diff options
Diffstat (limited to 'net')
-rw-r--r-- | net/Kconfig | 3 | ||||
-rw-r--r-- | net/dccp/Makefile | 2 | ||||
-rw-r--r-- | net/dccp/ackvec.c | 419 | ||||
-rw-r--r-- | net/dccp/ackvec.h | 133 | ||||
-rw-r--r-- | net/dccp/ccid.h | 31 | ||||
-rw-r--r-- | net/dccp/ccids/ccid3.c | 56 | ||||
-rw-r--r-- | net/dccp/ccids/ccid3.h | 23 | ||||
-rw-r--r-- | net/dccp/dccp.h | 91 | ||||
-rw-r--r-- | net/dccp/input.c | 89 | ||||
-rw-r--r-- | net/dccp/ipv4.c | 101 | ||||
-rw-r--r-- | net/dccp/minisocks.c | 19 | ||||
-rw-r--r-- | net/dccp/options.c | 443 | ||||
-rw-r--r-- | net/dccp/output.c | 26 | ||||
-rw-r--r-- | net/dccp/proto.c | 94 | ||||
-rw-r--r-- | net/ipv4/netfilter/Kconfig | 14 | ||||
-rw-r--r-- | net/ipv4/netfilter/ip_conntrack_core.c | 5 | ||||
-rw-r--r-- | net/ipv4/netfilter/ipt_CLUSTERIP.c | 223 | ||||
-rw-r--r-- | net/ipv6/udp.c | 10 | ||||
-rw-r--r-- | net/socket.c | 3 |
19 files changed, 1005 insertions, 780 deletions
diff --git a/net/Kconfig b/net/Kconfig index 2bdd562..60f6f32 100644 --- a/net/Kconfig +++ b/net/Kconfig @@ -140,6 +140,7 @@ config BRIDGE_NETFILTER If unsure, say N. +source "net/netfilter/Kconfig" source "net/ipv4/netfilter/Kconfig" source "net/ipv6/netfilter/Kconfig" source "net/decnet/netfilter/Kconfig" @@ -206,8 +207,6 @@ config NET_PKTGEN To compile this code as a module, choose M here: the module will be called pktgen. -source "net/netfilter/Kconfig" - endmenu endmenu diff --git a/net/dccp/Makefile b/net/dccp/Makefile index fb97bb0..344a8da 100644 --- a/net/dccp/Makefile +++ b/net/dccp/Makefile @@ -3,6 +3,8 @@ obj-$(CONFIG_IP_DCCP) += dccp.o dccp-y := ccid.o input.o ipv4.o minisocks.o options.o output.o proto.o \ timer.o +dccp-$(CONFIG_IP_DCCP_ACKVEC) += ackvec.o + obj-$(CONFIG_INET_DCCP_DIAG) += dccp_diag.o dccp_diag-y := diag.o diff --git a/net/dccp/ackvec.c b/net/dccp/ackvec.c new file mode 100644 index 0000000..6530283 --- /dev/null +++ b/net/dccp/ackvec.c @@ -0,0 +1,419 @@ +/* + * net/dccp/ackvec.c + * + * An implementation of the DCCP protocol + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@ghostprotocols.net> + * + * 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; version 2 of the License; + */ + +#include "ackvec.h" +#include "dccp.h" + +#include <linux/dccp.h> +#include <linux/skbuff.h> + +#include <net/sock.h> + +int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb) +{ + struct dccp_sock *dp = dccp_sk(sk); + struct dccp_ackvec *av = dp->dccps_hc_rx_ackvec; + int len = av->dccpav_vec_len + 2; + struct timeval now; + u32 elapsed_time; + unsigned char *to, *from; + + dccp_timestamp(sk, &now); + elapsed_time = timeval_delta(&now, &av->dccpav_time) / 10; + + if (elapsed_time != 0) + dccp_insert_option_elapsed_time(sk, skb, elapsed_time); + + if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) + return -1; + + /* + * XXX: now we have just one ack vector sent record, so + * we have to wait for it to be cleared. + * + * Of course this is not acceptable, but this is just for + * basic testing now. + */ + if (av->dccpav_ack_seqno != DCCP_MAX_SEQNO + 1) + return -1; + + DCCP_SKB_CB(skb)->dccpd_opt_len += len; + + to = skb_push(skb, len); + *to++ = DCCPO_ACK_VECTOR_0; + *to++ = len; + + len = av->dccpav_vec_len; + from = av->dccpav_buf + av->dccpav_buf_head; + + /* Check if buf_head wraps */ + if (av->dccpav_buf_head + len > av->dccpav_vec_len) { + const u32 tailsize = (av->dccpav_vec_len - av->dccpav_buf_head); + + memcpy(to, from, tailsize); + to += tailsize; + len -= tailsize; + from = av->dccpav_buf; + } + + memcpy(to, from, len); + /* + * From draft-ietf-dccp-spec-11.txt: + * + * For each acknowledgement it sends, the HC-Receiver will add an + * acknowledgement record. ack_seqno will equal the HC-Receiver + * sequence number it used for the ack packet; ack_ptr will equal + * buf_head; ack_ackno will equal buf_ackno; and ack_nonce will + * equal buf_nonce. + * + * This implemention uses just one ack record for now. + */ + av->dccpav_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq; + av->dccpav_ack_ptr = av->dccpav_buf_head; + av->dccpav_ack_ackno = av->dccpav_buf_ackno; + av->dccpav_ack_nonce = av->dccpav_buf_nonce; + av->dccpav_sent_len = av->dccpav_vec_len; + + dccp_pr_debug("%sACK Vector 0, len=%d, ack_seqno=%llu, " + "ack_ackno=%llu\n", + debug_prefix, av->dccpav_sent_len, + (unsigned long long)av->dccpav_ack_seqno, + (unsigned long long)av->dccpav_ack_ackno); + return -1; +} + +struct dccp_ackvec *dccp_ackvec_alloc(const unsigned int len, + const unsigned int __nocast priority) +{ + struct dccp_ackvec *av = kmalloc(sizeof(*av) + len, priority); + + if (av != NULL) { + av->dccpav_buf_len = len; + av->dccpav_buf_head = + av->dccpav_buf_tail = av->dccpav_buf_len - 1; + av->dccpav_buf_ackno = + av->dccpav_ack_ackno = av->dccpav_ack_seqno = ~0LLU; + av->dccpav_buf_nonce = av->dccpav_buf_nonce = 0; + av->dccpav_ack_ptr = 0; + av->dccpav_time.tv_sec = 0; + av->dccpav_time.tv_usec = 0; + av->dccpav_sent_len = av->dccpav_vec_len = 0; + } + + return av; +} + +void dccp_ackvec_free(struct dccp_ackvec *av) +{ + kfree(av); +} + +static inline u8 dccp_ackvec_state(const struct dccp_ackvec *av, + const unsigned int index) +{ + return av->dccpav_buf[index] & DCCP_ACKVEC_STATE_MASK; +} + +static inline u8 dccp_ackvec_len(const struct dccp_ackvec *av, + const unsigned int index) +{ + return av->dccpav_buf[index] & DCCP_ACKVEC_LEN_MASK; +} + +/* + * If several packets are missing, the HC-Receiver may prefer to enter multiple + * bytes with run length 0, rather than a single byte with a larger run length; + * this simplifies table updates if one of the missing packets arrives. + */ +static inline int dccp_ackvec_set_buf_head_state(struct dccp_ackvec *av, + const unsigned int packets, + const unsigned char state) +{ + unsigned int gap; + signed long new_head; + + if (av->dccpav_vec_len + packets > av->dccpav_buf_len) + return -ENOBUFS; + + gap = packets - 1; + new_head = av->dccpav_buf_head - packets; + + if (new_head < 0) { + if (gap > 0) { + memset(av->dccpav_buf, DCCP_ACKVEC_STATE_NOT_RECEIVED, + gap + new_head + 1); + gap = -new_head; + } + new_head += av->dccpav_buf_len; + } + + av->dccpav_buf_head = new_head; + + if (gap > 0) + memset(av->dccpav_buf + av->dccpav_buf_head + 1, + DCCP_ACKVEC_STATE_NOT_RECEIVED, gap); + + av->dccpav_buf[av->dccpav_buf_head] = state; + av->dccpav_vec_len += packets; + return 0; +} + +/* + * Implements the draft-ietf-dccp-spec-11.txt Appendix A + */ +int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk, + const u64 ackno, const u8 state) +{ + /* + * Check at the right places if the buffer is full, if it is, tell the + * caller to start dropping packets till the HC-Sender acks our ACK + * vectors, when we will free up space in dccpav_buf. + * + * We may well decide to do buffer compression, etc, but for now lets + * just drop. + * + * From Appendix A: + * + * Of course, the circular buffer may overflow, either when the + * HC-Sender is sending data at a very high rate, when the + * HC-Receiver's acknowledgements are not reaching the HC-Sender, + * or when the HC-Sender is forgetting to acknowledge those acks + * (so the HC-Receiver is unable to clean up old state). In this + * case, the HC-Receiver should either compress the buffer (by + * increasing run lengths when possible), transfer its state to + * a larger buffer, or, as a last resort, drop all received + * packets, without processing them whatsoever, until its buffer + * shrinks again. + */ + + /* See if this is the first ackno being inserted */ + if (av->dccpav_vec_len == 0) { + av->dccpav_buf[av->dccpav_buf_head] = state; + av->dccpav_vec_len = 1; + } else if (after48(ackno, av->dccpav_buf_ackno)) { + const u64 delta = dccp_delta_seqno(av->dccpav_buf_ackno, + ackno); + + /* + * Look if the state of this packet is the same as the + * previous ackno and if so if we can bump the head len. + */ + if (delta == 1 && + dccp_ackvec_state(av, av->dccpav_buf_head) == state && + (dccp_ackvec_len(av, av->dccpav_buf_head) < + DCCP_ACKVEC_LEN_MASK)) + av->dccpav_buf[av->dccpav_buf_head]++; + else if (dccp_ackvec_set_buf_head_state(av, delta, state)) + return -ENOBUFS; + } else { + /* + * A.1.2. Old Packets + * + * When a packet with Sequence Number S arrives, and + * S <= buf_ackno, the HC-Receiver will scan the table + * for the byte corresponding to S. (Indexing structures + * could reduce the complexity of this scan.) + */ + u64 delta = dccp_delta_seqno(ackno, av->dccpav_buf_ackno); + unsigned int index = av->dccpav_buf_head; + + while (1) { + const u8 len = dccp_ackvec_len(av, index); + const u8 state = dccp_ackvec_state(av, index); + /* + * valid packets not yet in dccpav_buf have a reserved + * entry, with a len equal to 0. + */ + if (state == DCCP_ACKVEC_STATE_NOT_RECEIVED && + len == 0 && delta == 0) { /* Found our + reserved seat! */ + dccp_pr_debug("Found %llu reserved seat!