| /** | |
| * \addtogroup uip | |
| * @{ | |
| */ | |
| /** | |
| * \file | |
| * The uIP TCP/IP stack code. | |
| * \author Adam Dunkels <adam@dunkels.com> | |
| */ | |
| /* | |
| * Copyright (c) 2001-2003, Adam Dunkels. | |
| * All rights reserved. | |
| * | |
| * Redistribution and use in source and binary forms, with or without | |
| * modification, are permitted provided that the following conditions | |
| * are met: | |
| * 1. Redistributions of source code must retain the above copyright | |
| * notice, this list of conditions and the following disclaimer. | |
| * 2. Redistributions in binary form must reproduce the above copyright | |
| * notice, this list of conditions and the following disclaimer in the | |
| * documentation and/or other materials provided with the distribution. | |
| * 3. The name of the author may not be used to endorse or promote | |
| * products derived from this software without specific prior | |
| * written permission. | |
| * | |
| * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS | |
| * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED | |
| * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | |
| * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY | |
| * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | |
| * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE | |
| * GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS | |
| * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, | |
| * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING | |
| * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS | |
| * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. | |
| * | |
| * This file is part of the uIP TCP/IP stack. | |
| * | |
| * $Id: uip.c,v 1.62.2.10 2003/10/07 13:23:01 adam Exp $ | |
| * | |
| */ | |
| /* | |
| This is a small implementation of the IP and TCP protocols (as well as | |
| some basic ICMP stuff). The implementation couples the IP, TCP and the | |
| application layers very tightly. To keep the size of the compiled code | |
| down, this code also features heavy usage of the goto statement. | |
| The principle is that we have a small buffer, called the uip_buf, in | |
| which the device driver puts an incoming packet. The TCP/IP stack | |
| parses the headers in the packet, and calls upon the application. If | |
| the remote host has sent data to the application, this data is present | |
| in the uip_buf and the application read the data from there. It is up | |
| to the application to put this data into a byte stream if needed. The | |
| application will not be fed with data that is out of sequence. | |
| If the application whishes to send data to the peer, it should put its | |
| data into the uip_buf, 40 bytes from the start of the buffer. The | |
| TCP/IP stack will calculate the checksums, and fill in the necessary | |
| header fields and finally send the packet back to the peer. | |
| */ | |
| #include "uip.h" | |
| #include "uipopt.h" | |
| #include "uip_arch.h" | |
| /*-----------------------------------------------------------------------------------*/ | |
| /* Variable definitions. */ | |
| /* The IP address of this host. If it is defined to be fixed (by setting UIP_FIXEDADDR to 1 in uipopt.h), the address is set here. Otherwise, the address */ | |
| #if UIP_FIXEDADDR > 0 | |
| const u16_t uip_hostaddr[2] = | |
| {HTONS((UIP_IPADDR0 << 8) | UIP_IPADDR1), | |
| HTONS((UIP_IPADDR2 << 8) | UIP_IPADDR3)}; | |
| const u16_t uip_arp_draddr[2] = | |
| {HTONS((UIP_DRIPADDR0 << 8) | UIP_DRIPADDR1), | |
| HTONS((UIP_DRIPADDR2 << 8) | UIP_DRIPADDR3)}; | |
| const u16_t uip_arp_netmask[2] = | |
| {HTONS((UIP_NETMASK0 << 8) | UIP_NETMASK1), | |
| HTONS((UIP_NETMASK2 << 8) | UIP_NETMASK3)}; | |
| #else | |
| u16_t uip_hostaddr[2]; | |
| u16_t uip_arp_draddr[2], uip_arp_netmask[2]; | |
| #endif /* UIP_FIXEDADDR */ | |
| u8_t uip_buf[UIP_BUFSIZE+2]; /* The packet buffer that contains | |
| incoming packets. */ | |
| volatile u8_t *uip_appdata; /* The uip_appdata pointer points to | |
| application data. */ | |
| volatile u8_t *uip_sappdata; /* The uip_appdata pointer points to the | |
| application data which is to be sent. */ | |
| #if UIP_URGDATA > 0 | |
| volatile u8_t *uip_urgdata; /* The uip_urgdata pointer points to | |
| urgent data (out-of-band data), if | |
| present. */ | |
| volatile u8_t uip_urglen, uip_surglen; | |
| #endif /* UIP_URGDATA > 0 */ | |
| volatile u16_t uip_len, uip_slen; | |
| /* The uip_len is either 8 or 16 bits, | |
| depending on the maximum packet | |
| size. */ | |
| volatile u8_t uip_flags; /* The uip_flags variable is used for | |
| communication between the TCP/IP stack | |
| and the application program. */ | |
| struct uip_conn *uip_conn; /* uip_conn always points to the current | |
| connection. */ | |
| struct uip_conn uip_conns[UIP_CONNS]; | |
| /* The uip_conns array holds all TCP | |
| connections. */ | |
| u16_t uip_listenports[UIP_LISTENPORTS]; | |
| /* The uip_listenports list all currently | |
| listning ports. */ | |
| #if UIP_UDP | |
| struct uip_udp_conn *uip_udp_conn; | |
| struct uip_udp_conn uip_udp_conns[UIP_UDP_CONNS]; | |
| #endif /* UIP_UDP */ | |
| static u16_t ipid; /* Ths ipid variable is an increasing | |
| number that is used for the IP ID | |
| field. */ | |
| static u8_t iss[4]; /* The iss variable is used for the TCP | |
| initial sequence number. */ | |
| #if UIP_ACTIVE_OPEN | |
| static u16_t lastport; /* Keeps track of the last port used for | |
| a new connection. */ | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| /* Temporary variables. */ | |
| volatile u8_t uip_acc32[4]; | |
| static u8_t c, opt; | |
| static u16_t tmp16; | |
| /* Structures and definitions. */ | |
| #define TCP_FIN 0x01 | |
| #define TCP_SYN 0x02 | |
| #define TCP_RST 0x04 | |
| #define TCP_PSH 0x08 | |
| #define TCP_ACK 0x10 | |
| #define TCP_URG 0x20 | |
| #define TCP_CTL 0x3f | |
| #define ICMP_ECHO_REPLY 0 | |
| #define ICMP_ECHO 8 | |
| /* Macros. */ | |
| #define BUF ((uip_tcpip_hdr *)&uip_buf[UIP_LLH_LEN]) | |
