Initial revision
diff --git a/include/net.h b/include/net.h
new file mode 100644
index 0000000..d4985ca
--- /dev/null
+++ b/include/net.h
@@ -0,0 +1,358 @@
+/*
+ * LiMon Monitor (LiMon) - Network.
+ *
+ * Copyright 1994 - 2000 Neil Russell.
+ * (See License)
+ *
+ *
+ * History
+ * 9/16/00 bor adapted to TQM823L/STK8xxL board, RARP/TFTP boot added
+ */
+
+#ifndef __NET_H__
+#define __NET_H__
+
+#if !defined(CONFIG_NET_MULTI) && defined(CONFIG_8xx)
+#include <commproc.h>
+#if defined(FEC_ENET) || defined(SCC_ENET)
+#define CONFIG_NET_MULTI
+#endif
+#endif
+#include <asm/byteorder.h> /* for nton* / ntoh* stuff */
+
+
+/*
+ * The number of receive packet buffers, and the required packet buffer
+ * alignment in memory.
+ *
+ */
+
+#define PKTBUFSRX 4
+#define PKTALIGN 32
+
+typedef ulong IPaddr_t;
+
+
+
+/*
+ * The current receive packet handler. Called with a pointer to the
+ * application packet, and a protocol type (PORT_BOOTPC or PORT_TFTP).
+ * All other packets are dealt with without calling the handler.
+ */
+typedef void rxhand_f(uchar *, unsigned, unsigned, unsigned);
+
+/*
+ * A timeout handler. Called after time interval has expired.
+ */
+typedef void thand_f(void);
+
+#ifdef CONFIG_NET_MULTI
+
+#define NAMESIZE 16
+
+enum eth_state_t {
+ ETH_STATE_INIT,
+ ETH_STATE_PASSIVE,
+ ETH_STATE_ACTIVE
+};
+
+struct eth_device {
+ char name[NAMESIZE];
+ unsigned char enetaddr[6];
+ int iobase;
+ int state;
+
+ int (*init) (struct eth_device*, bd_t*);
+ int (*send) (struct eth_device*, volatile void* pachet, int length);
+ int (*recv) (struct eth_device*);
+ void (*halt) (struct eth_device*);
+
+ struct eth_device *next;
+ void *priv;
+};
+
+extern int eth_initialize(bd_t *bis); /* Initialize network subsystem */
+extern int eth_register(struct eth_device* dev);/* Register network device */
+extern void eth_try_another(int first_restart); /* Change the device */
+extern struct eth_device *eth_get_dev(void); /* get the current device MAC */
+extern void eth_set_enetaddr(int num, char* a); /* Set new MAC address */
+#endif
+
+extern int eth_init(bd_t *bis); /* Initialize the device */
+extern int eth_send(volatile void *packet, int length); /* Send a packet */
+extern int eth_rx(void); /* Check for received packets */
+extern void eth_halt(void); /* stop SCC */
+
+
+/**********************************************************************/
+/*
+ * Protocol headers.
+ */
+
+/*
+ * Ethernet header
+ */
+typedef struct {
+ uchar et_dest[6]; /* Destination node */
+ uchar et_src[6]; /* Source node */
+ ushort et_protlen; /* Protocol or length */
+ uchar et_dsap; /* 802 DSAP */
+ uchar et_ssap; /* 802 SSAP */
+ uchar et_ctl; /* 802 control */
+ uchar et_snap1; /* SNAP */
+ uchar et_snap2;
+ uchar et_snap3;
+ ushort et_prot; /* 802 protocol */
+} Ethernet_t;
+
+#define ETHER_HDR_SIZE 14 /* Ethernet header size */
+#define E802_HDR_SIZE 22 /* 802 ethernet header size */
+#define PROT_IP 0x0800 /* IP protocol */
+#define PROT_ARP 0x0806 /* IP ARP protocol */
+#define PROT_RARP 0x8035 /* IP ARP protocol */
+
+#define IPPROTO_ICMP 1 /* Internet Control Message Protocol */
+#define IPPROTO_UDP 17 /* User Datagram Protocol */
+
+/*
+ * Internet Protocol (IP) header.
