rtoss - Rev 73

Subversion Repositories:
Rev:
/*
 * Copyright (c) 1995 Danny Gasparovski.
 *
 * Please read the file COPYRIGHT for the
 * terms and conditions of the copyright.
 */


#include "qemu-common.h"
#include <slirp.h>
#include "ip_icmp.h"
#ifdef __sun__
#include <sys/filio.h>
#endif

static void sofcantrcvmore(struct socket *so);
static void sofcantsendmore(struct socket *so);

struct socket *
solookup(struct socket *head, struct in_addr laddr, u_int lport,
         struct in_addr faddr, u_int fport)
{
        struct socket *so;

        for (so = head->so_next; so != head; so = so->so_next) {
                if (so->so_lport == lport &&
                    so->so_laddr.s_addr == laddr.s_addr &&
                    so->so_faddr.s_addr == faddr.s_addr &&
                    so->so_fport == fport)
                   break;
        }

        if (so == head)
           return (struct socket *)NULL;
        return so;

}

/*
 * Create a new socket, initialise the fields
 * It is the responsibility of the caller to
 * insque() it into the correct linked-list
 */

struct socket *
socreate(Slirp *slirp)
{
  struct socket *so;

  so = (struct socket *)malloc(sizeof(struct socket));
  if(so) {
    memset(so, 0, sizeof(struct socket));
    so->so_state = SS_NOFDREF;
    so->s = -1;
    so->slirp = slirp;
  }
  return(so);
}

/*
 * remque and free a socket, clobber cache
 */

void
sofree(struct socket *so)
{
  Slirp *slirp = so->slirp;

  if (so->so_emu==EMU_RSH && so->extra) {
        sofree(so->extra);
        so->extra=NULL;
  }
  if (so == slirp->tcp_last_so) {
      slirp->tcp_last_so = &slirp->tcb;
  } else if (so == slirp->udp_last_so) {
      slirp->udp_last_so = &slirp->udb;
  }
  m_free(so->so_m);

  if(so->so_next && so->so_prev)
    remque(so);  /* crashes if so is not in a queue */

  free(so);
}

size_t sopreprbuf(struct socket *so, struct iovec *iov, int *np)
{
        int n, lss, total;
        struct sbuf *sb = &so->so_snd;
        int len = sb->sb_datalen - sb->sb_cc;
        int mss = so->so_tcpcb->t_maxseg;

        DEBUG_CALL("sopreprbuf");
        DEBUG_ARG("so = %lx", (long )so);

        len = sb->sb_datalen - sb->sb_cc;

        if (len <= 0)
                return 0;

        iov[0].iov_base = sb->sb_wptr;
        iov[1].iov_base = NULL;
        iov[1].iov_len = 0;
        if (sb->sb_wptr < sb->sb_rptr) {
                iov[0].iov_len = sb->sb_rptr - sb->sb_wptr;
                /* Should never succeed, but... */
                if (iov[0].iov_len > len)
                   iov[0].iov_len = len;
                if (iov[0].iov_len > mss)
                   iov[0].iov_len -= iov[0].iov_len%mss;
                n = 1;
        } else {
                iov[0].iov_len = (sb->sb_data + sb->sb_datalen) - sb->sb_wptr;
                /* Should never succeed, but... */
                if (iov[0].iov_len > len) iov[0].iov_len = len;
                len -= iov[0].iov_len;
                if (len) {
                        iov[1].iov_base = sb->sb_data;
                        iov[1].iov_len = sb->sb_rptr - sb->sb_data;
                        if(iov[1].iov_len > len)
                           iov[1].iov_len = len;
                        total = iov[0].iov_len + iov[1].iov_len;
                        if (total > mss) {
                                lss = total%mss;
                                if (iov[1].iov_len > lss) {
                                        iov[1].iov_len -= lss;
                                        n = 2;
                                } else {
                                        lss -= iov[1].iov_len;
                                        iov[0].iov_len -= lss;
                                        n = 1;
                                }
                        } else
                                n = 2;
                } else {
                        if (iov[0].iov_len > mss)
                           iov[0].iov_len -= iov[0].iov_len%mss;
                        n = 1;
                }
        }
        if (np)
                *np = n;

