg1=icmp arg (same as deny) */
O_COUNT, /* none */
O_SKIPTO, /* arg1=next rule number */
O_PIPE, /* arg1=pipe number */
O_QUEUE, /* arg1=queue number */
O_DIVERT, /* arg1=port number */
O_TEE, /* arg1=port number */
O_FORWARD_IP, /* fwd sockaddr */
O_FORWARD_MAC, /* fwd mac */
O_NAT, /* nope */
O_REASS, /* none */
/*
* More opcodes.
*/
O_IPSEC, /* has ipsec history */
O_IP_SRC_LOOKUP, /* arg1=table number, u32=value */
O_IP_DST_LOOKUP, /* arg1=table number, u32=value */
O_ANTISPOOF, /* none */
O_JAIL, /* u32 = id */
O_ALTQ, /* u32 = altq classif. qid */
O_DIVERTED, /* arg1=bitmap (1:loop, 2:out) */
O_TCPDATALEN, /* arg1 = tcp data len */
O_IP6_SRC, /* address without mask */
O_IP6_SRC_ME, /* my addresses */
O_IP6_SRC_MASK, /* address with the mask */
O_IP6_DST,
O_IP6_DST_ME,
O_IP6_DST_MASK,
O_FLOW6ID, /* for flow id tag in the ipv6 pkt */
O_ICMP6TYPE, /* icmp6 packet type filtering */
O_EXT_HDR, /* filtering for ipv6 extension header */
O_IP6,
/*
* actions for ng_ipfw
*/
O_NETGRAPH, /* send to ng_ipfw */
O_NGTEE, /* copy to ng_ipfw */
O_IP4,
O_UNREACH6, /* arg1=icmpv6 code arg (deny) */
O_TAG, /* arg1=tag number */
O_TAGGED, /* arg1=tag number */
O_SETFIB, /* arg1=FIB number */
O_FIB, /* arg1=FIB desired fib number */
O_LAST_OPCODE /* not an opcode! */
};
/*
* The extension header are filtered only for presence using a bit
* vector with a flag for each header.
*/
#define EXT_FRAGMENT 0x1
#define EXT_HOPOPTS 0x2
#define EXT_ROUTING 0x4
#define EXT_AH 0x8
#define EXT_ESP 0x10
#define EXT_DSTOPTS 0x20
#define EXT_RTHDR0 0x40
#define EXT_RTHDR2 0x80
/*
* Template for instructions.
*
* ipfw_insn is used for all instructions which require no operands,
* a single 16-bit value (arg1), or a couple of 8-bit values.
*
* For other instructions which require different/larger arguments
* we have derived structures, ipfw_insn_*.
*
* The size of the instruction (in 32-bit words) is in the low
* 6 bits of "len". The 2 remaining bits are used to implement
* NOT and OR on individual instructions. Given a type, you can
* compute the length to be put in "len" using F_INSN_SIZE(t)
*
* F_NOT negates the match result of the instruction.
*
* F_OR is used to build or blocks. By default, instructions
* are evaluated as part of a logical AND. An "or" block
* { X or Y or Z } contains F_OR set in all but the last
* instruction of the block. A match will cause the code
* to skip past the last instruction of the block.
*
* NOTA BENE: in a couple of places we assume that
* sizeof(ipfw_insn) == sizeof(u_int32_t)
* this needs to be fixed.
*
*/
typedef struct _ipfw_insn { /* template for instructions */
u_int8_t opcode;
u_int8_t len; /* number of 32-bit words */
#define F_NOT 0x80
#define F_OR 0x40
#define F_LEN_MASK 0x3f
#define F_LEN(cmd) ((cmd)->len & F_LEN_MASK)
u_int16_t arg1;
} ipfw_insn;
/*
* The F_INSN_SIZE(type) computes the size, in 4-byte words, of
* a given type.
*/
#define F_INSN_SIZE(t) ((sizeof (t))/sizeof(u_int32_t))
/*
* This is used to store an array of 16-bit entries (ports etc.)
*/
typedef struct _ipfw_insn_u16 {
ipfw_insn o;
u_int16_t ports[2]; /* there may be more */
} ipfw_insn_u16;
/*
* This is used to store an array of 32-bit entries
* (uid, single IPv4 addresses etc.)
*/
typedef struct _ipfw_insn_u32 {
ipfw_insn o;
u_int32_t d[1]; /* one or more */
} ipfw_insn_u32;
/*
* This is used to store IP addr-mask pairs.
*/
typedef struct _ipfw_insn_ip {
ipfw_insn o;
struct in_addr addr;
struct in_addr mask;
} ipfw_insn_ip;
/*
* This is used to forward to a given address (ip).
*/
typedef struct _ipfw_insn_sa {
ipfw_insn o;
struct sockaddr_in sa;
} ipfw_insn_sa;
/*
* This is used for MAC addr-mask pairs.