+ */
+typedef struct {
+ uchar ip_hl_v; /* header length and version */
+ uchar ip_tos; /* type of service */
+ ushort ip_len; /* total length */
+ ushort ip_id; /* identification */
+ ushort ip_off; /* fragment offset field */
+ uchar ip_ttl; /* time to live */
+ uchar ip_p; /* protocol */
+ ushort ip_sum; /* checksum */
+ IPaddr_t ip_src; /* Source IP address */
+ IPaddr_t ip_dst; /* Destination IP address */
+ ushort udp_src; /* UDP source port */
+ ushort udp_dst; /* UDP destination port */
+ ushort udp_len; /* Length of UDP packet */
+ ushort udp_xsum; /* Checksum */
+} IP_t;
+
+#define IP_HDR_SIZE_NO_UDP (sizeof (IP_t) - 8)
+#define IP_HDR_SIZE (sizeof (IP_t))
+
+
+/*
+ * Address Resolution Protocol (ARP) header.
+ */
+typedef struct
+{
+ ushort ar_hrd; /* Format of hardware address */
+# define ARP_ETHER 1 /* Ethernet hardware address */
+ ushort ar_pro; /* Format of protocol address */
+ uchar ar_hln; /* Length of hardware address */
+ uchar ar_pln; /* Length of protocol address */
+ ushort ar_op; /* Operation */
+# define ARPOP_REQUEST 1 /* Request to resolve address */
+# define ARPOP_REPLY 2 /* Response to previous request */
+
+# define RARPOP_REQUEST 3 /* Request to resolve address */
+# define RARPOP_REPLY 4 /* Response to previous request */
+
+ /*
+ * The remaining fields are variable in size, according to
+ * the sizes above, and are defined as appropriate for
+ * specific hardware/protocol combinations.
+ */
+ uchar ar_data[0];
+#if 0
+ uchar ar_sha[]; /* Sender hardware address */
+ uchar ar_spa[]; /* Sender protocol address */
+ uchar ar_tha[]; /* Target hardware address */
+ uchar ar_tpa[]; /* Target protocol address */
+#endif /* 0 */
+} ARP_t;
+
+#define ARP_HDR_SIZE (8+20) /* Size assuming ethernet */
+
+/*
+ * ICMP stuff (just enough to handle (host) redirect messages)
+ */
+#define ICMP_REDIRECT 5 /* Redirect (change route) */
+
+/* Codes for REDIRECT. */
+#define ICMP_REDIR_NET 0 /* Redirect Net */
+#define ICMP_REDIR_HOST 1 /* Redirect Host */
+
+typedef struct icmphdr {
+ uchar type;
+ uchar code;
+ ushort checksum;
+ union {
+ struct {
+ ushort id;
+ ushort sequence;
+ } echo;
+ ulong gateway;
+ struct {
+ ushort __unused;
+ ushort mtu;
+ } frag;
+ } un;
+} ICMP_t;
+
+
+
+/*
+ * Maximum packet size; used to allocate packet storage.
+ * TFTP packets can be 524 bytes + IP header + ethernet header.
+ * Lets be conservative, and go for 38 * 16. (Must also be
+ * a multiple of 32 bytes).
+ */
+/*
+ * AS.HARNOIS : Better to set PKTSIZE to maximum size because
+ * traffic type is not always controlled
+ * maximum packet size = 1518
+ * maximum packet size and multiple of 32 bytes = 1536
+ */
+#define PKTSIZE 1518
+#define PKTSIZE_ALIGN 1536
+/*#define PKTSIZE 608*/
+
+/*
+ * Maximum receive ring size; that is, the number of packets
+ * we can buffer before overflow happens. Basically, this just
+ * needs to be enough to prevent a packet being discarded while
+ * we are processing the previous one.
+ */
+#define RINGSZ 4
+#define RINGSZ_LOG2 2
+
+/**********************************************************************/
+/*
+ * Globals.
+ *
+ * Note:
+ *
+ * All variables of type IPaddr_t are stored in NETWORK byte order
+ * (big endian).