        return iov[0].iov_len + (n - 1) * iov[1].iov_len;
}

/*
 * Read from so's socket into sb_snd, updating all relevant sbuf fields
 * NOTE: This will only be called if it is select()ed for reading, so
 * a read() of 0 (or less) means it's disconnected
 */

int
soread(struct socket *so)
{
        int n, nn;
        struct sbuf *sb = &so->so_snd;
        struct iovec iov[2];

        DEBUG_CALL("soread");
        DEBUG_ARG("so = %lx", (long )so);

        /*
         * No need to check if there's enough room to read.
         * soread wouldn't have been called if there weren't
         */

        sopreprbuf(so, iov, &n);

#ifdef HAVE_READV
        nn = readv(so->s, (struct iovec *)iov, n);
        DEBUG_MISC((dfd, " ... read nn = %d bytes\n", nn));
#else
        nn = recv(so->s, iov[0].iov_base, iov[0].iov_len,0);
#endif
        if (nn <= 0) {
                if (nn < 0 && (errno == EINTR || errno == EAGAIN))
                        return 0;
                else {
                        DEBUG_MISC((dfd, " --- soread() disconnected, nn = %d, errno = %d-%s\n", nn, errno,strerror(errno)));
                        sofcantrcvmore(so);
                        tcp_sockclosed(sototcpcb(so));
                        return -1;
                }
        }

#ifndef HAVE_READV
        /*
         * If there was no error, try and read the second time round
         * We read again if n = 2 (ie, there's another part of the buffer)
         * and we read as much as we could in the first read
         * We don't test for <= 0 this time, because there legitimately
         * might not be any more data (since the socket is non-blocking),
         * a close will be detected on next iteration.
         * A return of -1 wont (shouldn't) happen, since it didn't happen above
         */

        if (n == 2 && nn == iov[0].iov_len) {
            int ret;
            ret = recv(so->s, iov[1].iov_base, iov[1].iov_len,0);
            if (ret > 0)
                nn += ret;
        }

        DEBUG_MISC((dfd, " ... read nn = %d bytes\n", nn));
#endif

        /* Update fields */
        sb->sb_cc += nn;
        sb->sb_wptr += nn;
        if (sb->sb_wptr >= (sb->sb_data + sb->sb_datalen))
                sb->sb_wptr -= sb->sb_datalen;
        return nn;
}

int soreadbuf(struct socket *so, const char *buf, int size)
{
    int n, nn, copy = size;
        struct sbuf *sb = &so->so_snd;
        struct iovec iov[2];

        DEBUG_CALL("soreadbuf");
        DEBUG_ARG("so = %lx", (long )so);

        /*
         * No need to check if there's enough room to read.
         * soread wouldn't have been called if there weren't
         */

        if (sopreprbuf(so, iov, &n) < size)
        goto err;

    nn = MIN(iov[0].iov_len, copy);
    memcpy(iov[0].iov_base, buf, nn);

    copy -= nn;
    buf += nn;

    if (copy == 0)
        goto done;

    memcpy(iov[1].iov_base, buf, copy);

done:
    /* Update fields */
        sb->sb_cc += size;
        sb->sb_wptr += size;
        if (sb->sb_wptr >= (sb->sb_data + sb->sb_datalen))
                sb->sb_wptr -= sb->sb_datalen;
    return size;
err:

    sofcantrcvmore(so);
    tcp_sockclosed(sototcpcb(so));
    fprintf(stderr, "soreadbuf buffer to small");
    return -1;
}

/*
 * Get urgent data
 *
 * When the socket is created, we set it SO_OOBINLINE,
 * so when OOB data arrives, we soread() it and everything
 * in the send buffer is sent as urgent data
 */

void
sorecvoob(struct socket *so)
{
        struct tcpcb *tp = sototcpcb(so);