| #define FBUF ((uip_tcpip_hdr *)&uip_reassbuf[0]) | |
| #define ICMPBUF ((uip_icmpip_hdr *)&uip_buf[UIP_LLH_LEN]) | |
| #define UDPBUF ((uip_udpip_hdr *)&uip_buf[UIP_LLH_LEN]) | |
| #if UIP_STATISTICS == 1 | |
| struct uip_stats uip_stat; | |
| #define UIP_STAT(s) s | |
| #else | |
| #define UIP_STAT(s) | |
| #endif /* UIP_STATISTICS == 1 */ | |
| #if UIP_LOGGING == 1 | |
| #include <stdio.h> | |
| void uip_log(char *msg); | |
| #define UIP_LOG(m) uip_log(m) | |
| #else | |
| #define UIP_LOG(m) | |
| #endif /* UIP_LOGGING == 1 */ | |
| /*-----------------------------------------------------------------------------------*/ | |
| void | |
| uip_init(void) | |
| { | |
| for(c = 0; c < UIP_LISTENPORTS; ++c) { | |
| uip_listenports[c] = 0; | |
| } | |
| for(c = 0; c < UIP_CONNS; ++c) { | |
| uip_conns[c].tcpstateflags = CLOSED; | |
| } | |
| #if UIP_ACTIVE_OPEN | |
| lastport = 1024; | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| #if UIP_UDP | |
| for(c = 0; c < UIP_UDP_CONNS; ++c) { | |
| uip_udp_conns[c].lport = 0; | |
| } | |
| #endif /* UIP_UDP */ | |
| /* IPv4 initialization. */ | |
| #if UIP_FIXEDADDR == 0 | |
| uip_hostaddr[0] = uip_hostaddr[1] = 0; | |
| #endif /* UIP_FIXEDADDR */ | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| #if UIP_ACTIVE_OPEN | |
| struct uip_conn * | |
| uip_connect(u16_t *ripaddr, u16_t rport) | |
| { | |
| register struct uip_conn *conn, *cconn; | |
| /* Find an unused local port. */ | |
| again: | |
| ++lastport; | |
| if(lastport >= 32000) { | |
| lastport = 4096; | |
| } | |
| /* Check if this port is already in use, and if so try to find | |
| another one. */ | |
| for(c = 0; c < UIP_CONNS; ++c) { | |
| conn = &uip_conns[c]; | |
| if(conn->tcpstateflags != CLOSED && | |
| conn->lport == htons(lastport)) { | |
| goto again; | |
| } | |
| } | |
| conn = 0; | |
| for(c = 0; c < UIP_CONNS; ++c) { | |
| cconn = &uip_conns[c]; | |
| if(cconn->tcpstateflags == CLOSED) { | |
| conn = cconn; | |
| break; | |
| } | |
| if(cconn->tcpstateflags == TIME_WAIT) { | |
| if(conn == 0 || | |
| cconn->timer > uip_conn->timer) { | |
| conn = cconn; | |
| } | |
| } | |
| } | |
| if(conn == 0) { | |
| return 0; | |
| } | |
| conn->tcpstateflags = SYN_SENT; | |
| conn->snd_nxt[0] = iss[0]; | |
| conn->snd_nxt[1] = iss[1]; | |
| conn->snd_nxt[2] = iss[2]; | |
| conn->snd_nxt[3] = iss[3]; | |
| conn->initialmss = conn->mss = UIP_TCP_MSS; | |
| conn->len = 1; /* TCP length of the SYN is one. */ | |
| conn->nrtx = 0; | |
| conn->timer = 1; /* Send the SYN next time around. */ | |
| conn->rto = UIP_RTO; | |
| conn->sa = 0; | |
| conn->sv = 16; | |
| conn->lport = htons(lastport); | |
| conn->rport = rport; | |
| conn->ripaddr[0] = ripaddr[0]; | |
| conn->ripaddr[1] = ripaddr[1]; | |
| return conn; | |
| } | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| /*-----------------------------------------------------------------------------------*/ | |
| #if UIP_UDP | |
| struct uip_udp_conn * | |
| uip_udp_new(u16_t *ripaddr, u16_t rport) | |
| { | |
| register struct uip_udp_conn *conn; | |
| /* Find an unused local port. */ | |
| again: | |
| ++lastport; | |
| if(lastport >= 32000) { | |
| lastport = 4096; | |
| } | |
| for(c = 0; c < UIP_UDP_CONNS; ++c) { | |
| if(uip_udp_conns[c].lport == lastport) { | |
| goto again; | |
| } | |
| } | |
| conn = 0; | |
| for(c = 0; c < UIP_UDP_CONNS; ++c) { | |
| if(uip_udp_conns[c].lport == 0) { | |
| conn = &uip_udp_conns[c]; | |
| break; | |
| } | |
| } | |
| if(conn == 0) { | |
| return 0; | |
| } | |
| conn->lport = HTONS(lastport); | |
| conn->rport = HTONS(rport); | |
| conn->ripaddr[0] = ripaddr[0]; | |
| conn->ripaddr[1] = ripaddr[1]; | |
| return conn; | |
| } | |
| #endif /* UIP_UDP */ | |
| /*-----------------------------------------------------------------------------------*/ | |
| void | |
| uip_unlisten(u16_t port) | |
| { | |
| for(c = 0; c < UIP_LISTENPORTS; ++c) { | |
| if(uip_listenports[c] == port) { | |
| uip_listenports[c] = 0; | |
| return; | |
| } | |
| } | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| void | |
| uip_listen(u16_t port) | |
| { | |
| for(c = 0; c < UIP_LISTENPORTS; ++c) { | |
| if(uip_listenports[c] == 0) { | |
| uip_listenports[c] = port; | |
| return; | |
| } | |
| } | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| /* XXX: IP fragment reassembly: not well-tested. */ | |
| #if UIP_REASSEMBLY | |
| #define UIP_REASS_BUFSIZE (UIP_BUFSIZE - UIP_LLH_LEN) | |
| static u8_t uip_reassbuf[UIP_REASS_BUFSIZE]; | |
| static u8_t uip_reassbitmap[UIP_REASS_BUFSIZE / (8 * 8)]; | |
| static const u8_t bitmap_bits[8] = {0xff, 0x7f, 0x3f, 0x1f, | |
| 0x0f, 0x07, 0x03, 0x01}; | |
| static u16_t uip_reasslen; | |
| static u8_t uip_reassflags; | |
| #define UIP_REASS_FLAG_LASTFRAG 0x01 | |
| static u8_t uip_reasstmr; | |
| #define IP_HLEN 20 | |
| #define IP_MF 0x20 | |
| static u8_t | |
| uip_reass(void) | |
| { | |
| u16_t offset, len; | |
| u16_t i; | |
| /* If ip_reasstmr is zero, no packet is present in the buffer, so we | |
| write the IP header of the fragment into the reassembly | |
| buffer. The timer is updated with the maximum age. */ | |
| if(uip_reasstmr == 0) { | |
| memcpy(uip_reassbuf, &BUF->vhl, IP_HLEN); | |
| uip_reasstmr = UIP_REASS_MAXAGE; | |
| uip_reassflags = 0; | |
| /* Clear the bitmap. */ | |
| memset(uip_reassbitmap, sizeof(uip_reassbitmap), 0); | |
| } | |
| /* Check if the incoming fragment matches the one currently present | |
| in the reasembly buffer. If so, we proceed with copying the | |
| fragment into the buffer. */ | |
| if(BUF->srcipaddr[0] == FBUF->srcipaddr[0] && | |
| BUF->srcipaddr[1] == FBUF->srcipaddr[1] && | |
| BUF->destipaddr[0] == FBUF->destipaddr[0] && | |
| BUF->destipaddr[1] == FBUF->destipaddr[1] && | |
| BUF->ipid[0] == FBUF->ipid[0] && | |
| BUF->ipid[1] == FBUF->ipid[1]) { | |
| len = (BUF->len[0] << 8) + BUF->len[1] - (BUF->vhl & 0x0f) * 4; | |
| offset = (((BUF->ipoffset[0] & 0x3f) << 8) + BUF->ipoffset[1]) * 8; | |
| /* If the offset or the offset + fragment length overflows the | |
| reassembly buffer, we discard the entire packet. */ | |