+ */
+
+/* net.c */
+/** BOOTP EXTENTIONS **/
+extern IPaddr_t NetOurGatewayIP; /* Our gateway IP addresse */
+extern IPaddr_t NetOurSubnetMask; /* Our subnet mask (0 = unknown)*/
+extern IPaddr_t NetOurDNSIP; /* Our Domain Name Server (0 = unknown)*/
+extern char NetOurNISDomain[32]; /* Our NIS domain */
+extern char NetOurHostName[32]; /* Our hostname */
+extern char NetOurRootPath[64]; /* Our root path */
+extern ushort NetBootFileSize; /* Our boot file size in blocks */
+/** END OF BOOTP EXTENTIONS **/
+extern ulong NetBootFileXferSize; /* size of bootfile in bytes */
+extern uchar NetOurEther[6]; /* Our ethernet address */
+extern uchar NetServerEther[6]; /* Boot server enet address */
+extern IPaddr_t NetOurIP; /* Our IP addr (0 = unknown) */
+extern IPaddr_t NetServerIP; /* Server IP addr (0 = unknown) */
+extern volatile uchar * NetTxPacket; /* THE transmit packet */
+extern volatile uchar * NetRxPackets[PKTBUFSRX];/* Receive packets */
+extern volatile uchar * NetRxPkt; /* Current receive packet */
+extern int NetRxPktLen; /* Current rx packet length */
+extern unsigned NetIPID; /* IP ID (counting) */
+extern uchar NetBcastAddr[6]; /* Ethernet boardcast address */
+
+extern int NetState; /* Network loop state */
+#define NETLOOP_CONTINUE 1
+#define NETLOOP_RESTART 2
+#define NETLOOP_SUCCESS 3
+#define NETLOOP_FAIL 4
+
+#ifdef CONFIG_NET_MULTI
+extern int NetRestartWrap; /* Tried all network devices */
+#endif
+
+typedef enum { BOOTP, RARP, ARP, TFTP, DHCP } proto_t;
+
+/* from net/net.c */
+extern char BootFile[128]; /* Boot File name */
+
+/* Initialize the network adapter */
+extern int NetLoop(proto_t);
+
+/* Shutdown adapters and cleanup */
+extern void NetStop(void);
+
+/* Load failed. Start again. */
+extern void NetStartAgain(void);
+
+/* Set ethernet header */
+extern void NetSetEther(volatile uchar *, uchar *, uint);
+
+/* Set IP header */
+extern void NetSetIP(volatile uchar *, IPaddr_t, int, int, int);
+
+/* Checksum */
+extern int NetCksumOk(uchar *, int); /* Return true if cksum OK */
+extern uint NetCksum(uchar *, int); /* Calculate the checksum */
+
+/* Set callbacks */
+extern void NetSetHandler(rxhand_f *); /* Set RX packet handler */
+extern void NetSetTimeout(int, thand_f *); /* Set timeout handler */
+
+/* Transmit "NetTxPacket" */
+extern void NetSendPacket(volatile uchar *, int);
+
+/* Processes a received packet */
+extern void NetReceive(volatile uchar *, int);
+
+/* Print an IP address on the console */
+extern void print_IPaddr (IPaddr_t);
+
+/*
+ * The following functions are a bit ugly, but necessary to deal with
+ * alignment restrictions on ARM.
+ *
+ * We're using inline functions, which had the smallest memory
+ * footprint in our tests.
+ */
+/* return IP *in network byteorder* */
+static inline IPaddr_t NetReadIP(void *from)
+{
+ IPaddr_t ip;
+ memcpy((void*)&ip, from, sizeof(ip));
+ return ip;
+}
+
+/* return ulong *in network byteorder* */
+static inline ulong NetReadLong(ulong *from)
+{
+ ulong l;
+ memcpy((void*)&l, (void*)from, sizeof(l));
+ return l;
+}
+
+/* write IP *in network byteorder* */
+static inline void NetWriteIP(void *to, IPaddr_t ip)
+{
+ memcpy(to, (void*)&ip, sizeof(ip));
+}
+
+/* copy IP */
+static inline void NetCopyIP(void *to, void *from)
+{
+ memcpy(to, from, sizeof(IPaddr_t));
+}
+
+/* copy ulong */
+static inline void NetCopyLong(ulong *to, ulong *from)
+{
+ memcpy((void*)to, (void*)from, sizeof(ulong));
+}
+
+/* Convert an IP address to a string */
+extern void ip_to_string (IPaddr_t x, char *s);
+
+/* read an IP address from a environment variable */
+extern IPaddr_t getenv_IPaddr (char *);
+
+/* copy a filename (allow for "..." notation, limit length) */
+extern void copy_filename (uchar *dst, uchar *src, int size);
+
+/**********************************************************************/
+
+#endif /* __NET_H__ */
diff --git a/net/net.c b/net/net.c
new file mode 100644
index 0000000..cc3bec6
--- /dev/null
+++ b/net/net.c
@@ -0,0 +1,753 @@
+/*
+ * Copied from Linux Monitor (LiMon) - Networking.
+ *
+ * Copyright 1994 - 2000 Neil Russell.
+ * (See License)
+ * Copyright 2000 Roland Borde
+ * Copyright 2000 Paolo Scaffardi
+ * Copyright 2000-2002 Wolfgang Denk, wd@denx.de
+ */
+
+/*
+ * General Desription:
+ *
+ * The user interface supports commands for BOOTP, RARP, and TFTP.