        DEBUG_CALL("sorecvoob");
        DEBUG_ARG("so = %lx", (long)so);

        /*
         * We take a guess at how much urgent data has arrived.
         * In most situations, when urgent data arrives, the next
         * read() should get all the urgent data.  This guess will
         * be wrong however if more data arrives just after the
         * urgent data, or the read() doesn't return all the
         * urgent data.
         */

        soread(so);
        tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
        tp->t_force = 1;
        tcp_output(tp);
        tp->t_force = 0;
}

/*
 * Send urgent data
 * There's a lot duplicated code here, but...
 */

int
sosendoob(struct socket *so)
{
        struct sbuf *sb = &so->so_rcv;
        char buff[2048]; /* XXX Shouldn't be sending more oob data than this */

        int n, len;

        DEBUG_CALL("sosendoob");
        DEBUG_ARG("so = %lx", (long)so);
        DEBUG_ARG("sb->sb_cc = %d", sb->sb_cc);

        if (so->so_urgc > 2048)
           so->so_urgc = 2048; /* XXXX */

        if (sb->sb_rptr < sb->sb_wptr) {
                /* We can send it directly */
                n = slirp_send(so, sb->sb_rptr, so->so_urgc, (MSG_OOB)); /* |MSG_DONTWAIT)); */
                so->so_urgc -= n;

                DEBUG_MISC((dfd, " --- sent %d bytes urgent data, %d urgent bytes left\n", n, so->so_urgc));
        } else {
                /*
                 * Since there's no sendv or sendtov like writev,
                 * we must copy all data to a linear buffer then
                 * send it all
                 */

                len = (sb->sb_data + sb->sb_datalen) - sb->sb_rptr;
                if (len > so->so_urgc) len = so->so_urgc;
                memcpy(buff, sb->sb_rptr, len);
                so->so_urgc -= len;
                if (so->so_urgc) {
                        n = sb->sb_wptr - sb->sb_data;
                        if (n > so->so_urgc) n = so->so_urgc;
                        memcpy((buff + len), sb->sb_data, n);
                        so->so_urgc -= n;
                        len += n;
                }
                n = slirp_send(so, buff, len, (MSG_OOB)); /* |MSG_DONTWAIT)); */
#ifdef DEBUG
                if (n != len)
                   DEBUG_ERROR((dfd, "Didn't send all data urgently XXXXX\n"));
#endif
                DEBUG_MISC((dfd, " ---2 sent %d bytes urgent data, %d urgent bytes left\n", n, so->so_urgc));
        }

        sb->sb_cc -= n;
        sb->sb_rptr += n;
        if (sb->sb_rptr >= (sb->sb_data + sb->sb_datalen))
                sb->sb_rptr -= sb->sb_datalen;

        return n;
}

/*
 * Write data from so_rcv to so's socket,
 * updating all sbuf field as necessary
 */

int
sowrite(struct socket *so)
{
        int  n,nn;
        struct sbuf *sb = &so->so_rcv;
        int len = sb->sb_cc;
        struct iovec iov[2];

        DEBUG_CALL("sowrite");
        DEBUG_ARG("so = %lx", (long)so);

        if (so->so_urgc) {
                sosendoob(so);
                if (sb->sb_cc == 0)
                        return 0;
        }

        /*
         * No need to check if there's something to write,
         * sowrite wouldn't have been called otherwise
         */


        len = sb->sb_cc;

        iov[0].iov_base = sb->sb_rptr;
        iov[1].iov_base = NULL;
        iov[1].iov_len = 0;
        if (sb->sb_rptr < sb->sb_wptr) {
                iov[0].iov_len = sb->sb_wptr - sb->sb_rptr;
                /* Should never succeed, but... */
                if (iov[0].iov_len > len) iov[0].iov_len = len;
                n = 1;
        } else {
                iov[0].iov_len = (sb->sb_data + sb->sb_datalen) - sb->sb_rptr;
                if (iov[0].iov_len > len) iov[0].iov_len = len;
                len -= iov[0].iov_len;
                if (len) {
                        iov[1].iov_base = sb->sb_data;
                        iov[1].iov_len = sb->sb_wptr - sb->sb_data;
                        if (iov[1].iov_len > len) iov[1].iov_len = len;
                        n = 2;
                } else
                        n = 1;
        }
        /* Check if there's urgent data to send, and if so, send it */