\n", + (unsigned long long)ackno); + av->dccpav_buf[index] = state; + goto out; + } + /* len == 0 means one packet */ + if (delta < len + 1) + goto out_duplicate; + + delta -= len + 1; + if (++index == av->dccpav_buf_len) + index = 0; + } + } + + av->dccpav_buf_ackno = ackno; + dccp_timestamp(sk, &av->dccpav_time); +out: + dccp_pr_debug(""); + return 0; + +out_duplicate: + /* Duplicate packet */ + dccp_pr_debug("Received a dup or already considered lost " + "packet: %llu\n", (unsigned long long)ackno); + return -EILSEQ; +} + +#ifdef CONFIG_IP_DCCP_DEBUG +void dccp_ackvector_print(const u64 ackno, const unsigned char *vector, int len) +{ + if (!dccp_debug) + return; + + printk("ACK vector len=%d, ackno=%llu |", len, + (unsigned long long)ackno); + + while (len--) { + const u8 state = (*vector & DCCP_ACKVEC_STATE_MASK) >> 6; + const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK; + + printk("%d,%d|", state, rl); + ++vector; + } + + printk("\n"); +} + +void dccp_ackvec_print(const struct dccp_ackvec *av) +{ + dccp_ackvector_print(av->dccpav_buf_ackno, + av->dccpav_buf + av->dccpav_buf_head, + av->dccpav_vec_len); +} +#endif + +static void dccp_ackvec_trow_away_ack_record(struct dccp_ackvec *av) +{ + /* + * As we're keeping track of the ack vector size (dccpav_vec_len) and + * the sent ack vector size (dccpav_sent_len) we don't need + * dccpav_buf_tail at all, but keep this code here as in the future + * we'll implement a vector of ack records, as suggested in + * draft-ietf-dccp-spec-11.txt Appendix A. -acme + */ +#if 0 + av->dccpav_buf_tail = av->dccpav_ack_ptr + 1; + if (av->dccpav_buf_tail >= av->dccpav_vec_len) + av->dccpav_buf_tail -= av->dccpav_vec_len; +#endif + av->dccpav_vec_len -= av->dccpav_sent_len; +} + +void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, struct sock *sk, + const u64 ackno) +{ + /* Check if we actually sent an ACK vector */ + if (av->dccpav_ack_seqno == DCCP_MAX_SEQNO + 1) + return; + + if (ackno == av->dccpav_ack_seqno) { +#ifdef CONFIG_IP_DCCP_DEBUG + struct dccp_sock *dp = dccp_sk(sk); + const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ? + "CLIENT rx ack: " : "server rx ack: "; +#endif + dccp_pr_debug("%sACK packet 0, len=%d, ack_seqno=%llu, " + "ack_ackno=%llu, ACKED!\n", + debug_prefix, 1, + (unsigned long long)av->dccpav_ack_seqno, + (unsigned long long)av->dccpav_ack_ackno); + dccp_ackvec_trow_away_ack_record(av); + av->dccpav_ack_seqno = DCCP_MAX_SEQNO + 1; + } +} + +static void dccp_ackvec_check_rcv_ackvector(struct dccp_ackvec *av, + struct sock *sk, u64 ackno, + const unsigned char len, + const unsigned char *vector) +{ + unsigned char i; + + /* Check if we actually sent an ACK vector */ + if (av->dccpav_ack_seqno == DCCP_MAX_SEQNO + 1) + return; + /* + * We're in the receiver half connection, so if the received an ACK + * vector ackno (e.g. 50) before dccpav_ack_seqno (e.g. 52), we're + * not interested. + * + * Extra explanation with example: + * + * if we received an ACK vector with ackno 50, it can only be acking + * 50, 49, 48, etc, not 52 (the seqno for the ACK vector we sent). + */ + /* dccp_pr_debug("is %llu < %llu? ", ackno, av->dccpav_ack_seqno); */ + if (before48(ackno, av->dccpav_ack_seqno)) { + /* dccp_pr_debug_cat("yes\n"); */ + return; + } + /* dccp_pr_debug_cat("no\n"); */ + + i = len; + while (i--) { + const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK; + u64 ackno_end_rl; + + dccp_set_seqno(&ackno_end_rl, ackno - rl); + + /* + * dccp_pr_debug("is %llu <= %llu <= %llu? ", ackno_end_rl, + * av->dccpav_ack_seqno, ackno); + */ + if (between48(av->dccpav_ack_seqno, ackno_end_rl, ackno)) { + const u8 state = (*vector & + DCCP_ACKVEC_STATE_MASK) >> 6; + /* dccp_pr_debug_cat("yes\n"); */ + + if (state != DCCP_ACKVEC_STATE_NOT_RECEIVED) { +#ifdef CONFIG_IP_DCCP_DEBUG + struct dccp_sock *dp = dccp_sk(sk); + const char *debug_prefix = + dp->dccps_role == DCCP_ROLE_CLIENT ? + "CLIENT rx ack: " : "server rx ack: "; +#endif + dccp_pr_debug("%sACK vector 0, len=%d, " + "ack_seqno=%llu, ack_ackno=%llu, " + "ACKED!\n", + debug_prefix, len, + (unsigned long long) + av->dccpav_ack_seqno, + (unsigned long long) + av->dccpav_ack_ackno); + dccp_ackvec_trow_away_ack_record(av); + } + /* + * If dccpav_ack_seqno was not received, no problem + * we'll send another ACK vector. + */ + av->dccpav_ack_seqno = DCCP_MAX_SEQNO + 1; + break; + } + /* dccp_pr_debug_cat("no\n"); */ + + dccp_set_seqno(&ackno, ackno_end_rl - 1); + ++vector; + } +} + +int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb, + const u8 opt, const u8 *value, const u8 len) +{ + if (len > DCCP_MAX_ACKVEC_LEN) + return -1; + + /* dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq, value, len); */ + dccp_ackvec_check_rcv_ackvector(dccp_sk(sk)->dccps_hc_rx_ackvec, sk, + DCCP_SKB_CB(skb)->dccpd_ack_seq, + len, value); + return 0; +} diff --git a/net/dccp/ackvec.h b/net/dccp/ackvec.h new file mode 100644 index 0000000..8ca51c9 --- /dev/null +++ b/net/dccp/ackvec.h @@ -0,0 +1,133 @@ +#ifndef _ACKVEC_H +#define _ACKVEC_H +/* + * net/dccp/ackvec.h + * + * An implementation of the DCCP protocol + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@mandriva.com> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/config.h> +#include <linux/compiler.h> +#include <linux/time.h> +#include <linux/types.h> + +/* Read about the ECN nonce to see why it is 253 */ +#define DCCP_MAX_ACKVEC_LEN 253 + +#define DCCP_ACKVEC_STATE_RECEIVED 0 +#define DCCP_ACKVEC_STATE_ECN_MARKED (1 << 6) +#define DCCP_ACKVEC_STATE_NOT_RECEIVED (3 << 6) + +#define DCCP_ACKVEC_STATE_MASK 0xC0 /* 11000000 */ +#define DCCP_ACKVEC_LEN_MASK 0x3F /* 00111111 */ + +/** struct dccp_ackvec - ack vector + * + * This data structure is the one defined in the DCCP draft + * Appendix A. + * + * @dccpav_buf_head - circular buffer head + * @dccpav_buf_tail - circular buffer tail + * @dccpav_buf_ackno - ack # of the most recent packet acknowledgeable in the + * buffer (i.e. %dccpav_buf_head) + * @dccpav_buf_nonce - the one-bit sum of the ECN Nonces on all packets acked + * by the buffer with State 0 + * + * Additionally, the HC-Receiver must keep some information about the + * Ack Vectors it has recently sent. For each packet sent carrying an + * Ack Vector, it remembers four variables: + * + * @dccpav_ack_seqno - the Sequence Number used for the packet + * (HC-Receiver seqno) + * @dccpav_ack_ptr - the value of buf_head at the time of acknowledgement. + * @dccpav_ack_ackno - the Acknowledgement Number used for the packet + * (HC-Sender seqno) + * @dccpav_ack_nonce - the one-bit sum of the ECN Nonces for all State 0. + * + * @dccpav_buf_len - circular buffer length + * @dccpav_time - the time in usecs + * @dccpav_buf - circular buffer of acknowledgeable packets + */ +struct dccp_ackvec { + unsigned int dccpav_buf_head; + unsigned int dccpav_buf_tail; + u64 dccpav_buf_ackno; + u64 dccpav_ack_seqno; + u64 dccpav_ack_ackno; + unsigned int dccpav_ack_ptr; + unsigned int dccpav_sent_len; + unsigned int dccpav_vec_len; + unsigned int dccpav_buf_len; + struct timeval dccpav_time; + u8 dccpav_buf_nonce; + u8 dccpav_ack_nonce; + u8 dccpav_buf[0]; +}; + +struct sock; +struct sk_buff; + +#ifdef CONFIG_IP_DCCP_ACKVEC +extern struct dccp_ackvec *dccp_ackvec_alloc(unsigned int len, + const unsigned int __nocast priority); +extern void dccp_ackvec_free(struct dccp_ackvec *av); + +extern int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk, + const u64 ackno, const u8 state); + +extern void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, + struct sock *sk, const u64 ackno); +extern int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb, + const u8 opt, const u8 *value, const u8 len); + +extern int dccp_insert_option_ackvec(struct sock *sk, struct sk_buff *skb); + +static inline int dccp_ackvec_pending(const struct dccp_ackvec *av) +{ + return av->dccpav_sent_len != av->dccpav_vec_len; +} +#else /* CONFIG_IP_DCCP_ACKVEC */ +static inline struct dccp_ackvec *dccp_ackvec_alloc(unsigned int len, + const unsigned int __nocast priority) +{ + return NULL; +} + +static inline void dccp_ackvec_free(struct dccp_ackvec *av) +{ +} + +static inline int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk, + const u64 ackno, const u8 state) +{ + return -1; +} + +static inline void dccp_ackvec_check_rcv_ackno(struct dccp_ackvec *av, + struct sock *sk, const u64 ackno) +{ +} + +static inline int dccp_ackvec_parse(struct sock *sk, const struct sk_buff *skb, + const u8 opt, const u8 *value, const u8 len) +{ + return -1; +} + +static inline int dccp_insert_option_ackvec(const struct sock *sk, + const struct sk_buff *skb) +{ + return -1; +} + +static inline int dccp_ackvec_pending(const struct dccp_ackvec *av) +{ + return 0; +} +#endif /* CONFIG_IP_DCCP_ACKVEC */ +#endif /* _ACKVEC_H */ diff --git a/net/dccp/ccid.h b/net/dccp/ccid.h index 962f1e9..21e5514 100644 --- a/net/dccp/ccid.h +++ b/net/dccp/ccid.h @@ -14,6 +14,7 @@ */ #include <net/sock.h> +#include <linux/compiler.h> #include <linux/dccp.h> #include <linux/list.h> #include <linux/module.h> @@ -54,6 +55,14 @@ struct ccid { struct tcp_info *info); void (*ccid_hc_tx_get_info)(struct sock *sk, struct tcp_info *info); + int (*ccid_hc_rx_getsockopt)(struct sock *sk, + const int optname, int len, + u32 __user *optval, + int __user *optlen); + int (*ccid_hc_tx_getsockopt)(struct sock *sk, + const int optname, int len, + u32 __user *optval, + int __user *optlen); }; extern int ccid_register(struct ccid *ccid); @@ -177,4 +186,26 @@ static inline void ccid_hc_tx_get_info(struct ccid *ccid, struct sock *sk, if (ccid->ccid_hc_tx_get_info != NULL) ccid->ccid_hc_tx_get_info(sk, info); } + +static inline int ccid_hc_rx_getsockopt(struct ccid *ccid, struct sock *sk, + const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + int rc = -ENOPROTOOPT; + if (ccid->ccid_hc_rx_getsockopt != NULL) + rc = ccid->ccid_hc_rx_getsockopt(sk, optname, len, + optval, optlen); + return