+ * Also, we support ARP internally. Depending on available data,
+ * these interact as follows:
+ *
+ * BOOTP:
+ *
+ * Prerequisites: - own ethernet address
+ * We want: - own IP address
+ * - TFTP server IP address
+ * - name of bootfile
+ * Next step: ARP
+ *
+ * RARP:
+ *
+ * Prerequisites: - own ethernet address
+ * We want: - own IP address
+ * - TFTP server IP address
+ * Next step: ARP
+ *
+ * ARP:
+ *
+ * Prerequisites: - own ethernet address
+ * - own IP address
+ * - TFTP server IP address
+ * We want: - TFTP server ethernet address
+ * Next step: TFTP
+ *
+ * DHCP:
+ *
+ * Prerequisites: - own ethernet address
+ * We want: - IP, Netmask, ServerIP, Gateway IP
+ * - bootfilename, lease time
+ * Next step: - TFTP
+ *
+ * TFTP:
+ *
+ * Prerequisites: - own ethernet address
+ * - own IP address
+ * - TFTP server IP address
+ * - TFTP server ethernet address
+ * - name of bootfile (if unknown, we use a default name
+ * derived from our own IP address)
+ * We want: - load the boot file
+ * Next step: none
+ */
+
+
+#include <common.h>
+#include <watchdog.h>
+#include <command.h>
+#include <net.h>
+#include "bootp.h"
+#include "tftp.h"
+#include "rarp.h"
+#include "arp.h"
+
+#if (CONFIG_COMMANDS & CFG_CMD_NET)
+
+#if 0
+#define ET_DEBUG
+#endif
+
+/** BOOTP EXTENTIONS **/
+
+IPaddr_t NetOurSubnetMask=0; /* Our subnet mask (0=unknown) */
+IPaddr_t NetOurGatewayIP=0; /* Our gateways IP address */
+IPaddr_t NetOurDNSIP=0; /* Our DNS IP address */
+char NetOurNISDomain[32]={0,}; /* Our NIS domain */
+char NetOurHostName[32]={0,}; /* Our hostname */
+char NetOurRootPath[64]={0,}; /* Our bootpath */
+ushort NetBootFileSize=0; /* Our bootfile size in blocks */
+
+/** END OF BOOTP EXTENTIONS **/
+
+ulong NetBootFileXferSize; /* The actual transferred size of the bootfile (in bytes) */
+uchar NetOurEther[6]; /* Our ethernet address */
+uchar NetServerEther[6] = /* Boot server enet address */
+ { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
+IPaddr_t NetOurIP; /* Our IP addr (0 = unknown) */
+IPaddr_t NetServerIP; /* Our IP addr (0 = unknown) */
+volatile uchar *NetRxPkt; /* Current receive packet */
+int NetRxPktLen; /* Current rx packet length */
+unsigned NetIPID; /* IP packet ID */
+uchar NetBcastAddr[6] = /* Ethernet bcast address */
+ { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
+int NetState; /* Network loop state */
+#ifdef CONFIG_NET_MULTI
+int NetRestartWrap = 0; /* Tried all network devices */
+static int NetRestarted = 0; /* Network loop restarted */
+static int NetDevExists = 0; /* At least one device configured */
+#endif
+
+char BootFile[128]; /* Boot File name */
+
+volatile uchar PktBuf[(PKTBUFSRX+1) * PKTSIZE_ALIGN + PKTALIGN];
+
+volatile uchar *NetRxPackets[PKTBUFSRX]; /* Receive packets */
+
+static rxhand_f *packetHandler; /* Current RX packet handler */
+static thand_f *timeHandler; /* Current timeout handler */
+static ulong timeValue; /* Current timeout value */
+volatile uchar *NetTxPacket = 0; /* THE transmit packet */
+
+static int net_check_prereq (proto_t protocol);
+
+/**********************************************************************/
+/*
+ * Main network processing loop.
+ */
+
+int
+NetLoop(proto_t protocol)
+{
+ DECLARE_GLOBAL_DATA_PTR;
+
+ bd_t *bd = gd->bd;
+
+#ifdef CONFIG_NET_MULTI
+ NetRestarted = 0;
+ NetDevExists = 0;
+#endif
+
+ if (!NetTxPacket) {
+ int i;
+
+ /*
+ * Setup packet buffers, aligned correctly.