| if(offset > UIP_REASS_BUFSIZE || | |
| offset + len > UIP_REASS_BUFSIZE) { | |
| uip_reasstmr = 0; | |
| goto nullreturn; | |
| } | |
| /* Copy the fragment into the reassembly buffer, at the right | |
| offset. */ | |
| memcpy(&uip_reassbuf[IP_HLEN + offset], | |
| (char *)BUF + (int)((BUF->vhl & 0x0f) * 4), | |
| len); | |
| /* Update the bitmap. */ | |
| if(offset / (8 * 8) == (offset + len) / (8 * 8)) { | |
| /* If the two endpoints are in the same byte, we only update | |
| that byte. */ | |
| uip_reassbitmap[offset / (8 * 8)] |= | |
| bitmap_bits[(offset / 8 ) & 7] & | |
| ~bitmap_bits[((offset + len) / 8 ) & 7]; | |
| } else { | |
| /* If the two endpoints are in different bytes, we update the | |
| bytes in the endpoints and fill the stuff inbetween with | |
| 0xff. */ | |
| uip_reassbitmap[offset / (8 * 8)] |= | |
| bitmap_bits[(offset / 8 ) & 7]; | |
| for(i = 1 + offset / (8 * 8); i < (offset + len) / (8 * 8); ++i) { | |
| uip_reassbitmap[i] = 0xff; | |
| } | |
| uip_reassbitmap[(offset + len) / (8 * 8)] |= | |
| ~bitmap_bits[((offset + len) / 8 ) & 7]; | |
| } | |
| /* If this fragment has the More Fragments flag set to zero, we | |
| know that this is the last fragment, so we can calculate the | |
| size of the entire packet. We also set the | |
| IP_REASS_FLAG_LASTFRAG flag to indicate that we have received | |
| the final fragment. */ | |
| if((BUF->ipoffset[0] & IP_MF) == 0) { | |
| uip_reassflags |= UIP_REASS_FLAG_LASTFRAG; | |
| uip_reasslen = offset + len; | |
| } | |
| /* Finally, we check if we have a full packet in the buffer. We do | |
| this by checking if we have the last fragment and if all bits | |
| in the bitmap are set. */ | |
| if(uip_reassflags & UIP_REASS_FLAG_LASTFRAG) { | |
| /* Check all bytes up to and including all but the last byte in | |
| the bitmap. */ | |
| for(i = 0; i < uip_reasslen / (8 * 8) - 1; ++i) { | |
| if(uip_reassbitmap[i] != 0xff) { | |
| goto nullreturn; | |
| } | |
| } | |
| /* Check the last byte in the bitmap. It should contain just the | |
| right amount of bits. */ | |
| if(uip_reassbitmap[uip_reasslen / (8 * 8)] != | |
| (u8_t)~bitmap_bits[uip_reasslen / 8 & 7]) { | |
| goto nullreturn; | |
| } | |
| /* If we have come this far, we have a full packet in the | |
| buffer, so we allocate a pbuf and copy the packet into it. We | |
| also reset the timer. */ | |
| uip_reasstmr = 0; | |
| memcpy(BUF, FBUF, uip_reasslen); | |
| /* Pretend to be a "normal" (i.e., not fragmented) IP packet | |
| from now on. */ | |
| BUF->ipoffset[0] = BUF->ipoffset[1] = 0; | |
| BUF->len[0] = uip_reasslen >> 8; | |
| BUF->len[1] = uip_reasslen & 0xff; | |
| BUF->ipchksum = 0; | |
| BUF->ipchksum = ~(uip_ipchksum()); | |
| return uip_reasslen; | |
| } | |
| } | |
| nullreturn: | |
| return 0; | |
| } | |
| #endif /* UIP_REASSEMBL */ | |
| /*-----------------------------------------------------------------------------------*/ | |
| static void | |
| uip_add_rcv_nxt(u16_t n) | |
| { | |
| uip_add32(uip_conn->rcv_nxt, n); | |
| uip_conn->rcv_nxt[0] = uip_acc32[0]; | |
| uip_conn->rcv_nxt[1] = uip_acc32[1]; | |
| uip_conn->rcv_nxt[2] = uip_acc32[2]; | |
| uip_conn->rcv_nxt[3] = uip_acc32[3]; | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| void | |
| uip_process(u8_t flag) | |
| { | |
| register struct uip_conn *uip_connr = uip_conn; | |
| uip_appdata = &uip_buf[40 + UIP_LLH_LEN]; | |
| /* Check if we were invoked because of the perodic timer fireing. */ | |
| if(flag == UIP_TIMER) { | |
| #if UIP_REASSEMBLY | |
| if(uip_reasstmr != 0) { | |
| --uip_reasstmr; | |
| } | |
| #endif /* UIP_REASSEMBLY */ | |
| /* Increase the initial sequence number. */ | |
| if(++iss[3] == 0) { | |
| if(++iss[2] == 0) { | |
| if(++iss[1] == 0) { | |
| ++iss[0]; | |
| } | |
| } | |
| } | |
| uip_len = 0; | |
| if(uip_connr->tcpstateflags == TIME_WAIT || | |
| uip_connr->tcpstateflags == FIN_WAIT_2) { | |
| ++(uip_connr->timer); | |
| if(uip_connr->timer == UIP_TIME_WAIT_TIMEOUT) { | |
| uip_connr->tcpstateflags = CLOSED; | |
| } | |
| } else if(uip_connr->tcpstateflags != CLOSED) { | |
| /* If the connection has outstanding data, we increase the | |
| connection's timer and see if it has reached the RTO value | |
| in which case we retransmit. */ | |
| if(uip_outstanding(uip_connr)) { | |
| if(uip_connr->timer-- == 0) { | |
| if(uip_connr->nrtx == UIP_MAXRTX || | |
| ((uip_connr->tcpstateflags == SYN_SENT || | |
| uip_connr->tcpstateflags == SYN_RCVD) && | |
| uip_connr->nrtx == UIP_MAXSYNRTX)) { | |
| uip_connr->tcpstateflags = CLOSED; | |
| /* We call UIP_APPCALL() with uip_flags set to | |
| UIP_TIMEDOUT to inform the application that the | |
| connection has timed out. */ | |
| uip_flags = UIP_TIMEDOUT; | |
| UIP_APPCALL(); | |
| /* We also send a reset packet to the remote host. */ | |
| BUF->flags = TCP_RST | TCP_ACK; | |
| goto tcp_send_nodata; | |
| } | |
| /* Exponential backoff. */ | |
| uip_connr->timer = UIP_RTO << (uip_connr->nrtx > 4? | |
| 4: | |
| uip_connr->nrtx); | |
| ++(uip_connr->nrtx); | |
| /* Ok, so we need to retransmit. We do this differently | |
| depending on which state we are in. In ESTABLISHED, we | |
| call upon the application so that it may prepare the | |
| data for the retransmit. In SYN_RCVD, we resend the | |
| SYNACK that we sent earlier and in LAST_ACK we have to | |
| retransmit our FINACK. */ | |
| UIP_STAT(++uip_stat.tcp.rexmit); | |
| switch(uip_connr->tcpstateflags & TS_MASK) { | |
| case SYN_RCVD: | |
| /* In the SYN_RCVD state, we should retransmit our | |
| SYNACK. */ | |
| goto tcp_send_synack; | |
| #if UIP_ACTIVE_OPEN | |
| case SYN_SENT: | |
| /* In the SYN_SENT state, we retransmit out SYN. */ | |
| BUF->flags = 0; | |
| goto tcp_send_syn; | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| case ESTABLISHED: | |
| /* In the ESTABLISHED state, we call upon the application | |
| to do the actual retransmit after which we jump into | |
| the code for sending out the packet (the apprexmit | |
| label). */ | |
| uip_len = 0; | |
| uip_slen = 0; | |
| uip_flags = UIP_REXMIT; | |
| UIP_APPCALL(); | |