rc; +} + +static inline int ccid_hc_tx_getsockopt(struct ccid *ccid, struct sock *sk, + const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + int rc = -ENOPROTOOPT; + if (ccid->ccid_hc_tx_getsockopt != NULL) + rc = ccid->ccid_hc_tx_getsockopt(sk, optname, len, + optval, optlen); + return rc; +} #endif /* _CCID_H */ diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c index 38aa849..aa68e0a 100644 --- a/net/dccp/ccids/ccid3.c +++ b/net/dccp/ccids/ccid3.c @@ -1120,6 +1120,60 @@ static void ccid3_hc_tx_get_info(struct sock *sk, struct tcp_info *info) info->tcpi_rtt = hctx->ccid3hctx_rtt; } +static int ccid3_hc_rx_getsockopt(struct sock *sk, const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + const struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk); + const void *val; + + /* Listen socks doesn't have a private CCID block */ + if (sk->sk_state == DCCP_LISTEN) + return -EINVAL; + + switch (optname) { + case DCCP_SOCKOPT_CCID_RX_INFO: + if (len < sizeof(hcrx->ccid3hcrx_tfrc)) + return -EINVAL; + len = sizeof(hcrx->ccid3hcrx_tfrc); + val = &hcrx->ccid3hcrx_tfrc; + break; + default: + return -ENOPROTOOPT; + } + + if (put_user(len, optlen) || copy_to_user(optval, val, len)) + return -EFAULT; + + return 0; +} + +static int ccid3_hc_tx_getsockopt(struct sock *sk, const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + const struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk); + const void *val; + + /* Listen socks doesn't have a private CCID block */ + if (sk->sk_state == DCCP_LISTEN) + return -EINVAL; + + switch (optname) { + case DCCP_SOCKOPT_CCID_TX_INFO: + if (len < sizeof(hctx->ccid3hctx_tfrc)) + return -EINVAL; + len = sizeof(hctx->ccid3hctx_tfrc); + val = &hctx->ccid3hctx_tfrc; + break; + default: + return -ENOPROTOOPT; + } + + if (put_user(len, optlen) || copy_to_user(optval, val, len)) + return -EFAULT; + + return 0; +} + static struct ccid ccid3 = { .ccid_id = 3, .ccid_name = "ccid3", @@ -1139,6 +1193,8 @@ static struct ccid ccid3 = { .ccid_hc_rx_packet_recv = ccid3_hc_rx_packet_recv, .ccid_hc_rx_get_info = ccid3_hc_rx_get_info, .ccid_hc_tx_get_info = ccid3_hc_tx_get_info, + .ccid_hc_rx_getsockopt = ccid3_hc_rx_getsockopt, + .ccid_hc_tx_getsockopt = ccid3_hc_tx_getsockopt, }; module_param(ccid3_debug, int, 0444); diff --git a/net/dccp/ccids/ccid3.h b/net/dccp/ccids/ccid3.h index eb24877..0bde458 100644 --- a/net/dccp/ccids/ccid3.h +++ b/net/dccp/ccids/ccid3.h @@ -40,6 +40,7 @@ #include <linux/list.h> #include <linux/time.h> #include <linux/types.h> +#include <linux/tfrc.h> #define TFRC_MIN_PACKET_SIZE 16 #define TFRC_STD_PACKET_SIZE 256 @@ -93,12 +94,15 @@ struct ccid3_options_received { * @ccid3hctx_hist - Packet history */ struct ccid3_hc_tx_sock { - u32 ccid3hctx_x; - u32 ccid3hctx_x_recv; - u32 ccid3hctx_x_calc; + struct tfrc_tx_info ccid3hctx_tfrc; +#define ccid3hctx_x ccid3hctx_tfrc.tfrctx_x +#define ccid3hctx_x_recv ccid3hctx_tfrc.tfrctx_x_recv +#define ccid3hctx_x_calc ccid3hctx_tfrc.tfrctx_x_calc +#define ccid3hctx_rtt ccid3hctx_tfrc.tfrctx_rtt +#define ccid3hctx_p ccid3hctx_tfrc.tfrctx_p +#define ccid3hctx_t_rto ccid3hctx_tfrc.tfrctx_rto +#define ccid3hctx_t_ipi ccid3hctx_tfrc.tfrctx_ipi u16 ccid3hctx_s; - u32 ccid3hctx_rtt; - u32 ccid3hctx_p; u8 ccid3hctx_state; u8 ccid3hctx_last_win_count; u8 ccid3hctx_idle; @@ -106,19 +110,19 @@ struct ccid3_hc_tx_sock { struct timer_list ccid3hctx_no_feedback_timer; struct timeval ccid3hctx_t_ld; struct timeval ccid3hctx_t_nom; - u32 ccid3hctx_t_rto; - u32 ccid3hctx_t_ipi; u32 ccid3hctx_delta; struct list_head ccid3hctx_hist; struct ccid3_options_received ccid3hctx_options_received; }; struct ccid3_hc_rx_sock { + struct tfrc_rx_info ccid3hcrx_tfrc; +#define ccid3hcrx_x_recv ccid3hcrx_tfrc.tfrcrx_x_recv +#define ccid3hcrx_rtt ccid3hcrx_tfrc.tfrcrx_rtt +#define ccid3hcrx_p ccid3hcrx_tfrc.tfrcrx_p u64 ccid3hcrx_seqno_last_counter:48, ccid3hcrx_state:8, ccid3hcrx_last_counter:4; - u32 ccid3hcrx_rtt; - u32 ccid3hcrx_p; u32 ccid3hcrx_bytes_recv; struct timeval ccid3hcrx_tstamp_last_feedback; struct timeval ccid3hcrx_tstamp_last_ack; @@ -127,7 +131,6 @@ struct ccid3_hc_rx_sock { u16 ccid3hcrx_s; u32 ccid3hcrx_pinv; u32 ccid3hcrx_elapsed_time; - u32 ccid3hcrx_x_recv; }; static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk) diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h index 95c4630b..5871c02 100644 --- a/net/dccp/dccp.h +++ b/net/dccp/dccp.h @@ -17,6 +17,7 @@ #include <net/snmp.h> #include <net/sock.h> #include <net/tcp.h> +#include "ackvec.h" #ifdef CONFIG_IP_DCCP_DEBUG extern int dccp_debug; @@ -258,13 +259,12 @@ extern int dccp_v4_send_reset(struct sock *sk, extern void dccp_send_close(struct sock *sk, const int active); struct dccp_skb_cb { - __u8 dccpd_type; - __u8 dccpd_reset_code; - __u8 dccpd_service; - __u8 dccpd_ccval; + __u8 dccpd_type:4; + __u8 dccpd_ccval:4; + __u8 dccpd_reset_code; + __u16 dccpd_opt_len; __u64 dccpd_seq; __u64 dccpd_ack_seq; - int dccpd_opt_len; }; #define DCCP_SKB_CB(__skb) ((struct dccp_skb_cb *)&((__skb)->cb[0])) @@ -359,6 +359,17 @@ static inline void dccp_update_gss(struct sock *sk, u64 seq) (dp->dccps_gss - dp->dccps_options.dccpo_sequence_window + 1)); } + +static inline int dccp_ack_pending(const struct sock *sk) +{ + const struct dccp_sock *dp = dccp_sk(sk); + return dp->dccps_timestamp_echo != 0 || +#ifdef CONFIG_IP_DCCP_ACKVEC + (dp->dccps_options.dccpo_send_ack_vector && + dccp_ackvec_pending(dp->dccps_hc_rx_ackvec)) || +#endif + inet_csk_ack_scheduled(sk); +} extern void dccp_insert_options(struct sock *sk, struct sk_buff *skb); extern void dccp_insert_option_elapsed_time(struct sock *sk, @@ -372,65 +383,6 @@ extern void dccp_insert_option(struct sock *sk, struct sk_buff *skb, extern struct socket *dccp_ctl_socket; -#define DCCP_ACKPKTS_STATE_RECEIVED 0 -#define DCCP_ACKPKTS_STATE_ECN_MARKED (1 << 6) -#define DCCP_ACKPKTS_STATE_NOT_RECEIVED (3 << 6) - -#define DCCP_ACKPKTS_STATE_MASK 0xC0 /* 11000000 */ -#define DCCP_ACKPKTS_LEN_MASK 0x3F /* 00111111 */ - -/** struct dccp_ackpkts - acknowledgeable packets - * - * This data structure is the one defined in the DCCP draft - * Appendix A. - * - * @dccpap_buf_head - circular buffer head - * @dccpap_buf_tail - circular buffer tail - * @dccpap_buf_ackno - ack # of the most recent packet acknowledgeable in the - * buffer (i.e. %dccpap_buf_head) - * @dccpap_buf_nonce - the one-bit sum of the ECN Nonces on all packets acked - * by the buffer with State 0 - * - * Additionally, the HC-Receiver must keep some information about the - * Ack Vectors it has recently sent. For each packet sent carrying an - * Ack Vector, it remembers four variables: - * - * @dccpap_ack_seqno - the Sequence Number used for the packet - * (HC-Receiver seqno) - * @dccpap_ack_ptr - the value of buf_head at the time of acknowledgement. - * @dccpap_ack_ackno - the Acknowledgement Number used for the packet - * (HC-Sender seqno) - * @dccpap_ack_nonce - the one-bit sum of the ECN Nonces for all State 0. - * - * @dccpap_buf_len - circular buffer length - * @dccpap_time - the time in usecs - * @dccpap_buf - circular buffer of acknowledgeable packets - */ -struct dccp_ackpkts { - unsigned int dccpap_buf_head; - unsigned int dccpap_buf_tail; - u64 dccpap_buf_ackno; - u64 dccpap_ack_seqno; - u64 dccpap_ack_ackno; - unsigned int dccpap_ack_ptr; - unsigned int dccpap_buf_vector_len; - unsigned int dccpap_ack_vector_len; - unsigned int dccpap_buf_len; - struct timeval dccpap_time; - u8 dccpap_buf_nonce; - u8 dccpap_ack_nonce; - u8 dccpap_buf[0]; -}; - -extern struct dccp_ackpkts * - dccp_ackpkts_alloc(unsigned int len, - const unsigned int __nocast priority); -extern void dccp_ackpkts_free(struct dccp_ackpkts *ap); -extern int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk, - u64 ackno, u8 state); -extern void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap, - struct sock *sk, u64 ackno); - extern void dccp_timestamp(const struct sock *sk, struct timeval *tv); static inline suseconds_t timeval_usecs(const struct timeval *tv) @@ -471,15 +423,4 @@ static inline void timeval_sub_usecs(struct timeval *tv, } } -#ifdef CONFIG_IP_DCCP_DEBUG -extern void dccp_ackvector_print(const u64 ackno, - const unsigned char *vector, int len); -extern void dccp_ackpkts_print(const struct dccp_ackpkts *ap); -#else -static inline void dccp_ackvector_print(const u64 ackno, - const unsigned char *vector, - int len) { } -static inline void dccp_ackpkts_print(const struct dccp_ackpkts *ap) { } -#endif - #endif /* _DCCP_H */ diff --git a/net/dccp/input.c b/net/dccp/input.c index c74034c..1b6b2cb 100644 --- a/net/dccp/input.c +++ b/net/dccp/input.c @@ -16,6 +16,7 @@ #include <net/sock.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" @@ -60,8 +61,8 @@ static inline void dccp_event_ack_recv(struct sock *sk, struct sk_buff *skb) struct dccp_sock *dp = dccp_sk(sk); if (dp->dccps_options.dccpo_send_ack_vector) - dccp_ackpkts_check_rcv_ackno(dp->dccps_hc_rx_ackpkts, sk, - DCCP_SKB_CB(skb)->dccpd_ack_seq); + dccp_ackvec_check_rcv_ackno(dp->dccps_hc_rx_ackvec, sk, + DCCP_SKB_CB(skb)->dccpd_ack_seq); } static int dccp_check_seqno(struct sock *sk, struct sk_buff *skb) @@ -164,37 +165,11 @@ int dccp_rcv_established(struct sock *sk, struct sk_buff *skb, if (DCCP_SKB_CB(skb)->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ) dccp_event_ack_recv(sk, skb); - /* - * FIXME: check ECN to see if we should use - * DCCP_ACKPKTS_STATE_ECN_MARKED - */ - if (dp->dccps_options.dccpo_send_ack_vector) { - struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts; - - if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk, - DCCP_SKB_CB(skb)->dccpd_seq, - DCCP_ACKPKTS_STATE_RECEIVED)) { - LIMIT_NETDEBUG(KERN_WARNING "DCCP: acknowledgeable " - "packets buffer full!