+ */
+ NetTxPacket = &PktBuf[0] + (PKTALIGN - 1);
+ NetTxPacket -= (ulong)NetTxPacket % PKTALIGN;
+ for (i = 0; i < PKTBUFSRX; i++) {
+ NetRxPackets[i] = NetTxPacket + (i+1)*PKTSIZE_ALIGN;
+ }
+ }
+
+ eth_halt();
+ eth_init(bd);
+
+restart:
+#ifdef CONFIG_NET_MULTI
+ memcpy (NetOurEther, eth_get_dev()->enetaddr, 6);
+#else
+ memcpy (NetOurEther, bd->bi_enetaddr, 6);
+#endif
+
+ NetState = NETLOOP_CONTINUE;
+
+ /*
+ * Start the ball rolling with the given start function. From
+ * here on, this code is a state machine driven by received
+ * packets and timer events.
+ */
+
+ switch (protocol) {
+ case TFTP:
+ NetCopyIP(&NetOurIP, &bd->bi_ip_addr);
+ NetServerIP = getenv_IPaddr ("serverip");
+ NetOurGatewayIP = getenv_IPaddr ("gatewayip");
+ NetOurSubnetMask= getenv_IPaddr ("netmask");
+ break;
+ case BOOTP:
+ case RARP:
+ /*
+ * initialize our IP addr to 0 in order to accept ANY
+ * IP addr assigned to us by the BOOTP / RARP server
+ */
+ NetOurIP = 0;
+ NetServerIP = 0;
+ break;
+ default:
+ break;
+ }
+
+ switch (net_check_prereq (protocol)) {
+ case 1:
+ /* network not configured */
+ return 0;
+
+#ifdef CONFIG_NET_MULTI
+ case 2:
+ /* network device not configured */
+ break;
+#endif /* CONFIG_NET_MULTI */
+
+ case 0:
+#ifdef CONFIG_NET_MULTI
+ NetDevExists = 1;
+#endif
+ switch (protocol) {
+ case TFTP:
+ /* always use ARP to get server ethernet address */
+ ArpTry = 0;
+ ArpRequest ();
+ break;
+
+#if (CONFIG_COMMANDS & CFG_CMD_DHCP)
+ case DHCP:
+ /* Start with a clean slate... */
+ NetOurIP = 0;
+ NetServerIP = 0;
+ DhcpRequest(); /* Basically same as BOOTP */
+ break;
+#endif /* CFG_CMD_DHCP */
+
+ case BOOTP:
+ BootpTry = 0;
+ BootpRequest ();
+ break;
+
+ case RARP:
+ RarpTry = 0;
+ RarpRequest ();
+ break;
+ default:
+ break;
+ }
+
+ NetBootFileXferSize = 0;
+ break;
+ }
+
+
+ /*
+ * Main packet reception loop. Loop receiving packets until
+ * someone sets `NetQuit'.
+ */
+ for (;;) {
+ WATCHDOG_RESET();
+#ifdef CONFIG_SHOW_ACTIVITY
+ {
+ extern void show_activity(int arg);
+ show_activity(1);
+ }
+#endif
+ /*
+ * Check the ethernet for a new packet. The ethernet
+ * receive routine will process it.
+ */
+ eth_rx();
+
+ /*
+ * Abort if ctrl-c was pressed.
+ */
+ if (ctrlc()) {
+ eth_halt();
+ printf("\nAbort\n");
+ return 0;
+ }
+
+
+ /*
+ * Check for a timeout, and run the timeout handler
+ * if we have one.
+ */
+ if (timeHandler && (get_timer(0) > timeValue)) {
+ thand_f *x;
+
+ x = timeHandler;
+ timeHandler = (thand_f *)0;
+ (*x)();
+ }
+
+
+ switch (NetState) {
+
+ case NETLOOP_RESTART:
+#ifdef CONFIG_NET_MULTI
+ NetRestarted = 1;
+#endif
+ goto restart;
+
+ case NETLOOP_SUCCESS:
+ if (NetBootFileXferSize > 0) {
+ char buf[10];
+ printf("Bytes transferred = %ld (%lx hex)\n",
+ NetBootFileXferSize,
+ NetBootFileXferSize);
+ sprintf(buf, "%lx", NetBootFileXferSize);
+ setenv("filesize", buf);
+ }
+ eth_halt();
+ return NetBootFileXferSize;
+
+ case NETLOOP_FAIL:
+ return 0;
+ }
+ }
+}
+
+/**********************************************************************/
+
+static void
+startAgainTimeout(void)
+{
+ NetState = NETLOOP_RESTART;
+}
+
+static void
+startAgainHandler(uchar * pkt, unsigned dest, unsigned src, unsigned len)
+{
+ /* Totally ignore the packet */
+}
+
+void
+NetStartAgain(void)
+{
+#ifndef CONFIG_NET_MULTI
+ NetSetTimeout(10 * CFG_HZ, startAgainTimeout);
+ NetSetHandler(startAgainHandler);
+#else
+ DECLARE_GLOBAL_DATA_PTR;
+
+ eth_halt();
+ eth_try_another(!NetRestarted);
+ eth_init(gd->bd);
+ if (NetRestartWrap)
+ {
+ NetRestartWrap = 0;
+ if (NetDevExists)
+ {
+ NetSetTimeout(10 * CFG_HZ, startAgainTimeout);
+ NetSetHandler(startAgainHandler);
+ }
+ else
+ {
+ NetState = NETLOOP_FAIL;
+ }
+ }
+ else
+ {
+ NetState = NETLOOP_RESTART;
+ }
+#endif
+}
+
+/**********************************************************************/
+/*
+ * Miscelaneous bits.