| goto apprexmit; | |
| case FIN_WAIT_1: | |
| case CLOSING: | |
| case LAST_ACK: | |
| /* In all these states we should retransmit a FINACK. */ | |
| goto tcp_send_finack; | |
| } | |
| } | |
| } else if((uip_connr->tcpstateflags & TS_MASK) == ESTABLISHED) { | |
| /* If there was no need for a retransmission, we poll the | |
| application for new data. */ | |
| uip_len = 0; | |
| uip_slen = 0; | |
| uip_flags = UIP_POLL; | |
| UIP_APPCALL(); | |
| goto appsend; | |
| } | |
| } | |
| goto drop; | |
| } | |
| #if UIP_UDP | |
| if(flag == UIP_UDP_TIMER) { | |
| if(uip_udp_conn->lport != 0) { | |
| uip_appdata = &uip_buf[UIP_LLH_LEN + 28]; | |
| uip_len = uip_slen = 0; | |
| uip_flags = UIP_POLL; | |
| UIP_UDP_APPCALL(); | |
| goto udp_send; | |
| } else { | |
| goto drop; | |
| } | |
| } | |
| #endif | |
| /* This is where the input processing starts. */ | |
| UIP_STAT(++uip_stat.ip.recv); | |
| /* Start of IPv4 input header processing code. */ | |
| /* Check validity of the IP header. */ | |
| if(BUF->vhl != 0x45) { /* IP version and header length. */ | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_STAT(++uip_stat.ip.vhlerr); | |
| UIP_LOG("ip: invalid version or header length."); | |
| goto drop; | |
| } | |
| /* Check the size of the packet. If the size reported to us in | |
| uip_len doesn't match the size reported in the IP header, there | |
| has been a transmission error and we drop the packet. */ | |
| if(BUF->len[0] != (uip_len >> 8)) { /* IP length, high byte. */ | |
| uip_len = (uip_len & 0xff) | (BUF->len[0] << 8); | |
| } | |
| if(BUF->len[1] != (uip_len & 0xff)) { /* IP length, low byte. */ | |
| uip_len = (uip_len & 0xff00) | BUF->len[1]; | |
| } | |
| /* Check the fragment flag. */ | |
| if((BUF->ipoffset[0] & 0x3f) != 0 || | |
| BUF->ipoffset[1] != 0) { | |
| #if UIP_REASSEMBLY | |
| uip_len = uip_reass(); | |
| if(uip_len == 0) { | |
| goto drop; | |
| } | |
| #else | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_STAT(++uip_stat.ip.fragerr); | |
| UIP_LOG("ip: fragment dropped."); | |
| goto drop; | |
| #endif /* UIP_REASSEMBLY */ | |
| } | |
| /* If we are configured to use ping IP address configuration and | |
| hasn't been assigned an IP address yet, we accept all ICMP | |
| packets. */ | |
| #if UIP_PINGADDRCONF | |
| if((uip_hostaddr[0] | uip_hostaddr[1]) == 0) { | |
| if(BUF->proto == UIP_PROTO_ICMP) { | |
| UIP_LOG("ip: possible ping config packet received."); | |
| goto icmp_input; | |
| } else { | |
| UIP_LOG("ip: packet dropped since no address assigned."); | |
| goto drop; | |
| } | |
| } | |
| #endif /* UIP_PINGADDRCONF */ | |
| /* Check if the packet is destined for our IP address. */ | |
| if(BUF->destipaddr[0] != uip_hostaddr[0]) { | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_LOG("ip: packet not for us."); | |
| goto drop; | |
| } | |
| if(BUF->destipaddr[1] != uip_hostaddr[1]) { | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_LOG("ip: packet not for us."); | |
| goto drop; | |
| } | |
| #if 0 | |
| // IP checksum is wrong through Netgear DSL router | |
| if (uip_ipchksum() != 0xffff) { /* Compute and check the IP header | |
| checksum. */ | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_STAT(++uip_stat.ip.chkerr); | |
| UIP_LOG("ip: bad checksum."); | |
| goto drop; | |
| } | |
| #endif | |
| if(BUF->proto == UIP_PROTO_TCP) /* Check for TCP packet. If so, jump | |
| to the tcp_input label. */ | |
| goto tcp_input; | |
| #if UIP_UDP | |
| if(BUF->proto == UIP_PROTO_UDP) | |
| goto udp_input; | |
| #endif /* UIP_UDP */ | |
| if(BUF->proto != UIP_PROTO_ICMP) { /* We only allow ICMP packets from | |
| here. */ | |
| UIP_STAT(++uip_stat.ip.drop); | |
| UIP_STAT(++uip_stat.ip.protoerr); | |
| UIP_LOG("ip: neither tcp nor icmp."); | |
| goto drop; | |
| } | |
| #if UIP_PINGADDRCONF | |
| icmp_input: | |
| #endif | |
| UIP_STAT(++uip_stat.icmp.recv); | |
| /* ICMP echo (i.e., ping) processing. This is simple, we only change | |
| the ICMP type from ECHO to ECHO_REPLY and adjust the ICMP | |
| checksum before we return the packet. */ | |
| if(ICMPBUF->type != ICMP_ECHO) { | |
| UIP_STAT(++uip_stat.icmp.drop); | |
| UIP_STAT(++uip_stat.icmp.typeerr); | |
| UIP_LOG("icmp: not icmp echo."); | |
| goto drop; | |
| } | |
| /* If we are configured to use ping IP address assignment, we use | |
| the destination IP address of this ping packet and assign it to | |
| ourself. */ | |
| #if UIP_PINGADDRCONF | |
| if((uip_hostaddr[0] | uip_hostaddr[1]) == 0) { | |
| uip_hostaddr[0] = BUF->destipaddr[0]; | |
| uip_hostaddr[1] = BUF->destipaddr[1]; | |
| } | |
| #endif /* UIP_PINGADDRCONF */ | |
| ICMPBUF->type = ICMP_ECHO_REPLY; | |
| if(ICMPBUF->icmpchksum >= HTONS(0xffff - (ICMP_ECHO << 8))) { | |
| ICMPBUF->icmpchksum += HTONS(ICMP_ECHO << 8) + 1; | |
| } else { | |
| ICMPBUF->icmpchksum += HTONS(ICMP_ECHO << 8); | |
| } | |
| /* Swap IP addresses. */ | |
| tmp16 = BUF->destipaddr[0]; | |
| BUF->destipaddr[0] = BUF->srcipaddr[0]; | |
| BUF->srcipaddr[0] = tmp16; | |
| tmp16 = BUF->destipaddr[1]; | |
| BUF->destipaddr[1] = BUF->srcipaddr[1]; | |
| BUF->srcipaddr[1] = tmp16; | |
| UIP_STAT(++uip_stat.icmp.sent); | |
| goto send; | |
| /* End of IPv4 input header processing code. */ | |
| #if UIP_UDP | |
| /* UDP input processing. */ | |
| udp_input: | |
| /* UDP processing is really just a hack. We don't do anything to the | |
| UDP/IP headers, but let the UDP application do all the hard | |
| work. If the application sets uip_slen, it has a packet to | |
| send. */ | |
| #if UIP_UDP_CHECKSUMS | |
| if(uip_udpchksum() != 0xffff) { | |
| UIP_STAT(++uip_stat.udp.drop); | |
| UIP_STAT(++uip_stat.udp.chkerr); | |
| UIP_LOG("udp: bad checksum."); | |
| goto drop; | |
| } | |
| #endif /* UIP_UDP_CHECKSUMS */ | |
| /* Demultiplex this UDP packet between the UDP "connections". */ | |
| for(uip_udp_conn = &uip_udp_conns[0]; | |
| uip_udp_conn < &uip_udp_conns[UIP_UDP_CONNS]; | |
| ++uip_udp_conn) { | |
| if(uip_udp_conn->lport != 0 && | |
| UDPBUF->destport == uip_udp_conn->lport && | |
| (uip_udp_conn->rport == 0 || | |