\n"); - ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1; - inet_csk_schedule_ack(sk); - inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, - TCP_DELACK_MIN, - DCCP_RTO_MAX); - goto discard; - } - - /* - * FIXME: this activation is probably wrong, have to study more - * TCP delack machinery and how it fits into DCCP draft, but - * for now it kinda "works" 8) - */ - if (!inet_csk_ack_scheduled(sk)) { - inet_csk_schedule_ack(sk); - inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, 5 * HZ, - DCCP_RTO_MAX); - } - } + if (dp->dccps_options.dccpo_send_ack_vector && + dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk, + DCCP_SKB_CB(skb)->dccpd_seq, + DCCP_ACKVEC_STATE_RECEIVED)) + goto discard; ccid_hc_rx_packet_recv(dp->dccps_hc_rx_ccid, sk, skb); ccid_hc_tx_packet_recv(dp->dccps_hc_tx_ccid, sk, skb); @@ -384,9 +359,9 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk, } out_invalid_packet: - return 1; /* dccp_v4_do_rcv will send a reset, but... - FIXME: the reset code should be - DCCP_RESET_CODE_PACKET_ERROR */ + /* dccp_v4_do_rcv will send a reset */ + DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR; + return 1; } static int dccp_rcv_respond_partopen_state_process(struct sock *sk, @@ -433,6 +408,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb, struct dccp_hdr *dh, unsigned len) { struct dccp_sock *dp = dccp_sk(sk); + struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb); const int old_state = sk->sk_state; int queued = 0; @@ -473,7 +449,8 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb, if (dh->dccph_type == DCCP_PKT_RESET) goto discard; - /* Caller (dccp_v4_do_rcv) will send Reset(No Connection)*/ + /* Caller (dccp_v4_do_rcv) will send Reset */ + dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION; return 1; } @@ -487,36 +464,17 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb, if (dccp_parse_options(sk, skb)) goto discard; - if (DCCP_SKB_CB(skb)->dccpd_ack_seq != - DCCP_PKT_WITHOUT_ACK_SEQ) + if (dcb->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ) dccp_event_ack_recv(sk, skb); ccid_hc_rx_packet_recv(dp->dccps_hc_rx_ccid, sk, skb); ccid_hc_tx_packet_recv(dp->dccps_hc_tx_ccid, sk, skb); - /* - * FIXME: check ECN to see if we should use - * DCCP_ACKPKTS_STATE_ECN_MARKED - */ - if (dp->dccps_options.dccpo_send_ack_vector) { - if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk, - DCCP_SKB_CB(skb)->dccpd_seq, - DCCP_ACKPKTS_STATE_RECEIVED)) - goto discard; - /* - * FIXME: this activation is probably wrong, have to - * study more TCP delack machinery and how it fits into - * DCCP draft, but for now it kinda "works" 8) - */ - if ((dp->dccps_hc_rx_ackpkts->dccpap_ack_seqno == - DCCP_MAX_SEQNO + 1) && - !inet_csk_ack_scheduled(sk)) { - inet_csk_schedule_ack(sk); - inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, - TCP_DELACK_MIN, - DCCP_RTO_MAX); - } - } + if (dp->dccps_options.dccpo_send_ack_vector && + dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk, + DCCP_SKB_CB(skb)->dccpd_seq, + DCCP_ACKVEC_STATE_RECEIVED)) + goto discard; } /* @@ -551,8 +509,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb, dh->dccph_type == DCCP_PKT_REQUEST) || (sk->sk_state == DCCP_RESPOND && dh->dccph_type == DCCP_PKT_DATA)) { - dccp_send_sync(sk, DCCP_SKB_CB(skb)->dccpd_seq, - DCCP_PKT_SYNC); + dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNC); goto discard; } else if (dh->dccph_type == DCCP_PKT_CLOSEREQ) { dccp_rcv_closereq(sk, skb); @@ -563,13 +520,13 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb, } if (unlikely(dh->dccph_type == DCCP_PKT_SYNC)) { - dccp_send_sync(sk, DCCP_SKB_CB(skb)->dccpd_seq, - DCCP_PKT_SYNCACK); + dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNCACK); goto discard; } switch (sk->sk_state) { case DCCP_CLOSED: + dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION; return 1; case DCCP_REQUESTING: diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c index 2afaa46..40fe6af 100644 --- a/net/dccp/ipv4.c +++ b/net/dccp/ipv4.c @@ -23,6 +23,7 @@ #include <net/tcp_states.h> #include <net/xfrm.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" @@ -246,6 +247,9 @@ static int dccp_v4_connect(struct sock *sk, struct sockaddr *uaddr, dp->dccps_role = DCCP_ROLE_CLIENT; + if (dccp_service_not_initialized(sk)) + return -EPROTO; + if (addr_len < sizeof(struct sockaddr_in)) return -EINVAL; @@ -661,6 +665,16 @@ static inline u64 dccp_v4_init_sequence(const struct sock *sk, dccp_hdr(skb)->dccph_sport); } +static inline int dccp_bad_service_code(const struct sock *sk, + const __u32 service) +{ + const struct dccp_sock *dp = dccp_sk(sk); + + if (dp->dccps_service == service) + return 0; + return !dccp_list_has_service(dp->dccps_service_list, service); +} + int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb) { struct inet_request_sock *ireq; @@ -669,13 +683,22 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb) struct dccp_request_sock *dreq; const __u32 saddr = skb->nh.iph->saddr; const __u32 daddr = skb->nh.iph->daddr; + const __u32 service = dccp_hdr_request(skb)->dccph_req_service; + struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb); + __u8 reset_code = DCCP_RESET_CODE_TOO_BUSY; struct dst_entry *dst = NULL; /* Never answer to DCCP_PKT_REQUESTs send to broadcast or multicast */ if (((struct rtable *)skb->dst)->rt_flags & - (RTCF_BROADCAST | RTCF_MULTICAST)) + (RTCF_BROADCAST | RTCF_MULTICAST)) { + reset_code = DCCP_RESET_CODE_NO_CONNECTION; goto drop; + } + if (dccp_bad_service_code(sk, service)) { + reset_code = DCCP_RESET_CODE_BAD_SERVICE_CODE; + goto drop; + } /* * TW buckets are converted to open requests without * limitations, they conserve resources and peer is @@ -718,9 +741,9 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb) * dccp_create_openreq_child. */ dreq = dccp_rsk(req); - dreq->dreq_isr = DCCP_SKB_CB(skb)->dccpd_seq; - dreq->dreq_iss = dccp_v4_init_sequence(sk, skb); - dreq->dreq_service = dccp_hdr_request(skb)->dccph_req_service; + dreq->dreq_isr = dcb->dccpd_seq; + dreq->dreq_iss = dccp_v4_init_sequence(sk, skb); + dreq->dreq_service = service; if (dccp_v4_send_response(sk, req, dst)) goto drop_and_free; @@ -735,6 +758,7 @@ drop_and_free: __reqsk_free(req); drop: DCCP_INC_STATS_BH(DCCP_MIB_ATTEMPTFAILS); + dcb->dccpd_reset_code = reset_code; return -1; } @@ -1005,7 +1029,6 @@ int dccp_v4_do_rcv(struct sock *sk, struct sk_buff *skb) return 0; reset: - DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION; dccp_v4_ctl_send_reset(skb); discard: kfree_skb(skb); @@ -1090,45 +1113,7 @@ int dccp_v4_rcv(struct sk_buff *skb) goto discard_it; dh = dccp_hdr(skb); -#if 0 - /* - * Use something like this to simulate some DATA/DATAACK loss to test - * dccp_ackpkts_add, you'll get something like this on a session that - * sends 10 DATA/DATAACK packets: - * - * ackpkts_print: 281473596467422 |0,0|3,0|0,0|3,0|0,0|3,0|0,0|3,0|0,1| - * - * 0, 0 means: DCCP_ACKPKTS_STATE_RECEIVED, RLE == just this packet - * 0, 1 means: DCCP_ACKPKTS_STATE_RECEIVED, RLE == two adjacent packets - * with the same state - * 3, 0 means: DCCP_ACKPKTS_STATE_NOT_RECEIVED, RLE == just this packet - * - * So... - * - * 281473596467422 was received - * 281473596467421 was not received - * 281473596467420 was received - * 281473596467419 was not received - * 281473596467418 was received - * 281473596467417 was not received - * 281473596467416 was received - * 281473596467415 was not received - * 281473596467414 was received - * 281473596467413 was received (this one was the 3way handshake - * RESPONSE) - * - */ - if (dh->dccph_type == DCCP_PKT_DATA || - dh->dccph_type == DCCP_PKT_DATAACK) { - static int discard = 0; - if (discard) { - discard = 0; - goto discard_it; - } - discard = 1; - } -#endif DCCP_SKB_CB(skb)->dccpd_seq = dccp_hdr_seq(skb); DCCP_SKB_CB(skb)->dccpd_type = dh->dccph_type; @@ -1242,11 +1227,9 @@ static int dccp_v4_init_sock(struct sock *sk) do_gettimeofday(&dp->dccps_epoch); if (dp->dccps_options.dccpo_send_ack_vector) { - dp->dccps_hc_rx_ackpkts = - dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN, - GFP_KERNEL); - - if (dp->dccps_hc_rx_ackpkts == NULL) + dp->dccps_hc_rx_ackvec = dccp_ackvec_alloc(DCCP_MAX_ACKVEC_LEN, + GFP_KERNEL); + if (dp->dccps_hc_rx_ackvec == NULL) return -ENOMEM; } @@ -1258,16 +1241,18 @@ static int dccp_v4_init_sock(struct sock *sk) * setsockopt(CCIDs-I-want/accept). -acme */ if (likely(!dccp_ctl_socket_init)) { - dp->dccps_hc_rx_ccid = ccid_init(dp->dccps_options.dccpo_ccid, + dp->dccps_hc_rx_ccid = ccid_init(dp->dccps_options.dccpo_rx_ccid, sk); - dp->dccps_hc_tx_ccid = ccid_init(dp->dccps_options.dccpo_ccid, + dp->dccps_hc_tx_ccid = ccid_init(dp->dccps_options.dccpo_tx_ccid, sk); if (dp->dccps_hc_rx_ccid == NULL || dp->dccps_hc_tx_ccid == NULL) { ccid_exit(dp->dccps_hc_rx_ccid, sk); ccid_exit(dp->dccps_hc_tx_ccid, sk); - dccp_ackpkts_free(dp->dccps_hc_rx_ackpkts); - dp->dccps_hc_rx_ackpkts = NULL; + if (dp->dccps_options.dccpo_send_ack_vector) { + dccp_ackvec_free(dp->dccps_hc_rx_ackvec); + dp->dccps_hc_rx_ackvec = NULL; + } dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL; return -ENOMEM; } @@ -1280,6 +1265,7 @@ static int dccp_v4_init_sock(struct sock *sk) sk->sk_write_space = dccp_write_space; dp->dccps_mss_cache = 536; dp->dccps_role = DCCP_ROLE_UNDEFINED; + dp->dccps_service = DCCP_SERVICE_INVALID_VALUE; return 0; } @@ -1301,10 +1287,17 @@ static int dccp_v4_destroy_sock(struct sock *sk) if (inet_csk(sk)->icsk_bind_hash != NULL) inet_put_port(&dccp_hashinfo, sk); + if (dp->dccps_service_list != NULL) { + kfree(dp->dccps_service_list); + dp->dccps_service_list = NULL; + } + ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk); ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk); - dccp_ackpkts_free(dp->dccps_hc_rx_ackpkts); - dp->dccps_hc_rx_ackpkts = NULL; + if (dp->dccps_options.dccpo_send_ack_vector) { + dccp_ackvec_free(dp->dccps_hc_rx_ackvec); + dp->dccps_hc_rx_ackvec = NULL; + } ccid_exit(dp->dccps_hc_rx_ccid, sk); ccid_exit(dp->dccps_hc_tx_ccid, sk); dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL; diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c index 18461bc..1393461 100644 --- a/net/dccp/minisocks.c +++ b/net/dccp/minisocks.c @@ -19,6 +19,7 @@ #include <net/xfrm.h> #include <net/inet_timewait_sock.