diff options
Diffstat (limited to 'net/dccp/options.c')
| -rw-r--r-- | net/dccp/options.c | 958 |
1 files changed, 356 insertions, 602 deletions
diff --git a/net/dccp/options.c b/net/dccp/options.c index 382c5894acb..9bce31886bd 100644 --- a/net/dccp/options.c +++ b/net/dccp/options.c @@ -4,65 +4,54 @@ * An implementation of the DCCP protocol * Copyright (c) 2005 Aristeu Sergio Rozanski Filho <aris@cathedrallabs.org> * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@ghostprotocols.net> - * Copyright (c) 2005 Ian McDonald <iam4@cs.waikato.ac.nz> + * Copyright (c) 2005 Ian McDonald <ian.mcdonald@jandi.co.nz> * * 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; either version * 2 of the License, or (at your option) any later version. */ -#include <linux/config.h> #include <linux/dccp.h> #include <linux/module.h> #include <linux/types.h> +#include <asm/unaligned.h> #include <linux/kernel.h> #include <linux/skbuff.h> +#include "ackvec.h" #include "ccid.h" #include "dccp.h" +#include "feat.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_send_ack_vector = DCCPF_INITIAL_SEND_ACK_VECTOR, - .dccpo_send_ndp_count = DCCPF_INITIAL_SEND_NDP_COUNT, -}; - -void dccp_options_init(struct dccp_options *dccpo) +u64 dccp_decode_value_var(const u8 *bf, const u8 len) { - memcpy(dccpo, &dccpo_default_values, sizeof(*dccpo)); -} - -static u32 dccp_decode_value_var(const unsigned char *bf, const u8 len) -{ - u32 value = 0; + u64 value = 0; + if (len >= DCCP_OPTVAL_MAXLEN) + value += ((u64)*bf++) << 40; + if (len > 4) + value += ((u64)*bf++) << 32; if (len > 3) - value += *bf++ << 24; + value += ((u64)*bf++) << 24; if (len > 2) - value += *bf++ << 16; + value += ((u64)*bf++) << 16; if (len > 1) - value += *bf++ << 8; + value += ((u64)*bf++) << 8; if (len > 0) value += *bf; return value; } -int dccp_parse_options(struct sock *sk, struct sk_buff *skb) +/** + * dccp_parse_options - Parse DCCP options present in @skb + * @sk: client|server|listening dccp socket (when @dreq != NULL) + * @dreq: request socket to use during connection setup, or NULL + */ +int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq, + 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 rx opt: " : "server rx opt: "; -#endif const struct dccp_hdr *dh = dccp_hdr(skb); const u8 pkt_type = DCCP_SKB_CB(skb)->dccpd_type; unsigned char *options = (unsigned char *)dh + dccp_hdr_len(skb); @@ -71,10 +60,15 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb) (dh->dccph_doff * 4); struct dccp_options_received *opt_recv = &dp->dccps_options_received; unsigned char opt, len; - unsigned char *value; + unsigned char *uninitialized_var(value); + u32 elapsed_time; + __be32 opt_val; + int rc; + int mandatory = 0; memset(opt_recv, 0, sizeof(*opt_recv)); + opt = len = 0; while (opt_ptr != opt_end) { opt = *opt_ptr++; len = 0; @@ -83,11 +77,11 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb) /* Check if this isn't a single byte option */ if (opt > DCCPO_MAX_RESERVED) { if (opt_ptr == opt_end) - goto out_invalid_option; + goto out_nonsensical_length; len = *opt_ptr++; - if (len < 3) - goto out_invalid_option; + if (len < 2) + goto out_nonsensical_length; /* * Remove the type and len fields, leaving * just the value size @@ -97,145 +91,187 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb) opt_ptr += len; if (opt_ptr > opt_end) - goto out_invalid_option; + goto out_nonsensical_length; } + /* + * CCID-specific options are ignored during connection setup, as + * negotiation may still be in progress (see RFC 4340, 10.3). + * The same applies to Ack Vectors, as these depend on the CCID. + */ + if (dreq != NULL && (opt >= DCCPO_MIN_RX_CCID_SPECIFIC || + opt == DCCPO_ACK_VECTOR_0 || opt == DCCPO_ACK_VECTOR_1)) + goto ignore_option; + switch (opt) { case DCCPO_PADDING: break; + case DCCPO_MANDATORY: + if (mandatory) + goto out_invalid_option; + if (pkt_type != DCCP_PKT_DATA) + mandatory = 1; + break; case DCCPO_NDP_COUNT: - if (len > 3) + if (len > 6) goto out_invalid_option; opt_recv->dccpor_ndp = dccp_decode_value_var(value, len); - dccp_pr_debug("%sNDP count=%d\n", debug_prefix, - opt_recv->dccpor_ndp); + dccp_pr_debug("%s opt: NDP count=%llu\n", dccp_role(sk), + (unsigned long long)opt_recv->dccpor_ndp); break; - case DCCPO_ACK_VECTOR_0: - if (len > DCCP_MAX_ACK_VECTOR_LEN) + case DCCPO_CHANGE_L ... DCCPO_CONFIRM_R: + if (pkt_type == DCCP_PKT_DATA) /* RFC 4340, 6 */ + break; + if (len == 0) goto out_invalid_option; - - 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); + rc = dccp_feat_parse_options(sk, dreq, mandatory, opt, + *value, value + 1, len - 1); + if (rc) + goto out_featneg_failed; break; case DCCPO_TIMESTAMP: if (len != 4) goto out_invalid_option; + /* + * RFC 4340 13.1: "The precise time corresponding to + * Timestamp Value zero is not specified". We use + * zero to indicate absence of a meaningful timestamp. + */ + opt_val = get_unaligned((__be32 *)value); + if (unlikely(opt_val == 0)) { + DCCP_WARN("Timestamp with zero value\n"); + break; + } - opt_recv->dccpor_timestamp = ntohl(*(u32 *)value); - - dp->dccps_timestamp_echo = opt_recv->dccpor_timestamp; - do_gettimeofday(&dp->dccps_timestamp_time); - - dccp_pr_debug("%sTIMESTAMP=%u, ackno=%llu\n", - debug_prefix, opt_recv->dccpor_timestamp, + if (dreq != NULL) { + dreq->dreq_timestamp_echo = ntohl(opt_val); + dreq->dreq_timestamp_time = dccp_timestamp(); + } else { + opt_recv->dccpor_timestamp = + dp->dccps_timestamp_echo = ntohl(opt_val); + dp->dccps_timestamp_time = dccp_timestamp(); + } + dccp_pr_debug("%s rx opt: TIMESTAMP=%u, ackno=%llu\n", + dccp_role(sk), ntohl(opt_val), (unsigned long long) DCCP_SKB_CB(skb)->dccpd_ack_seq); + /* schedule an Ack in case this sender is quiescent */ + inet_csk_schedule_ack(sk); break; case DCCPO_TIMESTAMP_ECHO: if (len != 4 && len != 6 && len != 8) goto out_invalid_option; - opt_recv->dccpor_timestamp_echo = ntohl(*(u32 *)value); + opt_val = get_unaligned((__be32 *)value); + opt_recv->dccpor_timestamp_echo = ntohl(opt_val); - dccp_pr_debug("%sTIMESTAMP_ECHO=%u, len=%d, ackno=%llu, ", - debug_prefix, + dccp_pr_debug("%s rx opt: TIMESTAMP_ECHO=%u, len=%d, " + "ackno=%llu", dccp_role(sk), opt_recv->dccpor_timestamp_echo, len + 2, (unsigned long long) DCCP_SKB_CB(skb)->dccpd_ack_seq); - if (len > 4) { - if (len == 6) - opt_recv->dccpor_elapsed_time = - ntohs(*(u16 *)(value + 4)); - else - opt_recv->dccpor_elapsed_time = - ntohl(*(u32 *)(value + 