+ */
+
+void
+NetSetHandler(rxhand_f * f)
+{
+ packetHandler = f;
+}
+
+
+void
+NetSetTimeout(int iv, thand_f * f)
+{
+ if (iv == 0) {
+ timeHandler = (thand_f *)0;
+ } else {
+ timeHandler = f;
+ timeValue = get_timer(0) + iv;
+ }
+}
+
+
+void
+NetSendPacket(volatile uchar * pkt, int len)
+{
+ (void) eth_send(pkt, len);
+}
+
+
+
+void
+NetReceive(volatile uchar * pkt, int len)
+{
+ Ethernet_t *et;
+ IP_t *ip;
+ ARP_t *arp;
+ IPaddr_t tmp;
+ int x;
+
+
+ NetRxPkt = pkt;
+ NetRxPktLen = len;
+ et = (Ethernet_t *)pkt;
+
+ x = ntohs(et->et_protlen);
+
+ if (x < 1514) {
+ /*
+ * Got a 802 packet. Check the other protocol field.
+ */
+ x = ntohs(et->et_prot);
+ ip = (IP_t *)(pkt + E802_HDR_SIZE);
+ len -= E802_HDR_SIZE;
+ } else {
+ ip = (IP_t *)(pkt + ETHER_HDR_SIZE);
+ len -= ETHER_HDR_SIZE;
+ }
+
+#ifdef ET_DEBUG
+ printf("Receive from protocol 0x%x\n", x);
+#endif
+
+ switch (x) {
+
+ case PROT_ARP:
+ /*
+ * We have to deal with two types of ARP packets:
+ * - REQUEST packets will be answered by sending our
+ * IP address - if we know it.
+ * - REPLY packates are expected only after we asked
+ * for the TFTP server's or the gateway's ethernet
+ * address; so if we receive such a packet, we set
+ * the server ethernet address
+ */
+#ifdef ET_DEBUG
+ printf("Got ARP\n");
+#endif
+ arp = (ARP_t *)ip;
+ if (len < ARP_HDR_SIZE) {
+ printf("bad length %d < %d\n", len, ARP_HDR_SIZE);
+ return;
+ }
+ if (ntohs(arp->ar_hrd) != ARP_ETHER) {
+ return;
+ }
+ if (ntohs(arp->ar_pro) != PROT_IP) {
+ return;
+ }
+ if (arp->ar_hln != 6) {
+ return;
+ }
+ if (arp->ar_pln != 4) {
+ return;
+ }
+
+ if (NetOurIP == 0) {
+ return;
+ }
+
+ if (NetReadIP(&arp->ar_data[16]) != NetOurIP) {
+ return;
+ }
+
+ switch (ntohs(arp->ar_op)) {
+ case ARPOP_REQUEST: /* reply with our IP address */
+#ifdef ET_DEBUG
+ printf("Got ARP REQUEST, return our IP\n");
+#endif
+ NetSetEther((uchar *)et, et->et_src, PROT_ARP);
+ arp->ar_op = htons(ARPOP_REPLY);
+ memcpy (&arp->ar_data[10], &arp->ar_data[0], 6);
+ NetCopyIP(&arp->ar_data[16], &arp->ar_data[6]);
+ memcpy (&arp->ar_data[ 0], NetOurEther, 6);
+ NetCopyIP(&arp->ar_data[ 6], &NetOurIP);
+ NetSendPacket((uchar *)et,((uchar *)arp-pkt)+ARP_HDR_SIZE);
+ return;
+ case ARPOP_REPLY: /* set TFTP server eth addr */
+#ifdef ET_DEBUG
+ printf("Got ARP REPLY, set server/gtwy eth addr\n");
+#endif
+ memcpy (NetServerEther, &arp->ar_data[0], 6);
+ (*packetHandler)(0,0,0,0); /* start TFTP */