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" @@ -93,22 +94,24 @@ struct sock *dccp_create_openreq_child(struct sock *sk, struct inet_connection_sock *newicsk = inet_csk(sk); struct dccp_sock *newdp = dccp_sk(newsk); - newdp->dccps_hc_rx_ackpkts = NULL; - newdp->dccps_role = DCCP_ROLE_SERVER; - newicsk->icsk_rto = DCCP_TIMEOUT_INIT; + newdp->dccps_role = DCCP_ROLE_SERVER; + newdp->dccps_hc_rx_ackvec = NULL; + newdp->dccps_service_list = NULL; + newdp->dccps_service = dreq->dreq_service; + newicsk->icsk_rto = DCCP_TIMEOUT_INIT; do_gettimeofday(&newdp->dccps_epoch); if (newdp->dccps_options.dccpo_send_ack_vector) { - newdp->dccps_hc_rx_ackpkts = - dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN, - GFP_ATOMIC); + newdp->dccps_hc_rx_ackvec = + dccp_ackvec_alloc(DCCP_MAX_ACKVEC_LEN, + GFP_ATOMIC); /* * XXX: We're using the same CCIDs set on the parent, * i.e. sk_clone copied the master sock and left the * CCID pointers for this child, that is why we do the * __ccid_get calls. */ - if (unlikely(newdp->dccps_hc_rx_ackpkts == NULL)) + if (unlikely(newdp->dccps_hc_rx_ackvec == NULL)) goto out_free; } @@ -116,7 +119,7 @@ struct sock *dccp_create_openreq_child(struct sock *sk, newsk) != 0 || ccid_hc_tx_init(newdp->dccps_hc_tx_ccid, newsk) != 0)) { - dccp_ackpkts_free(newdp->dccps_hc_rx_ackpkts); + dccp_ackvec_free(newdp->dccps_hc_rx_ackvec); ccid_hc_rx_exit(newdp->dccps_hc_rx_ccid, newsk); ccid_hc_tx_exit(newdp->dccps_hc_tx_ccid, newsk); out_free: diff --git a/net/dccp/options.c b/net/dccp/options.c index d4c4242..0a76426 100644 --- a/net/dccp/options.c +++ b/net/dccp/options.c @@ -18,19 +18,15 @@ #include <linux/kernel.h> #include <linux/skbuff.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" -static void dccp_ackpkts_check_rcv_ackvector(struct dccp_ackpkts *ap, - struct sock *sk, - const u64 ackno, - const unsigned char len, - const unsigned char *vector); - /* stores the default values for new connection. may be changed with sysctl */ static const struct dccp_options dccpo_default_values = { .dccpo_sequence_window = DCCPF_INITIAL_SEQUENCE_WINDOW, - .dccpo_ccid = DCCPF_INITIAL_CCID, + .dccpo_rx_ccid = DCCPF_INITIAL_CCID, + .dccpo_tx_ccid = DCCPF_INITIAL_CCID, .dccpo_send_ack_vector = DCCPF_INITIAL_SEND_ACK_VECTOR, .dccpo_send_ndp_count = DCCPF_INITIAL_SEND_NDP_COUNT, }; @@ -113,25 +109,13 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb) opt_recv->dccpor_ndp); break; case DCCPO_ACK_VECTOR_0: - if (len > DCCP_MAX_ACK_VECTOR_LEN) - goto out_invalid_option; - + case DCCPO_ACK_VECTOR_1: if (pkt_type == DCCP_PKT_DATA) continue; - opt_recv->dccpor_ack_vector_len = len; - opt_recv->dccpor_ack_vector_idx = value - options; - - dccp_pr_debug("%sACK vector 0, len=%d, ack_ackno=%llu\n", - debug_prefix, len, - (unsigned long long) - DCCP_SKB_CB(skb)->dccpd_ack_seq); - dccp_ackvector_print(DCCP_SKB_CB(skb)->dccpd_ack_seq, - value, len); - dccp_ackpkts_check_rcv_ackvector(dp->dccps_hc_rx_ackpkts, - sk, - DCCP_SKB_CB(skb)->dccpd_ack_seq, - len, value); + if (dp->dccps_options.dccpo_send_ack_vector && + dccp_ackvec_parse(sk, skb, opt, value, len)) + goto out_invalid_option; break; case DCCPO_TIMESTAMP: if (len != 4) @@ -352,86 +336,6 @@ void dccp_insert_option_elapsed_time(struct sock *sk, EXPORT_SYMBOL_GPL(dccp_insert_option_elapsed_time); -static void dccp_insert_option_ack_vector(struct sock *sk, struct sk_buff *skb) -{ - struct dccp_sock *dp = dccp_sk(sk); -#ifdef CONFIG_IP_DCCP_DEBUG - const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ? - "CLIENT TX opt: " : "server TX opt: "; -#endif - struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts; - int len = ap->dccpap_buf_vector_len + 2; - struct timeval now; - u32 elapsed_time; - unsigned char *to, *from; - - dccp_timestamp(sk, &now); - elapsed_time = timeval_delta(&now, &ap->dccpap_time) / 10; - - if (elapsed_time != 0) - dccp_insert_option_elapsed_time(sk, skb, elapsed_time); - - if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) { - LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to " - "insert ACK Vector!\n"); - return; - } - - /* - * XXX: now we have just one ack vector sent record, so - * we have to wait for it to be cleared. - * - * Of course this is not acceptable, but this is just for - * basic testing now. - */ - if (ap->dccpap_ack_seqno != DCCP_MAX_SEQNO + 1) - return; - - DCCP_SKB_CB(skb)->dccpd_opt_len += len; - - to = skb_push(skb, len); - *to++ = DCCPO_ACK_VECTOR_0; - *to++ = len; - - len = ap->dccpap_buf_vector_len; - from = ap->dccpap_buf + ap->dccpap_buf_head; - - /* Check if buf_head wraps */ - if (ap->dccpap_buf_head + len > ap->dccpap_buf_len) { - const unsigned int tailsize = (ap->dccpap_buf_len - - ap->dccpap_buf_head); - - memcpy(to, from, tailsize); - to += tailsize; - len -= tailsize; - from = ap->dccpap_buf; - } - - memcpy(to, from, len); - /* - * From draft-ietf-dccp-spec-11.txt: - * - * For each acknowledgement it sends, the HC-Receiver will add an - * acknowledgement record. ack_seqno will equal the HC-Receiver - * sequence number it used for the ack packet; ack_ptr will equal - * buf_head; ack_ackno will equal buf_ackno; and ack_nonce will - * equal buf_nonce. - * - * This implemention uses just one ack record for now. - */ - ap->dccpap_ack_seqno = DCCP_SKB_CB(skb)->dccpd_seq; - ap->dccpap_ack_ptr = ap->dccpap_buf_head; - ap->dccpap_ack_ackno = ap->dccpap_buf_ackno; - ap->dccpap_ack_nonce = ap->dccpap_buf_nonce; - ap->dccpap_ack_vector_len = ap->dccpap_buf_vector_len; - - dccp_pr_debug("%sACK Vector 0, len=%d, ack_seqno=%llu, " - "ack_ackno=%llu\n", - debug_prefix, ap->dccpap_ack_vector_len, - (unsigned long long) ap->dccpap_ack_seqno, - (unsigned long long) ap->dccpap_ack_ackno); -} - void dccp_timestamp(const struct sock *sk, struct timeval *tv) { const struct dccp_sock *dp = dccp_sk(sk); @@ -528,9 +432,8 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb) if (!dccp_packet_without_ack(skb)) { if (dp->dccps_options.dccpo_send_ack_vector && - (dp->dccps_hc_rx_ackpkts->dccpap_buf_ackno != - DCCP_MAX_SEQNO + 1)) - dccp_insert_option_ack_vector(sk, skb); + dccp_ackvec_pending(dp->dccps_hc_rx_ackvec)) + dccp_insert_option_ackvec(sk, skb); if (dp->dccps_timestamp_echo != 0) dccp_insert_option_timestamp_echo(sk, skb); } @@ -557,331 +460,3 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb) } } } - -struct dccp_ackpkts *dccp_ackpkts_alloc(const unsigned int len, - const unsigned int __nocast priority) -{ - struct dccp_ackpkts *ap = kmalloc(sizeof(*ap) + len, priority); - - if (ap != NULL) { -#ifdef CONFIG_IP_DCCP_DEBUG - memset(ap->dccpap_buf, 0xFF, len); -#endif - ap->dccpap_buf_len = len; - ap->dccpap_buf_head = - ap->dccpap_buf_tail = - ap->dccpap_buf_len - 1; - ap->dccpap_buf_ackno = - ap->dccpap_ack_ackno = - ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1; - ap->dccpap_buf_nonce = ap->dccpap_buf_nonce = 0; - ap->dccpap_ack_ptr = 0; - ap->dccpap_time.tv_sec = 0; - ap->dccpap_time.tv_usec = 0; - ap->dccpap_buf_vector_len = ap->dccpap_ack_vector_len = 0; - } - - return ap; -} - -void dccp_ackpkts_free(struct dccp_ackpkts *ap) -{ - if (ap != NULL) { -#ifdef CONFIG_IP_DCCP_DEBUG - memset(ap, 0xFF, sizeof(*ap) + ap->dccpap_buf_len); -#endif - kfree(ap); - } -} - -static inline u8 dccp_ackpkts_state(const struct dccp_ackpkts *ap, - const unsigned int index) -{ - return ap->dccpap_buf[index] & DCCP_ACKPKTS_STATE_MASK; -} - -static inline u8 dccp_ackpkts_len(const struct dccp_ackpkts *ap, - const unsigned int index) -{ - return ap->dccpap_buf[index] & DCCP_ACKPKTS_LEN_MASK; -} - -/* - * If several packets are missing, the HC-Receiver may prefer to enter multiple - * bytes with run length 0, rather than a single byte with a larger run length; - * this simplifies table updates if one of the missing packets arrives. - */ -static inline int dccp_ackpkts_set_buf_head_state(struct dccp_ackpkts *ap, - const unsigned int packets, - const unsigned char state) -{ - unsigned int gap; - signed long new_head; - - if (ap->dccpap_buf_vector_len + packets > ap->dccpap_buf_len) - return -ENOBUFS; - - gap = packets - 1; - new_head = ap->dccpap_buf_head - packets; - - if (new_head < 0) { - if (gap > 0) { - memset(ap->dccpap_buf, DCCP_ACKPKTS_STATE_NOT_RECEIVED, - gap + new_head + 1); - gap = -new_head; - } - new_head += ap->dccpap_buf_len; - } - - ap->dccpap_buf_head = new_head; - - if (gap > 0) - memset(ap->dccpap_buf + ap->dccpap_buf_head + 1, - DCCP_ACKPKTS_STATE_NOT_RECEIVED, gap); - - ap->dccpap_buf[ap->dccpap_buf_head] = state; - ap->dccpap_buf_vector_len += packets; - return 0; -} - -/* - * Implements the draft-ietf-dccp-spec-11.txt Appendix A - */ -int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk, - u64 ackno, u8 state) -{ - /* - * Check at the right places if the buffer is full, if it is, tell the - * caller to start dropping packets till the HC-Sender acks our ACK - * vectors, when we will free up space in dccpap_buf. - * - * We may well decide to do buffer compression, etc, but for now lets - * just drop. - * - * From Appendix A: - * - * Of course, the circular buffer may overflow, either when the - * HC-Sender is sending data at a very high rate, when the - * HC-Receiver's acknowledgements are not reaching the HC-Sender, - * or when the HC-Sender is forgetting to acknowledge those acks - * (so the HC-Receiver is unable to clean up old state). In this - * case, the HC-Receiver should either compress the buffer (by - * increasing run lengths when possible), transfer its state to - * a larger buffer, or, as a last resort, drop all received - * packets, without processing them whatsoever, until its buffer - * shrinks again. - */ - - /* See if this is the first ackno being inserted */ - if (ap->dccpap_buf_vector_len == 0) { - ap->dccpap_buf[ap->dccpap_buf_head] = state; - ap->dccpap_buf_vector_len = 1; - } else if (after48(ackno, ap->dccpap_buf_ackno)) { - const u64 delta = dccp_delta_seqno(ap->dccpap_buf_ackno, - ackno); - - /* - * Look if the state of this packet is the same as the - * previous ackno and if so if we can bump the head len. - */ - if (delta == 1 && - dccp_ackpkts_state(ap, ap->dccpap_buf_head) == state && - (dccp_ackpkts_len(ap, ap->dccpap_buf_head) < - DCCP_ACKPKTS_LEN_MASK)) - ap->dccpap_buf[ap->dccpap_buf_head]++; - else if (dccp_ackpkts_set_buf_head_state(ap, delta, state)) - return -ENOBUFS; - } else { - /* - * A.1.2. Old Packets - * - * When a packet with Sequence Number S arrives, and - * S <= buf_ackno, the HC-Receiver will scan the table - * for the byte corresponding to S. (Indexing structures - * could reduce the complexity of this scan.) - */ - u64 delta = dccp_delta_seqno(ackno, ap->dccpap_buf_ackno); - unsigned int index = ap->dccpap_buf_head; - - while (1) { - const u8 len = dccp_ackpkts_len(ap, index); - const u8 state = dccp_ackpkts_state(ap, index); - /* - * valid packets not yet in dccpap_buf have a reserved - * entry, with a len equal to 0. - */ - if (state == DCCP_ACKPKTS_STATE_NOT_RECEIVED && - len == 0 && delta == 0) { /* Found our - reserved seat! */ - dccp_pr_debug("Found %llu reserved seat!\n", - (unsigned long long) ackno); - ap->dccpap_buf[index] = state; - goto out; - } - /* len == 0 means one packet */ - if (delta < len + 1) - goto out_duplicate; - - delta -= len + 1; - if (++index == ap->dccpap_buf_len) - index = 0; - } - } - - ap->dccpap_buf_ackno = ackno; - dccp_timestamp(sk, &ap->dccpap_time); -out: - dccp_pr_debug(""); - dccp_ackpkts_print(ap); - return 0; - -out_duplicate: - /* Duplicate packet */ - dccp_pr_debug("Received a dup or already considered lost " - "packet: %llu\n", (unsigned long long) ackno); - return -EILSEQ; -} - -#ifdef CONFIG_IP_DCCP_DEBUG -void dccp_ackvector_print(const u64 ackno, const unsigned char *vector, - int len) -{ - if (!dccp_debug) - return; - - printk("ACK vector len=%d, ackno=%llu |", len, - (unsigned long long) ackno); - - while (len--) { - const u8 state = (*vector & DCCP_ACKPKTS_STATE_MASK) >> 6; - const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK); - - printk("%d,%d|", state, rl); - ++vector; - } - - printk("\n"); -} - -void dccp_ackpkts_print(const struct dccp_ackpkts *ap) -{ - dccp_ackvector_print(ap->dccpap_buf_ackno, - ap->dccpap_buf + ap->dccpap_buf_head, - ap->dccpap_buf_vector_len); -} -#endif - -static void dccp_ackpkts_trow_away_ack_record(struct dccp_ackpkts *ap) -{ - /* - * As we're keeping track of the ack vector size - * (dccpap_buf_vector_len) and the sent ack vector size - * (dccpap_ack_vector_len) we don't need dccpap_buf_tail at all, but - * keep this code here as in the future we'll implement a vector of - * ack records, as suggested in draft-ietf-dccp-spec-11.txt - * Appendix A. -acme - */ -#if 0 - ap->dccpap_buf_tail = ap->dccpap_ack_ptr + 1; - if (ap->dccpap_buf_tail >= ap->dccpap_buf_len) - ap->dccpap_buf_tail -= ap->dccpap_buf_len; -#endif - ap->dccpap_buf_vector_len -= ap->dccpap_ack_vector_len; -} - -void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap, struct sock *sk, - u64 ackno) -{ - /* Check if we actually sent an ACK vector */ - if (ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1) - return; - - if (ackno == ap->dccpap_ack_seqno) { -#ifdef CONFIG_IP_DCCP_DEBUG - struct dccp_sock *dp = dccp_sk(sk); - const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ? - "CLIENT rx ack: " : "server rx ack: "; -#endif - dccp_pr_debug("%sACK packet 0, len=%d, ack_seqno=%llu, " - "ack_ackno=%llu, ACKED!\n", - debug_prefix, 1, - (unsigned long long) ap->dccpap_ack_seqno, - (unsigned long long) ap->dccpap_ack_ackno); - dccp_ackpkts_trow_away_ack_record(ap); - ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1; - } -} - -static void dccp_ackpkts_check_rcv_ackvector(struct dccp_ackpkts *ap, - struct sock *sk, u64 ackno, - const unsigned char len, - const unsigned char *vector) -{ - unsigned char i; - - /* Check if we actually sent an ACK vector */ - if (ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1) - return; - /* - * We're in the receiver half connection, so if the received an ACK - * vector ackno (e.g. 50) before dccpap_ack_seqno (e.g. 52), we're - * not interested. - * - * Extra explanation with example: - * - * if we received an ACK vector with ackno 50, it can only be acking - * 50, 49, 48, etc, not 52 (the seqno for the ACK vector we sent). - */ - /* dccp_pr_debug("is %llu < %llu? ", ackno, ap->dccpap_ack_seqno); */ - if (before48(ackno, ap->dccpap_ack_seqno)) { - /* dccp_pr_debug_cat("yes\n"); */ - return; - } - /* dccp_pr_debug_cat("no\n"); */ - - i = len; - while (i--) { - const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK); - u64 ackno_end_rl; - - dccp_set_seqno(&ackno_end_rl, ackno - rl); - - /* - * dccp_pr_debug("is %llu <= %llu <= %llu? ", ackno_end_rl, - * ap->dccpap_ack_seqno, ackno); - */ - if (between48(ap->dccpap_ack_seqno, ackno_end_rl, ackno)) { - const u8 state = (*vector & - DCCP_ACKPKTS_STATE_MASK) >> 6; - /* dccp_pr_debug_cat("yes\n"); */ - - if (state != DCCP_ACKPKTS_STATE_NOT_RECEIVED) { -#ifdef CONFIG_IP_DCCP_DEBUG - struct dccp_sock *dp = dccp_sk(sk); - const char *debug_prefix = - dp->dccps_role == DCCP_ROLE_CLIENT ? - "CLIENT rx ack: " : "server rx ack: "; -#endif - dccp_pr_debug("%sACK vector 0, len=%d, " - "ack_seqno=%llu, ack_ackno=%llu, " - "ACKED!\n", - debug_prefix, len, - (unsigned long long) - ap->dccpap_ack_seqno, - (unsigned long long) - ap->dccpap_ack_ackno); - dccp_ackpkts_trow_away_ack_record(ap); - } - /* - * If dccpap_ack_seqno was not received, no problem - * we'll send another ACK vector. - */ - ap->dccpap_ack_seqno = DCCP_MAX_SEQNO + 1; - break; - } - /* dccp_pr_debug_cat("no\n"); */ - - dccp_set_seqno(&ackno, ackno_end_rl - 1); - ++vector; - } -} diff --git a/net/dccp/output.c b/net/dccp/output.c index ea6d0e9..4786bdc 100644 --- a/net/dccp/output.c +++ b/net/dccp/output.c @@ -16,6 +16,7 @@ #include <net/sock.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" @@ -85,7 +86,7 @@ int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb) switch (dcb->dccpd_type) { case DCCP_PKT_REQUEST: dccp_hdr_request(skb)->dccph_req_service = - dcb->dccpd_service; + dp->dccps_service; break; case DCCP_PKT_RESET: dccp_hdr_reset(skb)->dccph_reset_code = @@ -225,7 +226,6 @@ int dccp_write_xmit(struct sock *sk, struct sk_buff *skb, long *timeo) err = dccp_wait_for_ccid(sk, skb, timeo); if (err == 0) { - const struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts; struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb); const int len = skb->len; @@ -236,15 +236,7 @@ int dccp_write_xmit(struct sock *sk, struct sk_buff *skb, long *timeo) inet_csk(sk)->icsk_rto, DCCP_RTO_MAX); dcb->dccpd_type = DCCP_PKT_DATAACK; - /* - * FIXME: we really should have a - * dccps_ack_pending or use icsk. - */ - } else if (inet_csk_ack_scheduled(sk) || - dp->dccps_timestamp_echo != 0 || - (dp->dccps_options.dccpo_send_ack_vector && - ap->dccpap_buf_ackno != DCCP_MAX_SEQNO + 1 && - ap->dccpap_ack_seqno == DCCP_MAX_SEQNO + 1)) + } else if (dccp_ack_pending(sk)) dcb->dccpd_type = DCCP_PKT_DATAACK; else dcb->dccpd_type = DCCP_PKT_DATA; @@ -270,6 +262,7 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, struct request_sock *req) { struct dccp_hdr *dh; + struct dccp_request_sock *dreq; const int dccp_header_size = sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext) + sizeof(struct dccp_hdr_response); @@ -285,8 +278,9 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, skb->dst = dst_clone(dst); skb->csum = 0; + dreq = dccp_rsk(req); DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE; - DCCP_SKB_CB(skb)->dccpd_seq = dccp_rsk(req)->dreq_iss; + DCCP_SKB_CB(skb)->dccpd_seq = dreq->dreq_iss; dccp_insert_options(sk, skb); skb->h.raw = skb_push(skb, dccp_header_size); @@ -300,8 +294,9 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, DCCP_SKB_CB(skb)->dccpd_opt_len) / 4; dh->dccph_type = DCCP_PKT_RESPONSE; dh->dccph_x = 1; - dccp_hdr_set_seq(dh, dccp_rsk(req)->dreq_iss); - dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dccp_rsk(req)->dreq_isr); + dccp_hdr_set_seq(dh, dreq->dreq_iss); + dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dreq->dreq_isr); + dccp_hdr_response(skb)->dccph_resp_service = dreq->dreq_service; dh->dccph_checksum = dccp_v4_checksum(skb, inet_rsk(req)->loc_addr, inet_rsk(req)->rmt_addr); @@ -397,9 +392,6 @@ int dccp_connect(struct sock *sk) skb_reserve(skb, MAX_DCCP_HEADER); DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST; - /* FIXME: set service to something meaningful, coming - * from userspace*/ - DCCP_SKB_CB(skb)->dccpd_service = 0; skb->csum = 0; skb_set_owner_w(skb, sk); diff --git a/net/dccp/proto.c b/net/dccp/proto.c index 18a0e69..a1cfd0e 100644 --- a/net/dccp/proto.c +++ b/net/dccp/proto.c @@ -94,7 +94,15 @@ EXPORT_SYMBOL_GPL(dccp_state_name); static inline int dccp_listen_start(struct sock *sk) { - dccp_sk(sk)->dccps_role = DCCP_ROLE_LISTEN; + struct dccp_sock *dp = dccp_sk(sk); + + dp->dccps_role = DCCP_ROLE_LISTEN; + /* + * Apps need to use setsockopt(DCCP_SOCKOPT_SERVICE) + * before calling listen() + */ + if (dccp_service_not_initialized(sk)) + return -EPROTO; return inet_csk_listen_start(sk, TCP_SYNQ_HSIZE); } @@ -202,6 +210,42 @@ int dccp_ioctl(struct sock *sk, int cmd, unsigned long arg) return -ENOIOCTLCMD; } +static int dccp_setsockopt_service(struct sock *sk, const u32 service, + char __user *optval, int optlen) +{ + struct dccp_sock *dp = dccp_sk(sk); + struct dccp_service_list *sl = NULL; + + if (service == DCCP_SERVICE_INVALID_VALUE || + optlen > DCCP_SERVICE_LIST_MAX_LEN * sizeof(u32)) + return -EINVAL; + + if (optlen > sizeof(service)) { + sl = kmalloc(optlen, GFP_KERNEL); + if (sl == NULL) + return -ENOMEM; + + sl->dccpsl_nr = optlen / sizeof(u32) - 1; + if (copy_from_user(sl->dccpsl_list, + optval + sizeof(service), + optlen - sizeof(service)) || + dccp_list_has_service(sl, DCCP_SERVICE_INVALID_VALUE)) { + kfree(sl); + return -EFAULT; + } + } + + lock_sock(sk); + dp->dccps_service = service; + + if (dp->dccps_service_list != NULL) + kfree(dp->dccps_service_list); + + dp->dccps_service_list = sl; + release_sock(sk); + return 0; +} + int dccp_setsockopt(struct sock *sk, int level, int optname, char __user *optval, int optlen) { @@ -218,8 +262,10 @@ int dccp_setsockopt(struct sock *sk, int level, int optname, if (get_user(val, (int __user *)optval)) return -EFAULT; - lock_sock(sk); + if (optname == DCCP_SOCKOPT_SERVICE) + return dccp_setsockopt_service(sk, val, optval, optlen); + lock_sock(sk); dp = dccp_sk(sk); err = 0; @@ -236,6 +282,37 @@ int dccp_setsockopt(struct sock *sk, int level, int optname, return err; } +static int dccp_getsockopt_service(struct sock *sk, int len, + u32 __user *optval, + int __user *optlen) +{ + const struct dccp_sock *dp = dccp_sk(sk); + const struct dccp_service_list *sl; + int err = -ENOENT, slen = 0, total_len = sizeof(u32); + + lock_sock(sk); + if (dccp_service_not_initialized(sk)) + goto out; + + if ((sl = dp->dccps_service_list) != NULL) { + slen = sl->dccpsl_nr * sizeof(u32); + total_len += slen; + } + + err = -EINVAL; + if (total_len > len) + goto out; + + err = 0; + if (put_user(total_len, optlen) || + put_user(dp->dccps_service, optval) || + (sl != NULL && copy_to_user(optval + 1, sl->dccpsl_list, slen))) + err = -EFAULT; +out: + release_sock(sk); + return err; +} + int dccp_getsockopt(struct sock *sk, int level, int optname, char __user *optval, int __user *optlen) { @@ -248,8 +325,7 @@ int dccp_getsockopt(struct sock *sk, int level, int optname, if (get_user(len, optlen)) return -EFAULT; - len = min_t(unsigned int, len, sizeof(int)); - if (len < 0) + if (len < sizeof(int)) return -EINVAL; dp = dccp_sk(sk); @@ -257,7 +333,17 @@ int dccp_getsockopt(struct sock *sk, int level, int optname, switch (optname) { case DCCP_SOCKOPT_PACKET_SIZE: val = dp->dccps_packet_size; + len = sizeof(dp->dccps_packet_size); break; + case DCCP_SOCKOPT_SERVICE: + return dccp_getsockopt_service(sk, len, + (u32 __user *)optval, optlen); + case 128 ... 191: + return ccid_hc_rx_getsockopt(dp->dccps_hc_rx_ccid, sk, optname, + len, (u32 __user *)optval, optlen); + case 192 ... 255: + return ccid_hc_tx_getsockopt(dp->dccps_hc_tx_ccid, sk, optname, + len, (u32 __user *)optval, optlen); default: return -ENOPROTOOPT; } diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig index 30aa8e2..e2162d2 100644 --- a/net/ipv4/netfilter/Kconfig +++ b/net/ipv4/netfilter/Kconfig @@ -51,6 +51,14 @@ config IP_NF_CONNTRACK_EVENTS IF unsure, say `N'. +config IP_NF_CONNTRACK_NETLINK + tristate 'Connection tracking netlink interface' + depends on IP_NF_CONNTRACK && NETFILTER_NETLINK + depends on IP_NF_CONNTRACK!=y || NETFILTER_NETLINK!=m + help + This option enables support for a netlink-based userspace interface + + config IP_NF_CT_PROTO_SCTP tristate 'SCTP protocol connection tracking support (EXPERIMENTAL)' depends on IP_NF_CONNTRACK && EXPERIMENTAL @@ -774,11 +782,5 @@ config IP_NF_ARP_MANGLE Allows altering the ARP packet payload: source and destination hardware and network addresses. -config IP_NF_CONNTRACK_NETLINK - tristate 'Connection tracking netlink interface' - depends on IP_NF_CONNTRACK && NETFILTER_NETLINK - help - This option enables support for a netlink-based userspace interface - endmenu diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c index 19cba16..f8cd8e4 100644 --- a/net/ipv4/netfilter/ip_conntrack_core.c +++ b/net/ipv4/netfilter/ip_conntrack_core.c @@ -1143,7 +1143,10 @@ void ip_ct_refresh_acct(struct ip_conntrack *ct, if (del_timer(&ct->timeout)) { ct->timeout.expires = jiffies + extra_jiffies; add_timer(&ct->timeout); - ip_conntrack_event_cache(IPCT_REFRESH, skb); + /* FIXME: We loose some REFRESH events if this function + * is called without an skb. I'll fix this later -HW */ + if (skb) + ip_conntrack_event_cache(IPCT_REFRESH, skb); } ct_add_counters(ct, ctinfo, skb); write_unlock_bh(&ip_conntrack_lock); diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c index 7d38913..9bcb398 100644 --- a/net/ipv4/netfilter/ipt_CLUSTERIP.c +++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c @@ -13,6 +13,7 @@ #include <linux/config.h> #include <linux/proc_fs.h> #include <linux/jhash.h> +#include <linux/bitops.h> #include <linux/skbuff.h> #include <linux/ip.h> #include <linux/tcp.h> @@ -30,7 +31,7 @@ #include <linux/netfilter_ipv4/ipt_CLUSTERIP.h> #include <linux/netfilter_ipv4/ip_conntrack.h> -#define CLUSTERIP_VERSION "0.7" +#define CLUSTERIP_VERSION "0.8" #define DEBUG_CLUSTERIP @@ -49,13 +50,14 @@ MODULE_DESCRIPTION("iptables target for CLUSTERIP"); struct clusterip_config { struct list_head list; /* list of all configs */ atomic_t refcount; /* reference count */ + atomic_t entries; /* number of entries/rules + * referencing us */ u_int32_t clusterip; /* the IP address */ u_int8_t clustermac[ETH_ALEN]; /* the MAC address */ struct net_device *dev; /* device */ u_int16_t num_total_nodes; /* total number of nodes */ - u_int16_t num_local_nodes; /* number of local nodes */ - u_int16_t local_nodes[CLUSTERIP_MAX_NODES]; /* node number array */ + unsigned long local_nodes; /* node number array */ #ifdef CONFIG_PROC_FS struct proc_dir_entry *pde; /* proc dir entry */ @@ -66,8 +68,7 @@ struct clusterip_config { static LIST_HEAD(clusterip_configs); -/* clusterip_lock protects the clusterip_configs list _AND_ the configurable - * data within all structurses (num_local_nodes, local_nodes[]) */ +/* clusterip_lock protects the clusterip_configs list */ static DEFINE_RWLOCK(clusterip_lock); #ifdef CONFIG_PROC_FS @@ -76,23 +77,48 @@ static struct proc_dir_entry *clusterip_procdir; #endif static inline void -clusterip_config_get(struct clusterip_config *c) { +clusterip_config_get(struct clusterip_config *c) +{ atomic_inc(&c->refcount); } static inline void -clusterip_config_put(struct clusterip_config *c) { - if (atomic_dec_and_test(&c->refcount)) { +clusterip_config_put(struct clusterip_config *c) +{ + if (atomic_dec_and_test(&c->refcount)) + kfree(c); +} + +/* increase the count of entries(rules) using/referencing this config */ +static inline void +clusterip_config_entry_get(struct clusterip_config *c) +{ + atomic_inc(&c->entries); +} + +/* decrease the count of entries using/referencing this config. If last + * entry(rule) is removed, remove the config from lists, but don't free it + * yet, since proc-files could still be holding references */ +static inline void +clusterip_config_entry_put(struct clusterip_config *c) +{ + if (atomic_dec_and_test(&c->entries)) { write_lock_bh(&clusterip_lock); list_del(&c->list); write_unlock_bh(&clusterip_lock); + dev_mc_delete(c->dev, c->clustermac, ETH_ALEN, 0); dev_put(c->dev); - kfree(c); + + /* In case anyone still accesses the file, the open/close + * functions are also incrementing the refcount on their own, + * so it's safe to remove the entry even if it's in use. */ +#ifdef CONFIG_PROC_FS + remove_proc_entry(c->pde->name, c->pde->parent); +#endif } } - static struct clusterip_config * __clusterip_config_find(u_int32_t clusterip) { @@ -111,7 +137,7 @@ __clusterip_config_find(u_int32_t clusterip) } static inline struct