4)); - - dccp_pr_debug("%sTIMESTAMP_ECHO ELAPSED_TIME=%d\n", - debug_prefix, - opt_recv->dccpor_elapsed_time); + value += 4; + + if (len == 4) { /* no elapsed time included */ + dccp_pr_debug_cat("\n"); + break; + } + + if (len == 6) { /* 2-byte elapsed time */ + __be16 opt_val2 = get_unaligned((__be16 *)value); + elapsed_time = ntohs(opt_val2); + } else { /* 4-byte elapsed time */ + opt_val = get_unaligned((__be32 *)value); + elapsed_time = ntohl(opt_val); } + + dccp_pr_debug_cat(", ELAPSED_TIME=%u\n", elapsed_time); + + /* Give precedence to the biggest ELAPSED_TIME */ + if (elapsed_time > opt_recv->dccpor_elapsed_time) + opt_recv->dccpor_elapsed_time = elapsed_time; break; case DCCPO_ELAPSED_TIME: - if (len != 2 && len != 4) + if (dccp_packet_without_ack(skb)) /* RFC 4340, 13.2 */ + break; + + if (len == 2) { + __be16 opt_val2 = get_unaligned((__be16 *)value); + elapsed_time = ntohs(opt_val2); + } else if (len == 4) { + opt_val = get_unaligned((__be32 *)value); + elapsed_time = ntohl(opt_val); + } else { goto out_invalid_option; + } - if (pkt_type == DCCP_PKT_DATA) - continue; - - if (len == 2) - opt_recv->dccpor_elapsed_time = - ntohs(*(u16 *)value); - else - opt_recv->dccpor_elapsed_time = - ntohl(*(u32 *)value); + if (elapsed_time > opt_recv->dccpor_elapsed_time) + opt_recv->dccpor_elapsed_time = elapsed_time; - dccp_pr_debug("%sELAPSED_TIME=%d\n", debug_prefix, - opt_recv->dccpor_elapsed_time); + dccp_pr_debug("%s rx opt: ELAPSED_TIME=%d\n", + dccp_role(sk), elapsed_time); break; - /* - * From draft-ietf-dccp-spec-11.txt: - * - * Option numbers 128 through 191 are for - * options sent from the HC-Sender to the - * HC-Receiver; option numbers 192 through 255 - * are for options sent from the HC-Receiver to - * the HC-Sender. - */ - case 128 ... 191: { - const u16 idx = value - options; - + case DCCPO_MIN_RX_CCID_SPECIFIC ... DCCPO_MAX_RX_CCID_SPECIFIC: if (ccid_hc_rx_parse_options(dp->dccps_hc_rx_ccid, sk, - opt, len, idx, - value) != 0) + pkt_type, opt, value, len)) goto out_invalid_option; - } break; - case 192 ... 255: { - const u16 idx = value - options; - + case DCCPO_ACK_VECTOR_0: + case DCCPO_ACK_VECTOR_1: + if (dccp_packet_without_ack(skb)) /* RFC 4340, 11.4 */ + break; + /* + * Ack vectors are processed by the TX CCID if it is + * interested. The RX CCID need not parse Ack Vectors, + * since it is only interested in clearing old state. + * Fall through. + */ + case DCCPO_MIN_TX_CCID_SPECIFIC ... DCCPO_MAX_TX_CCID_SPECIFIC: if (ccid_hc_tx_parse_options(dp->dccps_hc_tx_ccid, sk, - opt, len, idx, - value) != 0) + pkt_type, opt, value, len)) goto out_invalid_option; - } break; default: - pr_info("DCCP(%p): option %d(len=%d) not " - "implemented, ignoring\n", - sk, opt, len); + DCCP_CRIT("DCCP(%p): option %d(len=%d) not " + "implemented, ignoring", sk, opt, len); break; - } + } +ignore_option: + if (opt != DCCPO_MANDATORY) + mandatory = 0; } + /* mandatory was the last byte in option list -> reset connection */ + if (mandatory) + goto out_invalid_option; + +out_nonsensical_length: + /* RFC 4340, 5.8: ignore option and all remaining option space */ return 0; out_invalid_option: DCCP_INC_STATS_BH(DCCP_MIB_INVALIDOPT); - DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_OPTION_ERROR; - pr_info("DCCP(%p): invalid option %d, len=%d\n", sk, opt, len); + rc = DCCP_RESET_CODE_OPTION_ERROR; +out_featneg_failed: + DCCP_WARN("DCCP(%p): Option %d (len=%d) error=%u\n", sk, opt, len, rc); + DCCP_SKB_CB(skb)->dccpd_reset_code = rc; + DCCP_SKB_CB(skb)->dccpd_reset_data[0] = opt; + DCCP_SKB_CB(skb)->dccpd_reset_data[1] = len > 0 ? value[0] : 0; + DCCP_SKB_CB(skb)->dccpd_reset_data[2] = len > 1 ? value[1] : 0; return -1; } -static void dccp_encode_value_var(const u32 value, unsigned char *to, - const unsigned int len) +EXPORT_SYMBOL_GPL(dccp_parse_options); + +void dccp_encode_value_var(const u64 value, u8 *to, const u8 len) { + if (len >= DCCP_OPTVAL_MAXLEN) + *to++ = (value & 0xFF0000000000ull) >> 40; + if (len > 4) + *to++ = (value & 0xFF00000000ull) >> 32; if (len > 3) *to++ = (value & 0xFF000000) >> 24; if (len > 2) @@ -246,22 +282,20 @@ static void dccp_encode_value_var(const u32 value, unsigned char *to, *to++ = (value & 0xFF); } -static inline int dccp_ndp_len(const int ndp) +static inline u8 dccp_ndp_len(const u64 ndp) { - return likely(ndp <= 0xFF) ? 1 : ndp <= 0xFFFF ? 2 : 3; + if (likely(ndp <= 0xFF)) + return 1; + return likely(ndp <= USHRT_MAX) ? 2 : (ndp <= UINT_MAX ? 4 : 6); } -void dccp_insert_option(struct sock *sk, struct sk_buff *skb, - const unsigned char option, - const void *value, const unsigned char len) +int dccp_insert_option(struct sk_buff *skb, const unsigned char option, + const void *value, const unsigned char len) { unsigned char *to; - if (DCCP_SKB_CB(skb)->dccpd_opt_len + len + 2 > DCCP_MAX_OPT_LEN) { - LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to insert " - "%d option!\n", option); - return; - } + if (DCCP_SKB_CB(skb)->dccpd_opt_len + len + 2 > DCCP_MAX_OPT_LEN) + return -1; DCCP_SKB_CB(skb)->dccpd_opt_len += len + 2; @@ -270,14 +304,15 @@ void dccp_insert_option(struct sock *sk, struct sk_buff *skb, *to++ = len + 2; memcpy(to, value, len); + return 0; } EXPORT_SYMBOL_GPL(dccp_insert_option); -static void dccp_insert_option_ndp(struct sock *sk, struct sk_buff *skb) +static int dccp_insert_option_ndp(struct sock *sk, struct sk_buff *skb) { struct dccp_sock *dp = dccp_sk(sk); - int ndp = dp->dccps_ndp_count; + u64 ndp = dp->dccps_ndp_count; if (dccp_non_data_packet(skb)) ++dp->dccps_ndp_count; @@ -290,7 +325,7 @@ static void dccp_insert_option_ndp(struct sock *sk, struct sk_buff *skb) const int len = ndp_len + 2; if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) - return; + return -1; DCCP_SKB_CB(skb)->dccpd_opt_len += len; @@ -299,6 +334,8 @@ static void dccp_insert_option_ndp(struct sock *sk, struct sk_buff *skb) *ptr++ = len; dccp_encode_value_var(ndp, ptr, ndp_len); } + + return 0; } static inline int dccp_elapsed_time_len(const u32 elapsed_time) @@ -306,550 +343,267 @@ static inline int dccp_elapsed_time_len(const u32 elapsed_time) return elapsed_time == 0 ? 0 : elapsed_time <= 0xFFFF ? 