+ return;
+ default:
+#ifdef ET_DEBUG
+ printf("Unexpected ARP opcode 0x%x\n", ntohs(arp->ar_op));
+#endif
+ return;
+ }
+
+ case PROT_RARP:
+#ifdef ET_DEBUG
+ printf("Got RARP\n");
+#endif
+ arp = (ARP_t *)ip;
+ if (len < ARP_HDR_SIZE) {
+ printf("bad length %d < %d\n", len, ARP_HDR_SIZE);
+ return;
+ }
+
+ if ((ntohs(arp->ar_op) != RARPOP_REPLY) ||
+ (ntohs(arp->ar_hrd) != ARP_ETHER) ||
+ (ntohs(arp->ar_pro) != PROT_IP) ||
+ (arp->ar_hln != 6) || (arp->ar_pln != 4)) {
+
+ printf("invalid RARP header\n");
+ } else {
+ NetCopyIP(&NetOurIP, &arp->ar_data[16]);
+ NetCopyIP(&NetServerIP, &arp->ar_data[ 6]);
+ memcpy (NetServerEther, &arp->ar_data[ 0], 6);
+
+ (*packetHandler)(0,0,0,0);
+ }
+ break;
+
+ case PROT_IP:
+#ifdef ET_DEBUG
+ printf("Got IP\n");
+#endif
+ if (len < IP_HDR_SIZE) {
+ debug ("len bad %d < %d\n", len, IP_HDR_SIZE);
+ return;
+ }
+ if (len < ntohs(ip->ip_len)) {
+ printf("len bad %d < %d\n", len, ntohs(ip->ip_len));
+ return;
+ }
+ len = ntohs(ip->ip_len);
+#ifdef ET_DEBUG
+ printf("len=%d, v=%02x\n", len, ip->ip_hl_v & 0xff);
+#endif
+ if ((ip->ip_hl_v & 0xf0) != 0x40) {
+ return;
+ }
+ if (ip->ip_off & htons(0x1fff)) { /* Can't deal w/ fragments */
+ return;
+ }
+ if (!NetCksumOk((uchar *)ip, IP_HDR_SIZE_NO_UDP / 2)) {
+ printf("checksum bad\n");
+ return;
+ }
+ tmp = NetReadIP(&ip->ip_dst);
+ if (NetOurIP && tmp != NetOurIP && tmp != 0xFFFFFFFF) {
+ return;
+ }
+ /*
+ * watch for ICMP host redirects
+ *
+ * There is no real handler code (yet). We just watch
+ * for ICMP host redirect messages. In case anybody
+ * sees these messages: please contact me
+ * (wd@denx.de), or - even better - send me the
+ * necessary fixes :-)
+ *
+ * Note: in all cases where I have seen this so far
+ * it was a problem with the router configuration,
+ * for instance when a router was configured in the
+ * BOOTP reply, but the TFTP server was on the same
+ * subnet. So this is probably a warning that your
+ * configuration might be wrong. But I'm not really
+ * sure if there aren't any other situations.
+ */
+ if (ip->ip_p == IPPROTO_ICMP) {
+ ICMP_t *icmph = (ICMP_t *)&(ip->udp_src);
+
+ if (icmph->type != ICMP_REDIRECT)
+ return;
+ if (icmph->code != ICMP_REDIR_HOST)
+ return;
+ puts (" ICMP Host Redirect to ");
+ print_IPaddr(icmph->un.gateway);
+ putc(' ');
+ } else if (ip->ip_p != IPPROTO_UDP) { /* Only UDP packets */
+ return;
+ }
+
+ /*
+ * IP header OK. Pass the packet to the current handler.