*/
typedef struct _ipfw_insn_mac {
ipfw_insn o;
u_char addr[12]; /* dst[6] + src[6] */
u_char mask[12]; /* dst[6] + src[6] */
} ipfw_insn_mac;
/*
* This is used for interface match rules (recv xx, xmit xx).
*/
typedef struct _ipfw_insn_if {
ipfw_insn o;
union {
struct in_addr ip;
int glob;
} p;
char name[IFNAMSIZ];
} ipfw_insn_if;
/*
* This is used for storing an altq queue id number.
*/
typedef struct _ipfw_insn_altq {
ipfw_insn o;
u_int32_t qid;
} ipfw_insn_altq;
/*
* This is used for limit rules.
*/
typedef struct _ipfw_insn_limit {
ipfw_insn o;
u_int8_t _pad;
u_int8_t limit_mask; /* combination of DYN_* below */
#define DYN_SRC_ADDR 0x1
#define DYN_SRC_PORT 0x2
#define DYN_DST_ADDR 0x4
#define DYN_DST_PORT 0x8
u_int16_t conn_limit;
} ipfw_insn_limit;
/*
* This is used for log instructions.
*/
typedef struct _ipfw_insn_log {
ipfw_insn o;
u_int32_t max_log; /* how many do we log -- 0 = all */
u_int32_t log_left; /* how many left to log */
} ipfw_insn_log;
/*
* Data structures required by both ipfw(8) and ipfw(4) but not part of the
* management API are protected by IPFW_INTERNAL.
*/
#ifdef IPFW_INTERNAL
/* Server pool support (LSNAT). */
struct cfg_spool {
LIST_ENTRY(cfg_spool) _next; /* chain of spool instances */
struct in_addr addr;
u_short port;
};
#endif
/* Redirect modes id. */
#define REDIR_ADDR 0x01
#define REDIR_PORT 0x02
#define REDIR_PROTO 0x04
#ifdef IPFW_INTERNAL
/* Nat redirect configuration. */
struct cfg_redir {
LIST_ENTRY(cfg_redir) _next; /* chain of redir instances */
u_int16_t mode; /* type of redirect mode */
struct in_addr laddr; /* local ip address */
struct in_addr paddr; /* public ip address */
struct in_addr raddr; /* remote ip address */
u_short lport; /* local port */
u_short pport; /* public port */
u_short rport; /* remote port */
u_short pport_cnt; /* number of public ports */
u_short rport_cnt; /* number of remote ports */
int proto; /* protocol: tcp/udp */
struct alias_link **alink;
/* num of entry in spool chain */
u_int16_t spool_cnt;
/* chain of spool instances */
LIST_HEAD(spool_chain, cfg_spool) spool_chain;
};
#endif
#define NAT_BUF_LEN 1024
#ifdef IPFW_INTERNAL
/* Nat configuration data struct. */
struct cfg_nat {
/* chain of nat instances */
LIST_ENTRY(cfg_nat) _next;
int id; /* nat id */
struct in_addr ip; /* nat ip address */
char if_name[IF_NAMESIZE]; /* interface name */
int mode; /* aliasing mode */
struct libalias *lib; /* libalias instance */
/* number of entry in spool chain */
int redir_cnt;
/* chain of redir instances */
LIST_HEAD(redir_chain, cfg_redir) redir_chain;
};
#endif
#define SOF_NAT sizeof(struct cfg_nat)
#define SOF_REDIR sizeof(struct cfg_redir)
#define SOF_SPOOL sizeof(struct cfg_spool)
/* Nat command. */
typedef struct _ipfw_insn_nat {
ipfw_insn o;
struct cfg_nat *nat;
} ipfw_insn_nat;
/* Apply ipv6 mask on ipv6 addr */
#define APPLY_MASK(addr,mask) \
(addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \
(addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \
(addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \
(addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3];
/* Structure for ipv6 */
typedef struct _ipfw_insn_ip6 {
ipfw_insn o;
struct in6_addr addr6;
struct in6_addr mask6;
} ipfw_insn_ip6;
/* Used to support icmp6 types */
typedef struct _ipfw_insn_icmp6 {
ipfw_insn o;
uint32_t d[7]; /* XXX This number si related to the netinet/icmp6.h
* define ICMP6_MAXTYPE
* as follows: n = ICMP6_MAXTYPE/32 + 1
* Actually is 203
*/
} ipfw_insn_icmp6;
/*
* Here we have the structure representing an ipfw rule.
*
* It starts with a general area (with link fields and counters)
* followed by an array of one or more instructions, which the code
* accesses as an array of 32-bit values.
*
* Given a rule pointer r:
*
* r->cmd is the start of the first instruction.
* ACTION_PTR(r) is the start of the first action (things to do
* once a rule matched).