2 : 4; } -void dccp_insert_option_elapsed_time(struct sock *sk, - struct sk_buff *skb, - u32 elapsed_time) -{ -#ifdef CONFIG_IP_DCCP_DEBUG - struct dccp_sock *dp = dccp_sk(sk); - const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ? - "CLIENT TX opt: " : "server TX opt: "; -#endif - const int elapsed_time_len = dccp_elapsed_time_len(elapsed_time); - const int len = 2 + elapsed_time_len; - unsigned char *to; - - if (elapsed_time_len == 0) - return; - - if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) { - LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to " - "insert elapsed time!\n"); - return; - } - - DCCP_SKB_CB(skb)->dccpd_opt_len += len; - - to = skb_push(skb, len); - *to++ = DCCPO_ELAPSED_TIME; - *to++ = len; - - if (elapsed_time_len == 2) { - const u16 var16 = htons((u16)elapsed_time); - memcpy(to, &var16, 2); - } else { - const u32 var32 = htonl(elapsed_time); - memcpy(to, &var32, 4); - } - - dccp_pr_debug("%sELAPSED_TIME=%u, len=%d, seqno=%llu\n", - debug_prefix, elapsed_time, - len, - (unsigned long long) DCCP_SKB_CB(skb)->dccpd_seq); -} - -EXPORT_SYMBOL_GPL(dccp_insert_option_elapsed_time); - -static void dccp_insert_option_ack_vector(struct sock *sk, struct sk_buff *skb) +static int dccp_insert_option_timestamp(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; - const u32 elapsed_time = timeval_now_delta(&ap->dccpap_time) / 10; - unsigned char *to, *from; - - 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_insert_option_timestamp(struct sock *sk, struct sk_buff *skb) -{ - struct timeval tv; - u32 now; - - do_gettimeofday(&tv); - now = (tv.tv_sec * USEC_PER_SEC + tv.tv_usec) / 10; + __be32 now = htonl(dccp_timestamp()); /* yes this will overflow but that is the point as we want a * 10 usec 32 bit timer which mean it wraps every 11.9 hours */ - now = htonl(now); - dccp_insert_option(sk, skb, DCCPO_TIMESTAMP, &now, sizeof(now)); + return dccp_insert_option(skb, DCCPO_TIMESTAMP, &now, sizeof(now)); } -EXPORT_SYMBOL_GPL(dccp_insert_option_timestamp); - -static void dccp_insert_option_timestamp_echo(struct sock *sk, - struct sk_buff *skb) +static int dccp_insert_option_timestamp_echo(struct dccp_sock *dp, + struct dccp_request_sock *dreq, + 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 - u32 tstamp_echo; - const u32 elapsed_time = - timeval_now_delta(&dp->dccps_timestamp_time) / 10; - const int elapsed_time_len = dccp_elapsed_time_len(elapsed_time); - const int len = 6 + elapsed_time_len; + __be32 tstamp_echo; unsigned char *to; + u32 elapsed_time, elapsed_time_len, len; - if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) { - LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to insert " - "timestamp echo!\n"); - return; + if (dreq != NULL) { + elapsed_time = dccp_timestamp() - dreq->dreq_timestamp_time; + tstamp_echo = htonl(dreq->dreq_timestamp_echo); + dreq->dreq_timestamp_echo = 0; + } else { + elapsed_time = dccp_timestamp() - dp->dccps_timestamp_time; + tstamp_echo = htonl(dp->dccps_timestamp_echo); + dp->dccps_timestamp_echo = 0; } + elapsed_time_len = dccp_elapsed_time_len(elapsed_time); + len = 6 + elapsed_time_len; + + if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) + return -1; + DCCP_SKB_CB(skb)->dccpd_opt_len += len; to = skb_push(skb, len); *to++ = DCCPO_TIMESTAMP_ECHO; *to++ = len; - tstamp_echo = htonl(dp->dccps_timestamp_echo); memcpy(to, &tstamp_echo, 4); to += 4; - + if (elapsed_time_len == 2) { - const u16 var16 = htons((u16)elapsed_time); + const __be16 var16 = htons((u16)elapsed_time); memcpy(to, &var16, 2); } else if (elapsed_time_len == 4) { - const u32 var32 = htonl(elapsed_time); + const __be32 var32 = htonl(elapsed_time); memcpy(to, &var32, 4); } - dccp_pr_debug("%sTIMESTAMP_ECHO=%u, len=%d, seqno=%llu\n", - debug_prefix, dp->dccps_timestamp_echo, - len, - (unsigned long long) DCCP_SKB_CB(skb)->dccpd_seq); - - dp->dccps_timestamp_echo = 0; - dp->dccps_timestamp_time.tv_sec = 0; - dp->dccps_timestamp_time.tv_usec = 0; + return 0; } -void dccp_insert_options(struct sock *sk, struct sk_buff *skb) +static 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; + struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb); + const u16 buflen = dccp_ackvec_buflen(av); + /* Figure out how many options do we need to represent the ackvec */ + const u8 nr_opts = DIV_ROUND_UP(buflen, DCCP_SINGLE_OPT_MAXLEN); + u16 len = buflen + 2 * nr_opts; + u8 i, nonce = 0; + const unsigned char *tail, *from; + unsigned char *to; - DCCP_SKB_CB(skb)->dccpd_opt_len = 0; + if (dcb->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) { + DCCP_WARN("Lacking space for %u bytes on %s packet\n", len, + dccp_packet_name(dcb->dccpd_type)); + return -1; + } + /* + * Since Ack Vectors are variable-length, we can not always predict + * their size. To catch exception cases where the space is running out + * on the skb, a separate Sync is scheduled to carry the Ack Vector. + */ + if (len > DCCPAV_MIN_OPTLEN && + len + dcb->dccpd_opt_len + skb->len > dp->dccps_mss_cache) { + DCCP_WARN("No space left for Ack Vector (%u) on skb (%u+%u), " + "MPS=%u ==> reduce payload size?\n", len, skb->len, + dcb->dccpd_opt_len, dp->dccps_mss_cache); + dp->dccps_sync_scheduled = 1; + return 0; + } + dcb->dccpd_opt_len += len; - if (dp->dccps_options.dccpo_send_ndp_count) - dccp_insert_option_ndp(sk, skb); + to = skb_push(skb, len); + len = buflen; + from = av->av_buf + av->av_buf_head; + tail = av->av_buf + DCCPAV_MAX_ACKVEC_LEN; - 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); + for (i = 0; i < nr_opts; ++i) { + int copylen = len; - if (dp->dccps_timestamp_echo != 0) - dccp_insert_option_timestamp_echo(sk, skb); - } + if (len > DCCP_SINGLE_OPT_MAXLEN) + copylen = DCCP_SINGLE_OPT_MAXLEN; - ccid_hc_rx_insert_options(dp->dccps_hc_rx_ccid, sk, skb); - ccid_hc_tx_insert_options(dp->dccps_hc_tx_ccid, sk, skb); + /* + * RFC 4340, 12.2: Encode the Nonce Echo for this Ack Vector via + * its type; ack_nonce is the sum of all individual buf_nonce's. + */ + nonce ^= av->av_buf_nonce[i]; - /* XXX: insert other options when appropriate */ + *to++ = DCCPO_ACK_VECTOR_0 + av->av_buf_nonce[i]; + *to++ = copylen + 2; - if (DCCP_SKB_CB(skb)->dccpd_opt_len != 0) { - /* The length of all options has to be a multiple of 4 */ - int padding = DCCP_SKB_CB(skb)->dccpd_opt_len % 4; + /* Check if buf_head wraps */ + if (from + copylen > tail) { + const u16 tailsize = tail - from; - if (padding != 0) { - padding = 4 - padding; - memset(skb_push(skb, padding), 0, padding); - DCCP_SKB_CB(skb)->dccpd_opt_len += padding; + memcpy(to, from, tailsize); + to += tailsize; + len -= tailsize; + copylen -= tailsize; + from = av->av_buf; } - } -} -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); + memcpy(to, from, copylen); + from += copylen; + to += copylen; + len -= copylen; } + /* + * Each sent Ack Vector is recorded in the list, as per A.2 of RFC 4340. + */ + if (dccp_ackvec_update_records(av, dcb->dccpd_seq, nonce)) + return -ENOBUFS; + return 0; } -static inline u8 dccp_ackpkts_state(const struct dccp_ackpkts *ap, - const unsigned int index) +/** + * dccp_insert_option_mandatory - Mandatory option (5.8.2) + * Note that since we are using skb_push, this function needs to be called + * _after_ inserting the option it is supposed to influence (stack order). + */ +int dccp_insert_option_mandatory(struct sk_buff *skb) { - return ap->dccpap_buf[index] & DCCP_ACKPKTS_STATE_MASK; -} + if (DCCP_SKB_CB(skb)->dccpd_opt_len >= DCCP_MAX_OPT_LEN) + return -1; -static inline u8 dccp_ackpkts_len(const struct dccp_ackpkts *ap, - const unsigned int index) -{ - return ap->dccpap_buf[index] & DCCP_ACKPKTS_LEN_MASK; + DCCP_SKB_CB(skb)->dccpd_opt_len++; + *skb_push(skb, 1) = DCCPO_MANDATORY; + return 0; } -/* - * 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. +/** + * dccp_insert_fn_opt - Insert single Feature-Negotiation option into @skb + * @type: %DCCPO_CHANGE_L, %DCCPO_CHANGE_R, %DCCPO_CONFIRM_L, %DCCPO_CONFIRM_R + * @feat: one out of %dccp_feature_numbers + * @val: NN value or SP array (preferred element first) to copy + * @len: true length of @val in bytes (excluding first element repetition) + * @repeat_first: whether to copy the first element of @val twice + * + * The last argument is used to construct Confirm options, where the preferred + * value and the preference list appear separately (RFC 4340, 6.3.1). Preference + * lists are kept such that the preferred entry is always first, so we only need + * to copy twice, and avoid the overhead of cloning into a bigger array. */ -static inline int dccp_ackpkts_set_buf_head_state(struct dccp_ackpkts *ap, - const unsigned int packets, - const unsigned char state) +int dccp_insert_fn_opt(struct sk_buff *skb, u8 type, u8 feat, + u8 *val, u8 len, bool repeat_first) { - unsigned int gap; - signed long new_head; - - if (ap->dccpap_buf_vector_len + packets > ap->dccpap_buf_len) - return -ENOBUFS; + u8 tot_len, *to; - gap = packets - 1; - new_head = ap->dccpap_buf_head - packets; + /* take the `Feature' field and possible repetition into account */ + if (len > (DCCP_SINGLE_OPT_MAXLEN - 2)) { + DCCP_WARN("length %u for feature %u too large\n", len, feat); + return -1; + } - 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; - } + if (unlikely(val == NULL || len == 0)) + len = repeat_first = false; + tot_len = 3 + repeat_first + len; - ap->dccpap_buf_head = new_head; + if (DCCP_SKB_CB(skb)->dccpd_opt_len + tot_len > DCCP_MAX_OPT_LEN) { + DCCP_WARN("packet too small for feature %d option!