+ */
+ (*packetHandler)((uchar *)ip +IP_HDR_SIZE,
+ ntohs(ip->udp_dst),
+ ntohs(ip->udp_src),
+ ntohs(ip->udp_len) - 8);
+
+ break;
+ }
+}
+
+
+/**********************************************************************/
+
+static int net_check_prereq (proto_t protocol)
+{
+ switch (protocol) {
+ case ARP: /* nothing to do */
+ break;
+
+ case TFTP:
+ if (NetServerIP == 0) {
+ puts ("*** ERROR: `serverip' not set\n");
+ return (1);
+ }
+
+ if (NetOurIP == 0) {
+ puts ("*** ERROR: `ipaddr' not set\n");
+ return (1);
+ }
+ /* Fall through */
+
+ case DHCP:
+ case RARP:
+ case BOOTP:
+ if (memcmp(NetOurEther, "\0\0\0\0\0\0", 6) == 0) {
+#ifdef CONFIG_NET_MULTI
+ extern int eth_get_dev_index (void);
+ int num = eth_get_dev_index();
+
+ switch (num) {
+ case -1:
+ puts ("*** ERROR: No ethernet found.\n");
+ return (1);
+ case 0:
+ puts ("*** ERROR: `ethaddr' not set\n");
+ break;
+ default:
+ printf ("*** ERROR: `eth%daddr' not set\n",
+ num);
+ break;
+ }
+
+ NetStartAgain ();
+ return (2);
+#else
+ puts ("*** ERROR: `ethaddr' not set\n");
+ return (1);
+#endif
+ }
+ /* Fall through */
+ }
+ return (0); /* OK */
+}
+/**********************************************************************/
+
+int
+NetCksumOk(uchar * ptr, int len)
+{
+ return !((NetCksum(ptr, len) + 1) & 0xfffe);
+}
+
+
+unsigned
+NetCksum(uchar * ptr, int len)
+{
+ ulong xsum;
+
+ xsum = 0;
+ while (len-- > 0)
+ xsum += *((ushort *)ptr)++;
+ xsum = (xsum & 0xffff) + (xsum >> 16);
+ xsum = (xsum & 0xffff) + (xsum >> 16);
+ return (xsum & 0xffff);
+}
+
+
+void
+NetSetEther(volatile uchar * xet, uchar * addr, uint prot)
+{
+ Ethernet_t *et = (Ethernet_t *)xet;
+
+ memcpy (et->et_dest, addr, 6);
+ memcpy (et->et_src, NetOurEther, 6);
+ et->et_protlen = htons(prot);
+}
+
+
+void
+NetSetIP(volatile uchar * xip, IPaddr_t dest, int dport, int sport, int len)
+{
+ volatile IP_t *ip = (IP_t *)xip;
+
+ /*
+ * If the data is an odd number of bytes, zero the
+ * byte after the last byte so that the checksum
+ * will work.
+ */
+ if (len & 1)
+ xip[IP_HDR_SIZE + len] = 0;
+
+ /*
+ * Construct an IP and UDP header.
+ (need to set no fragment bit - XXX)
+ */
+ ip->ip_hl_v = 0x45; /* IP_HDR_SIZE / 4 (not including UDP) */
+ ip->ip_tos = 0;
+ ip->ip_len = htons(IP_HDR_SIZE + len);
+ ip->ip_id = htons(NetIPID++);
+ ip->ip_off = htons(0x4000); /* No fragmentation */
+ ip->ip_ttl = 255;
+ ip->ip_p = 17; /* UDP */
+ ip->ip_sum = 0;
+ NetCopyIP((void*)&ip->ip_src, &NetOurIP); /* already in network byte order */
+ NetCopyIP((void*)&ip->ip_dst, &dest); /* - "" - */
+ ip->udp_src = htons(sport);
+ ip->udp_dst = htons(dport);
+ ip->udp_len = htons(8 + len);
+ ip->udp_xsum = 0;
+ ip->ip_sum = ~NetCksum((uchar *)ip, IP_HDR_SIZE_NO_UDP / 2);
+}
+
+void copy_filename (uchar *dst, uchar *src, int size)
+{
+ if (*src && (*src == '"')) {
+ ++src;
+ --size;
+ }
+
+ while ((--size > 0) && *src && (*src != '"')) {
+ *dst++ = *src++;
+ }
+ *dst = '\0';
+}
+
+#endif /* CFG_CMD_NET */
+
+void ip_to_string (IPaddr_t x, char *s)
+{
+ x = ntohl(x);
+ sprintf (s,"%d.%d.%d.%d",
+ (int)((x >> 24) & 0xff),
+ (int)((x >> 16) & 0xff),
+ (int)((x >> 8) & 0xff),
+ (int)((x >> 0) & 0xff)
+ );
+}
+
+void print_IPaddr (IPaddr_t x)
+{
+ char tmp[16];
+
+ ip_to_string(x, tmp);
+
+ puts(tmp);
+}
+
+IPaddr_t getenv_IPaddr (char *var)
+{
+ IPaddr_t addr;
+ char *s, *e;
+ int i;
+
+ s = getenv (var);
+
+ for (addr=0, i=0; i<4; ++i) {
+ ulong val = s ? simple_strtoul(s, &e, 10) : 0;
+ addr <<= 8;
+ addr |= (val & 0xFF);
+ if (s) {
+ s = (*e) ? e+1 : e;
+ }
+ }
+
+ return (htonl(addr));
+}