#ifdef HAVE_READV
        nn = writev(so->s, (const struct iovec *)iov, n);

        DEBUG_MISC((dfd, "  ... wrote nn = %d bytes\n", nn));
#else
        nn = slirp_send(so, iov[0].iov_base, iov[0].iov_len,0);
#endif
        /* This should never happen, but people tell me it does *shrug* */
        if (nn < 0 && (errno == EAGAIN || errno == EINTR))
                return 0;

        if (nn <= 0) {
                DEBUG_MISC((dfd, " --- sowrite disconnected, so->so_state = %x, errno = %d\n",
                        so->so_state, errno));
                sofcantsendmore(so);
                tcp_sockclosed(sototcpcb(so));
                return -1;
        }

#ifndef HAVE_READV
        if (n == 2 && nn == iov[0].iov_len) {
            int ret;
            ret = slirp_send(so, iov[1].iov_base, iov[1].iov_len,0);
            if (ret > 0)
                nn += ret;
        }
        DEBUG_MISC((dfd, "  ... wrote nn = %d bytes\n", nn));
#endif

        /* Update sbuf */
        sb->sb_cc -= nn;
        sb->sb_rptr += nn;
        if (sb->sb_rptr >= (sb->sb_data + sb->sb_datalen))
                sb->sb_rptr -= sb->sb_datalen;

        /*
         * If in DRAIN mode, and there's no more data, set
         * it CANTSENDMORE
         */

        if ((so->so_state & SS_FWDRAIN) && sb->sb_cc == 0)
                sofcantsendmore(so);

        return nn;
}

/*
 * recvfrom() a UDP socket
 */

void
sorecvfrom(struct socket *so)
{
        struct sockaddr_in addr;
        socklen_t addrlen = sizeof(struct sockaddr_in);

        DEBUG_CALL("sorecvfrom");
        DEBUG_ARG("so = %lx", (long)so);

        if (so->so_type == IPPROTO_ICMP) {   /* This is a "ping" reply */
          char buff[256];
          int len;

          len = recvfrom(so->s, buff, 256, 0,
                         (struct sockaddr *)&addr, &addrlen);
          /* XXX Check if reply is "correct"? */

          if(len == -1 || len == 0) {
            u_char code=ICMP_UNREACH_PORT;

            if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
            else if(errno == ENETUNREACH) code=ICMP_UNREACH_NET;

            DEBUG_MISC((dfd," udp icmp rx errno = %d-%s\n",
                        errno,strerror(errno)));
            icmp_error(so->so_m, ICMP_UNREACH,code, 0,strerror(errno));
          } else {
            icmp_reflect(so->so_m);
            so->so_m = NULL; /* Don't m_free() it again! */
          }
          /* No need for this socket anymore, udp_detach it */
          udp_detach(so);
        } else {                                /* A "normal" UDP packet */
          struct mbuf *m;
          int len;
#ifdef _WIN32
          unsigned long n;
#else
          int n;
#endif

          m = m_get(so->slirp);
          if (!m) {
              return;
          }
          m->m_data += IF_MAXLINKHDR;