clusterip_config * -clusterip_config_find_get(u_int32_t clusterip) +clusterip_config_find_get(u_int32_t clusterip, int entry) { struct clusterip_config *c; @@ -122,11 +148,24 @@ clusterip_config_find_get(u_int32_t clusterip) return NULL; } atomic_inc(&c->refcount); + if (entry) + atomic_inc(&c->entries); read_unlock_bh(&clusterip_lock); return c; } +static void +clusterip_config_init_nodelist(struct clusterip_config *c, + const struct ipt_clusterip_tgt_info *i) +{ + int n; + + for (n = 0; n < i->num_local_nodes; n++) { + set_bit(i->local_nodes[n] - 1, &c->local_nodes); + } +} + static struct clusterip_config * clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip, struct net_device *dev) @@ -143,11 +182,11 @@ clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip, c->clusterip = ip; memcpy(&c->clustermac, &i->clustermac, ETH_ALEN); c->num_total_nodes = i->num_total_nodes; - c->num_local_nodes = i->num_local_nodes; - memcpy(&c->local_nodes, &i->local_nodes, sizeof(c->local_nodes)); + clusterip_config_init_nodelist(c, i); c->hash_mode = i->hash_mode; c->hash_initval = i->hash_initval; atomic_set(&c->refcount, 1); + atomic_set(&c->entries, 1); #ifdef CONFIG_PROC_FS /* create proc dir entry */ @@ -171,53 +210,28 @@ clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip, static int clusterip_add_node(struct clusterip_config *c, u_int16_t nodenum) { - int i; - - write_lock_bh(&clusterip_lock); - if (c->num_local_nodes >= CLUSTERIP_MAX_NODES - || nodenum > CLUSTERIP_MAX_NODES) { - write_unlock_bh(&clusterip_lock); + if (nodenum == 0 || + nodenum > c->num_total_nodes) return 1; - } - - /* check if we alrady have this number in our array */ - for (i = 0; i < c->num_local_nodes; i++) { - if (c->local_nodes[i] == nodenum) { - write_unlock_bh(&clusterip_lock); - return 1; - } - } - c->local_nodes[c->num_local_nodes++] = nodenum; + /* check if we already have this number in our bitfield */ + if (test_and_set_bit(nodenum - 1, &c->local_nodes)) + return 1; - write_unlock_bh(&clusterip_lock); return 0; } static int clusterip_del_node(struct clusterip_config *c, u_int16_t nodenum) { - int i; - - write_lock_bh(&clusterip_lock); - - if (c->num_local_nodes <= 1 || nodenum > CLUSTERIP_MAX_NODES) { - write_unlock_bh(&clusterip_lock); + if (nodenum == 0 || + nodenum > c->num_total_nodes) return 1; - } - for (i = 0; i < c->num_local_nodes; i++) { - if (c->local_nodes[i] == nodenum) { - int size = sizeof(u_int16_t)*(c->num_local_nodes-(i+1)); - memmove(&c->local_nodes[i], &c->local_nodes[i+1], size); - c->num_local_nodes--; - write_unlock_bh(&clusterip_lock); - return 0; - } - } + if (test_and_clear_bit(nodenum - 1, &c->local_nodes)) + return 0; - write_unlock_bh(&clusterip_lock); return 1; } @@ -285,25 +299,7 @@ clusterip_hashfn(struct sk_buff *skb, struct clusterip_config *config) static inline int clusterip_responsible(struct clusterip_config *config, u_int32_t hash) { - int i; - - read_lock_bh(&clusterip_lock); - - if (config->num_local_nodes == 0) { - read_unlock_bh(&clusterip_lock); - return 0; - } - - for (i = 0; i < config->num_local_nodes; i++) { - if (config->local_nodes[i] == hash) { - read_unlock_bh(&clusterip_lock); - return 1; - } - } - - read_unlock_bh(&clusterip_lock); - - return 0; + return test_bit(hash - 1, &config->local_nodes); } /*********************************************************************** @@ -415,8 +411,26 @@ checkentry(const char *tablename, /* FIXME: further sanity checks */ - config = clusterip_config_find_get(e->ip.dst.s_addr); - if (!config) { + config = clusterip_config_find_get(e->ip.dst.s_addr, 1); + if (config) { + if (cipinfo->config != NULL) { + /* Case A: This is an entry that gets reloaded, since + * it still has a cipinfo->config pointer. Simply + * increase the entry refcount and return */ + if (cipinfo->config != config) { + printk(KERN_ERR "CLUSTERIP: Reloaded entry " + "has invalid config pointer!\n"); + return 0; + } + clusterip_config_entry_get(cipinfo->config); + } else { + /* Case B: This is a new rule referring to an existing + * clusterip config. */ + cipinfo->config = config; + clusterip_config_entry_get(cipinfo->config); + } + } else { + /* Case C: This is a completely new clusterip config */ if (!(cipinfo->flags & CLUSTERIP_FLAG_NEW)) { printk(KERN_WARNING "CLUSTERIP: no config found for %u.%u.%u.%u, need 'new'\n", NIPQUAD(e->ip.dst.s_addr)); return 0; @@ -443,10 +457,9 @@ checkentry(const char *tablename, } dev_mc_add(config->dev,config->clustermac, ETH_ALEN, 0); } + cipinfo->config = config; } - cipinfo->config = config; - return 1; } @@ -455,13 +468,10 @@ static void destroy(void *matchinfo, unsigned int matchinfosize) { struct ipt_clusterip_tgt_info *cipinfo = matchinfo; - /* we first remove the proc entry and then drop the reference - * count. In case anyone still accesses the file, the open/close - * functions are also incrementing the refcount on their own */ -#ifdef CONFIG_PROC_FS - remove_proc_entry(cipinfo->config->pde->name, - cipinfo->config->pde->parent); -#endif + /* if no more entries are referencing the config, remove it + * from the list and destroy the proc entry */ + clusterip_config_entry_put(cipinfo->config); + clusterip_config_put(cipinfo->config); } @@ -533,7 +543,7 @@ arp_mangle(unsigned int hook, /* if there is no clusterip configuration for the arp reply's * source ip, we don't want to mangle it */ - c = clusterip_config_find_get(payload->src_ip); + c = clusterip_config_find_get(payload->src_ip, 0); if (!c) return NF_ACCEPT; @@ -574,56 +584,69 @@ static struct nf_hook_ops cip_arp_ops = { #ifdef CONFIG_PROC_FS +struct clusterip_seq_position { + unsigned int pos; /* position */ + unsigned int weight; /* number of bits set == size */ + unsigned int bit; /* current bit */ + unsigned long val; /* current value */ +}; + static void *clusterip_seq_start(struct seq_file *s, loff_t *pos) { struct proc_dir_entry *pde = s->private; struct clusterip_config *c = pde->data; - unsigned int *nodeidx; - - read_lock_bh(&clusterip_lock); - if (*pos >= c->num_local_nodes) + unsigned int weight; + u_int32_t local_nodes; + struct clusterip_seq_position *idx; + + /* FIXME: possible race */ + local_nodes = c->local_nodes; + weight = hweight32(local_nodes); + if (*pos >= weight) return NULL; - nodeidx = kmalloc(sizeof(unsigned int), GFP_KERNEL); - if (!nodeidx) + idx = kmalloc(sizeof(struct clusterip_seq_position), GFP_KERNEL); + if (!idx) return ERR_PTR(-ENOMEM); - *nodeidx = *pos; - return nodeidx; + idx->pos = *pos; + idx->weight = weight; + idx->bit = ffs(local_nodes); + idx->val = local_nodes; + clear_bit(idx->bit - 1, &idx->val); + + return idx; } static void *clusterip_seq_next(struct seq_file *s, void *v, loff_t *pos) { - struct proc_dir_entry *pde = s->private; - struct clusterip_config *c = pde->data; - unsigned int *nodeidx = (unsigned int *)v; + struct clusterip_seq_position *idx = (struct clusterip_seq_position *)v; - *pos = ++(*nodeidx); - if (*pos >= c->num_local_nodes) { + *pos = ++idx->pos; + if (*pos >= idx->weight) { kfree(v); return NULL; } - return nodeidx; + idx->bit = ffs(idx->val); + clear_bit(idx->bit - 1, &idx->val); + return idx; } static void clusterip_seq_stop(struct seq_file *s, void *v) { kfree(v); - - read_unlock_bh(&clusterip_lock); } static int clusterip_seq_show(struct seq_file *s, void *v) { - struct proc_dir_entry *pde = s->private; - struct clusterip_config *c = pde->data; - unsigned int *nodeidx = (unsigned int *)v; + struct clusterip_seq_position *idx = (struct clusterip_seq_position *)v; - if (*nodeidx != 0) + if (idx->pos != 0) seq_putc(s, ','); - seq_printf(s, "%u", c->local_nodes[*nodeidx]); - if (*nodeidx == c->num_local_nodes-1) + seq_printf(s, "%u", idx->bit); + + if (idx->pos == idx->weight - 1) seq_putc(s, '\n'); return 0; diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 2b9bf9b..6001948 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -639,6 +639,7 @@ static int udpv6_sendmsg(struct kiocb *iocb, struct sock *sk, int tclass = -1; int corkreq = up->corkflag || msg->msg_flags&MSG_MORE; int err; + int connected = 0; /* destination address check */ if (sin6) { @@ -748,6 +749,7 @@ do_udp_sendmsg: fl->fl_ip_dport = inet->dport; daddr = &np->daddr; fl->fl6_flowlabel = np->flow_label; + connected = 1; } if (!fl->oif) @@ -770,6 +772,7 @@ do_udp_sendmsg: } if (!(opt->opt_nflen|opt->opt_flen)) opt = NULL; + connected = 0; } if (opt == NULL) opt = np->opt; @@ -787,10 +790,13 @@ do_udp_sendmsg: ipv6_addr_copy(&final, &fl->fl6_dst); ipv6_addr_copy(&fl->fl6_dst, rt0->addr); final_p = &final; + connected = 0; } - if (!fl->oif && ipv6_addr_is_multicast(&fl->fl6_dst)) + if (!fl->oif && ipv6_addr_is_multicast(&fl->fl6_dst)) { fl->oif = np->mcast_oif; + connected = 0; + } err = ip6_dst_lookup(sk, &dst, fl); if (err) @@ -846,7 +852,7 @@ do_append_data: else if (!corkreq) err = udp_v6_push_pending_frames(sk, up); - if (dst) + if (dst && connected) ip6_dst_store(sk, dst, ipv6_addr_equal(&fl->fl6_dst, &np->daddr) ? &np->daddr : NULL); diff --git a/net/socket.c b/net/socket.c index c699e93..f926447 100644 --- a/net/socket.c +++ b/net/socket.c @@ -1862,7 +1862,8 @@ asmlinkage long sys_recvmsg(int fd, struct msghdr __user *msg, unsigned int flag if (err < 0) goto out_freeiov; } - err = __put_user(msg_sys.msg_flags, COMPAT_FLAGS(msg)); + err = __put_user((msg_sys.msg_flags & ~MSG_CMSG_COMPAT), + COMPAT_FLAGS(msg)); if (err) goto out_freeiov; if (MSG_CMSG_COMPAT & flags) |