*
* When assembling instruction, remember the following:
*
* + if a rule has a "keep-state" (or "limit") option, then the
* first instruction (at r->cmd) MUST BE an O_PROBE_STATE
* + if a rule has a "log" option, then the first action
* (at ACTION_PTR(r)) MUST be O_LOG
* + if a rule has an "altq" option, it comes after "log"
* + if a rule has an O_TAG option, it comes after "log" and "altq"
*
* NOTE: we use a simple linked list of rules because we never need
* to delete a rule without scanning the list. We do not use
* queue(3) macros for portability and readability.
*/
struct ip_fw {
struct ip_fw *x_next; /* linked list of rules */
struct ip_fw *next_rule; /* ptr to next [skipto] rule */
/* 'next_rule' is used to pass up 'set_disable' status */
uint16_t act_ofs; /* offset of action in 32-bit units */
uint16_t cmd_len; /* # of 32-bit words in cmd */
uint16_t rulenum; /* rule number */
uint8_t set; /* rule set (0..31) */
#define RESVD_SET 31 /* set for default and persistent rules */
uint8_t _pad; /* padding */
uint32_t id; /* rule id */
/* These fields are present in all rules. */
uint64_t pcnt; /* Packet counter */
uint64_t bcnt; /* Byte counter */
uint32_t timestamp; /* tv_sec of last match */
ipfw_insn cmd[1]; /* storage for commands */
};
#define ACTION_PTR(rule) \
(ipfw_insn *)( (u_int32_t *)((rule)->cmd) + ((rule)->act_ofs) )
#define RULESIZE(rule) (sizeof(struct ip_fw) + \
((struct ip_fw *)(rule))->cmd_len * 4 - 4)
/*
* This structure is used as a flow mask and a flow id for various
* parts of the code.
*/
struct ipfw_flow_id {
u_int32_t dst_ip;
u_int32_t src_ip;
u_int16_t dst_port;
u_int16_t src_port;
u_int8_t fib;
u_int8_t proto;
u_int8_t flags; /* protocol-specific flags */
uint8_t addr_type; /* 4 = ipv4, 6 = ipv6, 1=ether ? */
struct in6_addr dst_ip6; /* could also store MAC addr! */
struct in6_addr src_ip6;
u_int32_t flow_id6;
u_int32_t frag_id6;
};
#define IS_IP6_FLOW_ID(id) ((id)->addr_type == 6)
/*
* Dynamic ipfw rule.
*/
typedef struct _ipfw_dyn_rule ipfw_dyn_rule;
struct _ipfw_dyn_rule {
ipfw_dyn_rule *next; /* linked list of rules. */
struct ip_fw *rule; /* pointer to rule */
/* 'rule' is used to pass up the rule number (from the parent) */
ipfw_dyn_rule *parent; /* pointer to parent rule */
u_int64_t pcnt; /* packet match counter */
u_int64_t bcnt; /* byte match counter */
struct ipfw_flow_id id; /* (masked) flow id */
u_int32_t expire; /* expire time */
u_int32_t bucket; /* which bucket in hash table */
u_int32_t state; /* state of this rule (typically a
* combination of TCP flags)
*/
u_int32_t ack_fwd; /* most recent ACKs in forward */
u_int32_t ack_rev; /* and reverse directions (used */
/* to generate keepalives) */
u_int16_t dyn_type; /* rule type */
u_int16_t count; /* refcount */
};
/*
* Definitions for IP option names.
*/
#define IP_FW_IPOPT_LSRR 0x01
#define IP_FW_IPOPT_SSRR 0x02
#define IP_FW_IPOPT_RR 0x04
#define IP_FW_IPOPT_TS 0x08
/*
* Definitions for TCP option names.
*/
#define IP_FW_TCPOPT_MSS 0x01
#define IP_FW_TCPOPT_WINDOW 0x02
#define IP_FW_TCPOPT_SACK 0x04
#define IP_FW_TCPOPT_TS 0x08
#define IP_FW_TCPOPT_CC 0x10
#define ICMP_REJECT_RST 0x100 /* fake ICMP code (send a TCP RST) */
#define ICMP6_UNREACH_RST 0x100 /* fake ICMPv6 code (send a TCP RST) */
/*
* These are used for lookup tables.
*/
typedef struct _ipfw_table_entry {
in_addr_t addr; /* network address */
u_int32_t value; /* value */
u_int16_t tbl; /* table number */
u_int8_t masklen; /* mask length */
} ipfw_table_entry;
typedef struct _ipfw_table {
u_int32_t size; /* size of entries in bytes */
u_int32_t cnt; /* # of entries */
u_int16_t tbl; /* table number */
ipfw_table_entry ent[0]; /* entries */
} ipfw_table;
#endif /* _IPFW2_H */
Reference in New Issue View Git Blame Copy Permalink