          /*
           * XXX Shouldn't FIONREAD packets destined for port 53,
           * but I don't know the max packet size for DNS lookups
           */

          len = M_FREEROOM(m);
          /* if (so->so_fport != htons(53)) { */
          ioctlsocket(so->s, FIONREAD, &n);

          if (n > len) {
            n = (m->m_data - m->m_dat) + m->m_len + n + 1;
            m_inc(m, n);
            len = M_FREEROOM(m);
          }
          /* } */

          m->m_len = recvfrom(so->s, m->m_data, len, 0,
                              (struct sockaddr *)&addr, &addrlen);
          DEBUG_MISC((dfd, " did recvfrom %d, errno = %d-%s\n",
                      m->m_len, errno,strerror(errno)));
          if(m->m_len<0) {
            u_char code=ICMP_UNREACH_PORT;

            if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
            else if(errno == ENETUNREACH) code=ICMP_UNREACH_NET;

            DEBUG_MISC((dfd," rx error, tx icmp ICMP_UNREACH:%i\n", code));
            icmp_error(so->so_m, ICMP_UNREACH,code, 0,strerror(errno));
            m_free(m);
          } else {
          /*
           * Hack: domain name lookup will be used the most for UDP,
           * and since they'll only be used once there's no need
           * for the 4 minute (or whatever) timeout... So we time them
           * out much quicker (10 seconds  for now...)
           */

            if (so->so_expire) {
              if (so->so_fport == htons(53))
                so->so_expire = curtime + SO_EXPIREFAST;
              else
                so->so_expire = curtime + SO_EXPIRE;
            }

            /*
             * If this packet was destined for CTL_ADDR,
             * make it look like that's where it came from, done by udp_output
             */

            udp_output(so, m, &addr);
          } /* rx error */
        } /* if ping packet */
}

/*
 * sendto() a socket
 */

int
sosendto(struct socket *so, struct mbuf *m)
{
        Slirp *slirp = so->slirp;
        int ret;
        struct sockaddr_in addr;

        DEBUG_CALL("sosendto");
        DEBUG_ARG("so = %lx", (long)so);
        DEBUG_ARG("m = %lx", (long)m);

        addr.sin_family = AF_INET;
        if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) ==
            slirp->vnetwork_addr.s_addr) {
          /* It's an alias */
          if (so->so_faddr.s_addr == slirp->vnameserver_addr.s_addr) {
            if (get_dns_addr(&addr.sin_addr) < 0)
              addr.sin_addr = loopback_addr;
          } else {
            addr.sin_addr = loopback_addr;
          }
        } else
          addr.sin_addr = so->so_faddr;
        addr.sin_port = so->so_fport;

        DEBUG_MISC((dfd, " sendto()ing, addr.sin_port=%d, addr.sin_addr.s_addr=%.16s\n", ntohs(addr.sin_port), inet_ntoa(addr.sin_addr)));

        /* Don't care what port we get */
        ret = sendto(so->s, m->m_data, m->m_len, 0,
                     (struct sockaddr *)&addr, sizeof (struct sockaddr));
        if (ret < 0)
                return -1;

        /*
         * Kill the socket if there's no reply in 4 minutes,
         * but only if it's an expirable socket
         */

        if (so->so_expire)
                so->so_expire = curtime + SO_EXPIRE;
        so->so_state &= SS_PERSISTENT_MASK;
        so->so_state |= SS_ISFCONNECTED; /* So that it gets select()ed */
        return 0;
}

/*
 * Listen for incoming TCP connections
 */

struct socket *
tcp_listen(Slirp *slirp, u_int32_t haddr, u_int hport, u_int32_t laddr,
           u_int lport, int flags)
{
        struct sockaddr_in addr;
        struct socket *so;
        int s, opt = 1;
        socklen_t addrlen = sizeof(addr);

        DEBUG_CALL("tcp_listen");
        DEBUG_ARG("haddr = %x", haddr);
        DEBUG_ARG("hport = %d", hport);
        DEBUG_ARG("laddr = %x", laddr);
        DEBUG_ARG("lport = %d", lport);
        DEBUG_ARG("flags = %x", flags);

        so = socreate(slirp);
        if (!so) {
          return NULL;
        }

        /* Don't tcp_attach... we don't need so_snd nor so_rcv */
        if ((so->so_tcpcb = tcp_newtcpcb(so)) == NULL) {
                free(so);
                return NULL;
        }
        insque(so, &slirp->tcb);