\n", feat); + return -1; + } + DCCP_SKB_CB(skb)->dccpd_opt_len += tot_len; - if (gap > 0) - memset(ap->dccpap_buf + ap->dccpap_buf_head + 1, - DCCP_ACKPKTS_STATE_NOT_RECEIVED, gap); + to = skb_push(skb, tot_len); + *to++ = type; + *to++ = tot_len; + *to++ = feat; - ap->dccpap_buf[ap->dccpap_buf_head] = state; - ap->dccpap_buf_vector_len += packets; + if (repeat_first) + *to++ = *val; + if (len) + memcpy(to, val, len); return 0; } -/* - * Implements the draft-ietf-dccp-spec-11.txt Appendix A - */ -int dccp_ackpkts_add(struct dccp_ackpkts *ap, u64 ackno, u8 state) +/* The length of all options needs to be a multiple of 4 (5.8) */ +static void dccp_insert_option_padding(struct sk_buff *skb) { - /* - * 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); + int padding = DCCP_SKB_CB(skb)->dccpd_opt_len % 4; - /* - * 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; - } + if (padding != 0) { + padding = 4 - padding; + memset(skb_push(skb, padding), 0, padding); + DCCP_SKB_CB(skb)->dccpd_opt_len += padding; } - - ap->dccpap_buf_ackno = ackno; - do_gettimeofday(&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) +int dccp_insert_options(struct sock *sk, struct sk_buff *skb) { - if (!dccp_debug) - return; + struct dccp_sock *dp = dccp_sk(sk); - printk("ACK vector len=%d, ackno=%llu |", len, - (unsigned long long) ackno); + DCCP_SKB_CB(skb)->dccpd_opt_len = 0; - while (len--) { - const u8 state = (*vector & DCCP_ACKPKTS_STATE_MASK) >> 6; - const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK); + if (dp->dccps_send_ndp_count && dccp_insert_option_ndp(sk, skb)) + return -1; - printk("%d,%d|", state, rl); - ++vector; - } + if (DCCP_SKB_CB(skb)->dccpd_type != DCCP_PKT_DATA) { - printk("\n"); -} + /* Feature Negotiation */ + if (dccp_feat_insert_opts(dp, NULL, skb)) + return -1; -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 + if (DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_REQUEST) { + /* + * Obtain RTT sample from Request/Response exchange. + * This is currently used for TFRC initialisation. + */ + if (dccp_insert_option_timestamp(skb)) + return -1; -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; -} + } else if (dccp_ackvec_pending(sk) && + dccp_insert_option_ackvec(sk, skb)) { + return -1; + } + } -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; + if (dp->dccps_hc_rx_insert_options) { + if (ccid_hc_rx_insert_options(dp->dccps_hc_rx_ccid, sk, skb)) + return -1; + dp->dccps_hc_rx_insert_options = 0; } + + if (dp->dccps_timestamp_echo != 0 && + dccp_insert_option_timestamp_echo(dp, NULL, skb)) + return -1; + + dccp_insert_option_padding(skb); + return 0; } -static void dccp_ackpkts_check_rcv_ackvector(struct dccp_ackpkts *ap, - struct sock *sk, u64 ackno, - const unsigned char len, - const unsigned char *vector) +int dccp_insert_options_rsk(struct dccp_request_sock *dreq, struct sk_buff *skb) { - 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"); */ + DCCP_SKB_CB(skb)->dccpd_opt_len = 0; - i = len; - while (i--) { - const u8 rl = (*vector & DCCP_ACKPKTS_LEN_MASK); - u64 ackno_end_rl; + if (dccp_feat_insert_opts(NULL, dreq, skb)) + return -1; - dccp_set_seqno(&ackno_end_rl, ackno - rl); + /* Obtain RTT sample from Response/Ack exchange (used by TFRC). */ + if (dccp_insert_option_timestamp(skb)) + return -1; - /* - * 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"); */ + if (dreq->dreq_timestamp_echo != 0 && + dccp_insert_option_timestamp_echo(NULL, dreq, skb)) + return -1; - dccp_set_seqno(&ackno, ackno_end_rl - 1); - ++vector; - } + dccp_insert_option_padding(skb); + return 0; } |