| UDPBUF->srcport == uip_udp_conn->rport) && | |
| BUF->srcipaddr[0] == uip_udp_conn->ripaddr[0] && | |
| BUF->srcipaddr[1] == uip_udp_conn->ripaddr[1]) { | |
| goto udp_found; | |
| } | |
| } | |
| goto drop; | |
| udp_found: | |
| uip_len = uip_len - 28; | |
| uip_appdata = &uip_buf[UIP_LLH_LEN + 28]; | |
| uip_flags = UIP_NEWDATA; | |
| uip_slen = 0; | |
| UIP_UDP_APPCALL(); | |
| udp_send: | |
| if(uip_slen == 0) { | |
| goto drop; | |
| } | |
| uip_len = uip_slen + 28; | |
| BUF->len[0] = (uip_len >> 8); | |
| BUF->len[1] = (uip_len & 0xff); | |
| BUF->proto = UIP_PROTO_UDP; | |
| UDPBUF->udplen = HTONS(uip_slen + 8); | |
| UDPBUF->udpchksum = 0; | |
| #if UIP_UDP_CHECKSUMS | |
| /* Calculate UDP checksum. */ | |
| UDPBUF->udpchksum = ~(uip_udpchksum()); | |
| if(UDPBUF->udpchksum == 0) { | |
| UDPBUF->udpchksum = 0xffff; | |
| } | |
| #endif /* UIP_UDP_CHECKSUMS */ | |
| BUF->srcport = uip_udp_conn->lport; | |
| BUF->destport = uip_udp_conn->rport; | |
| BUF->srcipaddr[0] = uip_hostaddr[0]; | |
| BUF->srcipaddr[1] = uip_hostaddr[1]; | |
| BUF->destipaddr[0] = uip_udp_conn->ripaddr[0]; | |
| BUF->destipaddr[1] = uip_udp_conn->ripaddr[1]; | |
| uip_appdata = &uip_buf[UIP_LLH_LEN + 40]; | |
| goto ip_send_nolen; | |
| #endif /* UIP_UDP */ | |
| /* TCP input processing. */ | |
| tcp_input: | |
| UIP_STAT(++uip_stat.tcp.recv); | |
| /* Start of TCP input header processing code. */ | |
| #if 1 // FIXME | |
| if(uip_tcpchksum() != 0xffff) { /* Compute and check the TCP | |
| checksum. */ | |
| UIP_STAT(++uip_stat.tcp.drop); | |
| UIP_STAT(++uip_stat.tcp.chkerr); | |
| UIP_LOG("tcp: bad checksum."); | |
| goto drop; | |
| } | |
| #endif | |
| /* Demultiplex this segment. */ | |
| /* First check any active connections. */ | |
| for(uip_connr = &uip_conns[0]; uip_connr < &uip_conns[UIP_CONNS]; ++uip_connr) { | |
| if(uip_connr->tcpstateflags != CLOSED && | |
| BUF->destport == uip_connr->lport && | |
| BUF->srcport == uip_connr->rport && | |
| BUF->srcipaddr[0] == uip_connr->ripaddr[0] && | |
| BUF->srcipaddr[1] == uip_connr->ripaddr[1]) { | |
| goto found; | |
| } | |
| } | |
| /* If we didn't find and active connection that expected the packet, | |
| either this packet is an old duplicate, or this is a SYN packet | |
| destined for a connection in LISTEN. If the SYN flag isn't set, | |
| it is an old packet and we send a RST. */ | |
| if((BUF->flags & TCP_CTL) != TCP_SYN) | |
| goto reset; | |
| tmp16 = BUF->destport; | |
| /* Next, check listening connections. */ | |
| for(c = 0; c < UIP_LISTENPORTS; ++c) { | |
| if(tmp16 == uip_listenports[c]) | |
| goto found_listen; | |
| } | |
| /* No matching connection found, so we send a RST packet. */ | |
| UIP_STAT(++uip_stat.tcp.synrst); | |
| reset: | |
| /* We do not send resets in response to resets. */ | |
| if(BUF->flags & TCP_RST) | |
| goto drop; | |
| UIP_STAT(++uip_stat.tcp.rst); | |
| BUF->flags = TCP_RST | TCP_ACK; | |
| uip_len = 40; | |
| BUF->tcpoffset = 5 << 4; | |
| /* Flip the seqno and ackno fields in the TCP header. */ | |
| c = BUF->seqno[3]; | |
| BUF->seqno[3] = BUF->ackno[3]; | |
| BUF->ackno[3] = c; | |
| c = BUF->seqno[2]; | |
| BUF->seqno[2] = BUF->ackno[2]; | |
| BUF->ackno[2] = c; | |
| c = BUF->seqno[1]; | |
| BUF->seqno[1] = BUF->ackno[1]; | |
| BUF->ackno[1] = c; | |
| c = BUF->seqno[0]; | |
| BUF->seqno[0] = BUF->ackno[0]; | |
| BUF->ackno[0] = c; | |
| /* We also have to increase the sequence number we are | |
| acknowledging. If the least significant byte overflowed, we need | |
| to propagate the carry to the other bytes as well. */ | |
| if(++BUF->ackno[3] == 0) { | |
| if(++BUF->ackno[2] == 0) { | |
| if(++BUF->ackno[1] == 0) { | |
| ++BUF->ackno[0]; | |
| } | |
| } | |
| } | |
| /* Swap port numbers. */ | |
| tmp16 = BUF->srcport; | |
| BUF->srcport = BUF->destport; | |
| BUF->destport = tmp16; | |
| /* Swap IP addresses. */ | |
| tmp16 = BUF->destipaddr[0]; | |
| BUF->destipaddr[0] = BUF->srcipaddr[0]; | |
| BUF->srcipaddr[0] = tmp16; | |
| tmp16 = BUF->destipaddr[1]; | |
| BUF->destipaddr[1] = BUF->srcipaddr[1]; | |
| BUF->srcipaddr[1] = tmp16; | |
| /* And send out the RST packet! */ | |
| goto tcp_send_noconn; | |
| /* This label will be jumped to if we matched the incoming packet | |
| with a connection in LISTEN. In that case, we should create a new | |
| connection and send a SYNACK in return. */ | |
| found_listen: | |
| /* First we check if there are any connections avaliable. Unused | |
| connections are kept in the same table as used connections, but | |
| unused ones have the tcpstate set to CLOSED. Also, connections in | |
| TIME_WAIT are kept track of and we'll use the oldest one if no | |
| CLOSED connections are found. Thanks to Eddie C. Dost for a very | |
| nice algorithm for the TIME_WAIT search. */ | |
| uip_connr = 0; | |
| for(c = 0; c < UIP_CONNS; ++c) { | |
| if(uip_conns[c].tcpstateflags == CLOSED) { | |
| uip_connr = &uip_conns[c]; | |
| break; | |
| } | |
| if(uip_conns[c].tcpstateflags == TIME_WAIT) { | |
| if(uip_connr == 0 || | |
| uip_conns[c].timer > uip_connr->timer) { | |
| uip_connr = &uip_conns[c]; | |
| } | |
| } | |
| } | |
| if(uip_connr == 0) { | |
| /* All connections are used already, we drop packet and hope that | |
| the remote end will retransmit the packet at a time when we | |
| have more spare connections. */ | |
| UIP_STAT(++uip_stat.tcp.syndrop); | |
| UIP_LOG("tcp: found no unused connections."); | |
| goto drop; | |
| } | |
| uip_conn = uip_connr; | |
| /* Fill in the necessary fields for the new connection. */ | |
| uip_connr->rto = uip_connr->timer = UIP_RTO; | |
| uip_connr->sa = 0; | |
| uip_connr->sv = 4; | |
| uip_connr->nrtx = 0; | |
| uip_connr->lport = BUF->destport; | |
| uip_connr->rport = BUF->srcport; | |
| uip_connr->ripaddr[0] = BUF->srcipaddr[0]; | |
| uip_connr->ripaddr[1] = BUF->srcipaddr[1]; | |
| uip_connr->tcpstateflags = SYN_RCVD; | |
| uip_connr->snd_nxt[0] = iss[0]; | |
| uip_connr->snd_nxt[1] = iss[1]; | |
| uip_connr->snd_nxt[2] = iss[2]; | |
| uip_connr->snd_nxt[3] = iss[3]; | |
| uip_connr->len = 1; | |