        /*
         * SS_FACCEPTONCE sockets must time out.
         */

        if (flags & SS_FACCEPTONCE)
           so->so_tcpcb->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT*2;

        so->so_state &= SS_PERSISTENT_MASK;
        so->so_state |= (SS_FACCEPTCONN | flags);
        so->so_lport = lport; /* Kept in network format */
        so->so_laddr.s_addr = laddr; /* Ditto */

        addr.sin_family = AF_INET;
        addr.sin_addr.s_addr = haddr;
        addr.sin_port = hport;

        if (((s = socket(AF_INET,SOCK_STREAM,0)) < 0) ||
            (setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&opt,sizeof(int)) < 0) ||
            (bind(s,(struct sockaddr *)&addr, sizeof(addr)) < 0) ||
            (listen(s,1) < 0)) {
                int tmperrno = errno; /* Don't clobber the real reason we failed */

                close(s);
                sofree(so);
                /* Restore the real errno */
#ifdef _WIN32
                WSASetLastError(tmperrno);
#else
                errno = tmperrno;
#endif
                return NULL;
        }
        setsockopt(s,SOL_SOCKET,SO_OOBINLINE,(char *)&opt,sizeof(int));

        getsockname(s,(struct sockaddr *)&addr,&addrlen);
        so->so_fport = addr.sin_port;
        if (addr.sin_addr.s_addr == 0 || addr.sin_addr.s_addr == loopback_addr.s_addr)
           so->so_faddr = slirp->vhost_addr;
        else
           so->so_faddr = addr.sin_addr;

        so->s = s;
        return so;
}

/*
 * Various session state calls
 * XXX Should be #define's
 * The socket state stuff needs work, these often get call 2 or 3
 * times each when only 1 was needed
 */

void
soisfconnecting(struct socket *so)
{
        so->so_state &= ~(SS_NOFDREF|SS_ISFCONNECTED|SS_FCANTRCVMORE|
                          SS_FCANTSENDMORE|SS_FWDRAIN);
        so->so_state |= SS_ISFCONNECTING; /* Clobber other states */
}

void
soisfconnected(struct socket *so)
{
        so->so_state &= ~(SS_ISFCONNECTING|SS_FWDRAIN|SS_NOFDREF);
        so->so_state |= SS_ISFCONNECTED; /* Clobber other states */
}

static void
sofcantrcvmore(struct socket *so)
{
        if ((so->so_state & SS_NOFDREF) == 0) {
                shutdown(so->s,0);
                if(global_writefds) {
                  FD_CLR(so->s,global_writefds);
                }
        }
        so->so_state &= ~(SS_ISFCONNECTING);
        if (so->so_state & SS_FCANTSENDMORE) {
           so->so_state &= SS_PERSISTENT_MASK;
           so->so_state |= SS_NOFDREF; /* Don't select it */
        } else {
           so->so_state |= SS_FCANTRCVMORE;
        }
}

static void
sofcantsendmore(struct socket *so)
{
        if ((so->so_state & SS_NOFDREF) == 0) {
            shutdown(so->s,1);           /* send FIN to fhost */
            if (global_readfds) {
                FD_CLR(so->s,global_readfds);
            }
            if (global_xfds) {
                FD_CLR(so->s,global_xfds);
            }
        }
        so->so_state &= ~(SS_ISFCONNECTING);
        if (so->so_state & SS_FCANTRCVMORE) {
           so->so_state &= SS_PERSISTENT_MASK;
           so->so_state |= SS_NOFDREF; /* as above */
        } else {
           so->so_state |= SS_FCANTSENDMORE;
        }
}

/*
 * Set write drain mode
 * Set CANTSENDMORE once all data has been write()n
 */

void
sofwdrain(struct socket *so)
{
        if (so->so_rcv.sb_cc)
                so->so_state |= SS_FWDRAIN;
        else
                sofcantsendmore(so);
}