| /* rcv_nxt should be the seqno from the incoming packet + 1. */ | |
| uip_connr->rcv_nxt[3] = BUF->seqno[3]; | |
| uip_connr->rcv_nxt[2] = BUF->seqno[2]; | |
| uip_connr->rcv_nxt[1] = BUF->seqno[1]; | |
| uip_connr->rcv_nxt[0] = BUF->seqno[0]; | |
| uip_add_rcv_nxt(1); | |
| /* Parse the TCP MSS option, if present. */ | |
| if((BUF->tcpoffset & 0xf0) > 0x50) { | |
| for(c = 0; c < ((BUF->tcpoffset >> 4) - 5) << 2 ;) { | |
| opt = uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + c]; | |
| if(opt == 0x00) { | |
| /* End of options. */ | |
| break; | |
| } else if(opt == 0x01) { | |
| ++c; | |
| /* NOP option. */ | |
| } else if(opt == 0x02 && | |
| uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c] == 0x04) { | |
| /* An MSS option with the right option length. */ | |
| tmp16 = ((u16_t)uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 2 + c] << 8) | | |
| (u16_t)uip_buf[40 + UIP_LLH_LEN + 3 + c]; | |
| uip_connr->initialmss = uip_connr->mss = | |
| tmp16 > UIP_TCP_MSS? UIP_TCP_MSS: tmp16; | |
| /* And we are done processing options. */ | |
| break; | |
| } else { | |
| /* All other options have a length field, so that we easily | |
| can skip past them. */ | |
| if(uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c] == 0) { | |
| /* If the length field is zero, the options are malformed | |
| and we don't process them further. */ | |
| break; | |
| } | |
| c += uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c]; | |
| } | |
| } | |
| } | |
| /* Our response will be a SYNACK. */ | |
| #if UIP_ACTIVE_OPEN | |
| tcp_send_synack: | |
| BUF->flags = TCP_ACK; | |
| tcp_send_syn: | |
| BUF->flags |= TCP_SYN; | |
| #else /* UIP_ACTIVE_OPEN */ | |
| tcp_send_synack: | |
| BUF->flags = TCP_SYN | TCP_ACK; | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| /* We send out the TCP Maximum Segment Size option with our | |
| SYNACK. */ | |
| BUF->optdata[0] = 2; | |
| BUF->optdata[1] = 4; | |
| BUF->optdata[2] = (UIP_TCP_MSS) / 256; | |
| BUF->optdata[3] = (UIP_TCP_MSS) & 255; | |
| uip_len = 44; | |
| BUF->tcpoffset = 6 << 4; | |
| goto tcp_send; | |
| /* This label will be jumped to if we found an active connection. */ | |
| found: | |
| uip_conn = uip_connr; | |
| uip_flags = 0; | |
| /* We do a very naive form of TCP reset processing; we just accept | |
| any RST and kill our connection. We should in fact check if the | |
| sequence number of this reset is wihtin our advertised window | |
| before we accept the reset. */ | |
| if(BUF->flags & TCP_RST) { | |
| uip_connr->tcpstateflags = CLOSED; | |
| UIP_LOG("tcp: got reset, aborting connection."); | |
| uip_flags = UIP_ABORT; | |
| UIP_APPCALL(); | |
| goto drop; | |
| } | |
| /* Calculated the length of the data, if the application has sent | |
| any data to us. */ | |
| c = (BUF->tcpoffset >> 4) << 2; | |
| /* uip_len will contain the length of the actual TCP data. This is | |
| calculated by subtracing the length of the TCP header (in | |
| c) and the length of the IP header (20 bytes). */ | |
| uip_len = uip_len - c - 20; | |
| /* First, check if the sequence number of the incoming packet is | |
| what we're expecting next. If not, we send out an ACK with the | |
| correct numbers in. */ | |
| if(uip_len > 0 && | |
| (BUF->seqno[0] != uip_connr->rcv_nxt[0] || | |
| BUF->seqno[1] != uip_connr->rcv_nxt[1] || | |
| BUF->seqno[2] != uip_connr->rcv_nxt[2] || | |
| BUF->seqno[3] != uip_connr->rcv_nxt[3])) { | |
| goto tcp_send_ack; | |
| } | |
| /* Next, check if the incoming segment acknowledges any outstanding | |
| data. If so, we update the sequence number, reset the length of | |
| the outstanding data, calculate RTT estimations, and reset the | |
| retransmission timer. */ | |
| if((BUF->flags & TCP_ACK) && uip_outstanding(uip_connr)) { | |
| uip_add32(uip_connr->snd_nxt, uip_connr->len); | |
| if(BUF->ackno[0] == uip_acc32[0] && | |
| BUF->ackno[1] == uip_acc32[1] && | |
| BUF->ackno[2] == uip_acc32[2] && | |
| BUF->ackno[3] == uip_acc32[3]) { | |
| /* Update sequence number. */ | |
| uip_connr->snd_nxt[0] = uip_acc32[0]; | |
| uip_connr->snd_nxt[1] = uip_acc32[1]; | |
| uip_connr->snd_nxt[2] = uip_acc32[2]; | |
| uip_connr->snd_nxt[3] = uip_acc32[3]; | |
| /* Do RTT estimation, unless we have done retransmissions. */ | |
| if(uip_connr->nrtx == 0) { | |
| signed char m; | |
| m = uip_connr->rto - uip_connr->timer; | |
| /* This is taken directly from VJs original code in his paper */ | |
| m = m - (uip_connr->sa >> 3); | |
| uip_connr->sa += m; | |
| if(m < 0) { | |
| m = -m; | |
| } | |
| m = m - (uip_connr->sv >> 2); | |
| uip_connr->sv += m; | |
| uip_connr->rto = (uip_connr->sa >> 3) + uip_connr->sv; | |
| } | |
| /* Set the acknowledged flag. */ | |
| uip_flags = UIP_ACKDATA; | |
| /* Reset the retransmission timer. */ | |
| uip_connr->timer = uip_connr->rto; | |
| } | |
| } | |
| /* Do different things depending on in what state the connection is. */ | |
| switch(uip_connr->tcpstateflags & TS_MASK) { | |
| /* CLOSED and LISTEN are not handled here. CLOSE_WAIT is not | |
| implemented, since we force the application to close when the | |
| peer sends a FIN (hence the application goes directly from | |
| ESTABLISHED to LAST_ACK). */ | |
| case SYN_RCVD: | |
| /* In SYN_RCVD we have sent out a SYNACK in response to a SYN, and | |
| we are waiting for an ACK that acknowledges the data we sent | |
| out the last time. Therefore, we want to have the UIP_ACKDATA | |
| flag set. If so, we enter the ESTABLISHED state. */ | |
| if(uip_flags & UIP_ACKDATA) { | |
| uip_connr->tcpstateflags = ESTABLISHED; | |
| uip_flags = UIP_CONNECTED; | |
| uip_connr->len = 0; | |
| if(uip_len > 0) { | |
| uip_flags |= UIP_NEWDATA; | |
| uip_add_rcv_nxt(uip_len); | |
| } | |
| uip_slen = 0; | |
| UIP_APPCALL(); | |
| goto appsend; | |
| } | |
| goto drop; | |
| #if UIP_ACTIVE_OPEN | |
| case SYN_SENT: | |
| /* In SYN_SENT, we wait for a SYNACK that is sent in response to | |
| our SYN. The rcv_nxt is set to sequence number in the SYNACK | |
| plus one, and we send an ACK. We move into the ESTABLISHED | |
| state. */ | |
| if((uip_flags & UIP_ACKDATA) && | |
| BUF->flags == (TCP_SYN | TCP_ACK)) { | |
| /* Parse the TCP MSS option, if present. */ | |
| if((BUF->tcpoffset & 0xf0) > 0x50) { | |
| for(c = 0; c < ((BUF->tcpoffset >> 4) - 5) << 2 ;) { | |
| opt = uip_buf[40 + UIP_LLH_LEN + c]; | |
| if(opt == 0x00) { | |
| /* End of options. */ | |
| break; | |
| } else if(opt == 0x01) { | |
| ++c; | |
| /* NOP option. */ | |
| } else if(opt == 0x02 && | |
| uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c] == 0x04) { | |
| /* An MSS option with the right option length. */ | |
| tmp16 = (uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 2 + c] << 8) | | |
| uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 3 + c]; | |
| uip_connr->initialmss = | |
| uip_connr->mss = tmp16 > UIP_TCP_MSS? UIP_TCP_MSS: tmp16; | |
| /* And we are done processing options. */ | |
| break; | |
| } else { | |
| /* All other options have a length field, so that we easily | |
| can skip past them. */ | |
| if(uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c] == 0) { | |
| /* If the length field is zero, the options are malformed | |
| and we don't process them further. */ | |
| break; | |
| } | |
| c += uip_buf[UIP_TCPIP_HLEN + UIP_LLH_LEN + 1 + c]; | |
| } | |
| } | |
| } | |
| uip_connr->tcpstateflags = ESTABLISHED; | |
| uip_connr->rcv_nxt[0] = BUF->seqno[0]; | |
| uip_connr->rcv_nxt[1] = BUF->seqno[1]; | |
| uip_connr->rcv_nxt[2] = BUF->seqno[2]; | |
| uip_connr->rcv_nxt[3] = BUF->seqno[3]; | |
| uip_add_rcv_nxt(1); | |
| uip_flags = UIP_CONNECTED | UIP_NEWDATA; | |
| uip_connr->len = 0; | |
| uip_len = 0; | |
| uip_slen = 0; | |
| UIP_APPCALL(); | |
| goto appsend; | |
| } | |
| goto reset; | |
| #endif /* UIP_ACTIVE_OPEN */ | |
| case ESTABLISHED: | |
| /* In the ESTABLISHED state, we call upon the application to feed | |
| data into the uip_buf. If the UIP_ACKDATA flag is set, the | |
| application should put new data into the buffer, otherwise we are | |
| retransmitting an old segment, and the application should put that | |
| data into the buffer. | |
| If the incoming packet is a FIN, we should close the connection on | |
| this side as well, and we send out a FIN and enter the LAST_ACK | |
| state. We require that there is no outstanding data; otherwise the | |
| sequence numbers will be screwed up. */ | |
| if(BUF->flags & TCP_FIN) { | |
| if(uip_outstanding(uip_connr)) { | |
| goto drop; | |
| } | |
| uip_add_rcv_nxt(1 + uip_len); | |
| uip_flags = UIP_CLOSE; | |
| if(uip_len > 0) { | |
| uip_flags |= UIP_NEWDATA; | |
| } | |
| UIP_APPCALL(); | |
| uip_connr->len = 1; | |
| uip_connr->tcpstateflags = LAST_ACK; | |
| uip_connr->nrtx = 0; | |
| tcp_send_finack: | |
| BUF->flags = TCP_FIN | TCP_ACK; | |
| goto tcp_send_nodata; | |
| } | |
| /* Check the URG flag. If this is set, the segment carries urgent | |
| data that we must pass to the application. */ | |
| if(BUF->flags & TCP_URG) { | |
| #if UIP_URGDATA > 0 | |
| uip_urglen = (BUF->urgp[0] << 8) | BUF->urgp[1]; | |
| if(uip_urglen > uip_len) { | |
| /* There is more urgent data in the next segment to come. */ | |
| uip_urglen = uip_len; | |
| } | |
| uip_add_rcv_nxt(uip_urglen); | |
| uip_len -= uip_urglen; | |
| uip_urgdata = uip_appdata; | |
| uip_appdata += uip_urglen; | |
| } else { | |
| uip_urglen = 0; | |
| #endif /* UIP_URGDATA > 0 */ | |
| uip_appdata += (BUF->urgp[0] << 8) | BUF->urgp[1]; | |
| uip_len -= (BUF->urgp[0] << 8) | BUF->urgp[1]; | |
| } | |
| /* If uip_len > 0 we have TCP data in the packet, and we flag this | |
| by setting the UIP_NEWDATA flag and update the sequence number | |
| we acknowledge. If the application has stopped the dataflow | |
| using uip_stop(), we must not accept any data packets from the | |
| remote host. */ | |
| if(uip_len > 0 && !(uip_connr->tcpstateflags & UIP_STOPPED)) { | |
| uip_flags |= UIP_NEWDATA; | |
| uip_add_rcv_nxt(uip_len); | |
| } | |
| /* Check if the available buffer space advertised by the other end | |
| is smaller than the initial MSS for this connection. If so, we | |
| set the current MSS to the window size to ensure that the | |
| application does not send more data than the other end can | |
| handle. | |
| If the remote host advertises a zero window, we set the MSS to | |
| the initial MSS so that the application will send an entire MSS | |
| of data. This data will not be acknowledged by the receiver, | |
| and the application will retransmit it. This is called the | |
| "persistent timer" and uses the retransmission mechanim. | |
| */ | |
| tmp16 = ((u16_t)BUF->wnd[0] << 8) + (u16_t)BUF->wnd[1]; | |
| if(tmp16 > uip_connr->initialmss || | |
| tmp16 == 0) { | |
| tmp16 = uip_connr->initialmss; | |
| } | |
| uip_connr->mss = tmp16; | |
| /* If this packet constitutes an ACK for outstanding data (flagged | |
| by the UIP_ACKDATA flag, we should call the application since it | |
| might want to send more data. If the incoming packet had data | |
| from the peer (as flagged by the UIP_NEWDATA flag), the | |
| application must also be notified. | |
| When the application is called, the global variable uip_len | |
| contains the length of the incoming data. The application can | |
| access the incoming data through the global pointer | |
| uip_appdata, which usually points 40 bytes into the uip_buf | |
| array. | |
| If the application wishes to send any data, this data should be | |
| put into the uip_appdata and the length of the data should be | |
| put into uip_len. If the application don't have any data to | |
| send, uip_len must be set to 0. */ | |
| if(uip_flags & (UIP_NEWDATA | UIP_ACKDATA)) { | |
| uip_slen = 0; | |
| UIP_APPCALL(); | |
| appsend: | |
| if(uip_flags & UIP_ABORT) { | |
| uip_slen = 0; | |
| uip_connr->tcpstateflags = CLOSED; | |
| BUF->flags = TCP_RST | TCP_ACK; | |
| goto tcp_send_nodata; | |
| } | |
| if(uip_flags & UIP_CLOSE) { | |
| uip_slen = 0; | |
| uip_connr->len = 1; | |
| uip_connr->tcpstateflags = FIN_WAIT_1; | |
| uip_connr->nrtx = 0; | |
| BUF->flags = TCP_FIN | TCP_ACK; | |
| goto tcp_send_nodata; | |
| } | |
| /* If uip_slen > 0, the application has data to be sent. */ | |
| if(uip_slen > 0) { | |
| /* If the connection has acknowledged data, the contents of | |
| the ->len variable should be discarded. */ | |
| if((uip_flags & UIP_ACKDATA) != 0) { | |
| uip_connr->len = 0; | |
| } | |
| /* If the ->len variable is non-zero the connection has | |
| already data in transit and cannot send anymore right | |
| now. */ | |
| if(uip_connr->len == 0) { | |
| /* The application cannot send more than what is allowed by | |
| the mss (the minumum of the MSS and the available | |
| window). */ | |
| if(uip_slen > uip_connr->mss) { | |
| uip_slen = uip_connr->mss; | |
| } | |
| /* Remember how much data we send out now so that we know | |
| when everything has been acknowledged. */ | |
| uip_connr->len = uip_slen; | |
| } else { | |
| /* If the application already had unacknowledged data, we | |
| make sure that the application does not send (i.e., | |
| retransmit) out more than it previously sent out. */ | |
| uip_slen = uip_connr->len; | |
| } | |
| } else { | |
| uip_connr->len = 0; | |
| } | |
| uip_connr->nrtx = 0; | |
| apprexmit: | |
| uip_appdata = uip_sappdata; | |
| /* If the application has data to be sent, or if the incoming | |
| packet had new data in it, we must send out a packet. */ | |
| if(uip_slen > 0 && uip_connr->len > 0) { | |
| /* Add the length of the IP and TCP headers. */ | |
| uip_len = uip_connr->len + UIP_TCPIP_HLEN; | |
| /* We always set the ACK flag in response packets. */ | |
| BUF->flags = TCP_ACK | TCP_PSH; | |
| /* Send the packet. */ | |
| goto tcp_send_noopts; | |
| } | |
| /* If there is no data to send, just send out a pure ACK if | |
| there is newdata. */ | |
| if(uip_flags & UIP_NEWDATA) { | |
| uip_len = UIP_TCPIP_HLEN; | |
| BUF->flags = TCP_ACK; | |
| goto tcp_send_noopts; | |
| } | |
| } | |
| goto drop; | |
| case LAST_ACK: | |
| /* We can close this connection if the peer has acknowledged our | |
| FIN. This is indicated by the UIP_ACKDATA flag. */ | |
| if(uip_flags & UIP_ACKDATA) { | |
| uip_connr->tcpstateflags = CLOSED; | |
| uip_flags = UIP_CLOSE; | |
| UIP_APPCALL(); | |
| } | |
| break; | |
| case FIN_WAIT_1: | |
| /* The application has closed the connection, but the remote host | |
| hasn't closed its end yet. Thus we do nothing but wait for a | |
| FIN from the other side. */ | |
| if(uip_len > 0) { | |
| uip_add_rcv_nxt(uip_len); | |
| } | |
| if(BUF->flags & TCP_FIN) { | |
| if(uip_flags & UIP_ACKDATA) { | |
| uip_connr->tcpstateflags = TIME_WAIT; | |
| uip_connr->timer = 0; | |
| uip_connr->len = 0; | |
| } else { | |
| uip_connr->tcpstateflags = CLOSING; | |
| } | |
| uip_add_rcv_nxt(1); | |
| uip_flags = UIP_CLOSE; | |
| UIP_APPCALL(); | |
| goto tcp_send_ack; | |
| } else if(uip_flags & UIP_ACKDATA) { | |
| uip_connr->tcpstateflags = FIN_WAIT_2; | |
| uip_connr->len = 0; | |
| goto drop; | |
| } | |
| if(uip_len > 0) { | |
| goto tcp_send_ack; | |
| } | |
| goto drop; | |
| case FIN_WAIT_2: | |
| if(uip_len > 0) { | |
| uip_add_rcv_nxt(uip_len); | |
| } | |
| if(BUF->flags & TCP_FIN) { | |
| uip_connr->tcpstateflags = TIME_WAIT; | |
| uip_connr->timer = 0; | |
| uip_add_rcv_nxt(1); | |
| uip_flags = UIP_CLOSE; | |
| UIP_APPCALL(); | |
| goto tcp_send_ack; | |
| } | |
| if(uip_len > 0) { | |
| goto tcp_send_ack; | |
| } | |
| goto drop; | |
| case TIME_WAIT: | |
| goto tcp_send_ack; | |
| case CLOSING: | |
| if(uip_flags & UIP_ACKDATA) { | |
| uip_connr->tcpstateflags = TIME_WAIT; | |
| uip_connr->timer = 0; | |
| } | |
| } | |
| goto drop; | |
| /* We jump here when we are ready to send the packet, and just want | |
| to set the appropriate TCP sequence numbers in the TCP header. */ | |
| tcp_send_ack: | |
| BUF->flags = TCP_ACK; | |
| tcp_send_nodata: | |
| uip_len = 40; | |
| tcp_send_noopts: | |
| BUF->tcpoffset = 5 << 4; | |
| tcp_send: | |
| /* We're done with the input processing. We are now ready to send a | |
| reply. Our job is to fill in all the fields of the TCP and IP | |
| headers before calculating the checksum and finally send the | |
| packet. */ | |
| BUF->ackno[0] = uip_connr->rcv_nxt[0]; | |
| BUF->ackno[1] = uip_connr->rcv_nxt[1]; | |
| BUF->ackno[2] = uip_connr->rcv_nxt[2]; | |
| BUF->ackno[3] = uip_connr->rcv_nxt[3]; | |
| BUF->seqno[0] = uip_connr->snd_nxt[0]; | |
| BUF->seqno[1] = uip_connr->snd_nxt[1]; | |
| BUF->seqno[2] = uip_connr->snd_nxt[2]; | |
| BUF->seqno[3] = uip_connr->snd_nxt[3]; | |
| BUF->proto = UIP_PROTO_TCP; | |
| BUF->srcport = uip_connr->lport; | |
| BUF->destport = uip_connr->rport; | |
| BUF->srcipaddr[0] = uip_hostaddr[0]; | |
| BUF->srcipaddr[1] = uip_hostaddr[1]; | |
| BUF->destipaddr[0] = uip_connr->ripaddr[0]; | |
| BUF->destipaddr[1] = uip_connr->ripaddr[1]; | |
| if(uip_connr->tcpstateflags & UIP_STOPPED) { | |
| /* If the connection has issued uip_stop(), we advertise a zero | |
| window so that the remote host will stop sending data. */ | |
| BUF->wnd[0] = BUF->wnd[1] = 0; | |
| } else { | |
| BUF->wnd[0] = ((UIP_RECEIVE_WINDOW) >> 8); | |
| BUF->wnd[1] = ((UIP_RECEIVE_WINDOW) & 0xff); | |
| } | |
| tcp_send_noconn: | |
| BUF->len[0] = (uip_len >> 8); | |
| BUF->len[1] = (uip_len & 0xff); | |
| /* Calculate TCP checksum. */ | |
| BUF->tcpchksum = 0; | |
| BUF->tcpchksum = ~(uip_tcpchksum()); | |
| #if UIP_UDP | |
| ip_send_nolen: | |
| #endif | |
| BUF->vhl = 0x45; | |
| BUF->tos = 0; | |
| BUF->ipoffset[0] = BUF->ipoffset[1] = 0; | |
| BUF->ttl = UIP_TTL; | |
| ++ipid; | |
| BUF->ipid[0] = ipid >> 8; | |
| BUF->ipid[1] = ipid & 0xff; | |
| /* Calculate IP checksum. */ | |
| BUF->ipchksum = 0; | |
| BUF->ipchksum = ~(uip_ipchksum()); | |
| UIP_STAT(++uip_stat.tcp.sent); | |
| send: | |
| UIP_STAT(++uip_stat.ip.sent); | |
| /* Return and let the caller do the actual transmission. */ | |
| return; | |
| drop: | |
| uip_len = 0; | |
| return; | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| u16_t | |
| htons(u16_t val) | |
| { | |
| return HTONS(val); | |
| } | |
| /*-----------------------------------------------------------------------------------*/ | |
| /** @} */ |