slirp: Forward ICMP echo requests via unprivileged sockets

Linux 3.0 gained support for unprivileged ICMP ping sockets. Use this
feature to forward guest pings to the outer world. The host admin has to
set the ping_group_range in order to grant access to those sockets. To
allow ping for the users group (GID 100):

echo 100 100 > /proc/sys/net/ipv4/ping_group_range

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

7 files changed, 147 insertions, 1 deletions

diff --git a/slirp/ip_icmp.c b/slirp/ip_icmp.c
index 4f10826..14a5312 100644
--- a/slirp/ip_icmp.c
+++ b/slirp/ip_icmp.c
@@ -60,6 +60,52 @@ static const int icmp_flush[19] = {
 /* ADDR MASK REPLY (18) */ 0
 };
 
+void icmp_init(Slirp *slirp)
+{
+    slirp->icmp.so_next = slirp->icmp.so_prev = &slirp->icmp;
+    slirp->icmp_last_so = &slirp->icmp;
+}
+
+static int icmp_send(struct socket *so, struct mbuf *m, int hlen)
+{
+    struct ip *ip = mtod(m, struct ip *);
+    struct sockaddr_in addr;
+
+    so->s = qemu_socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP);
+    if (so->s == -1) {
+        return -1;
+    }
+
+    so->so_m = m;
+    so->so_faddr = ip->ip_dst;
+    so->so_laddr = ip->ip_src;
+    so->so_iptos = ip->ip_tos;
+    so->so_type = IPPROTO_ICMP;
+    so->so_state = SS_ISFCONNECTED;
+    so->so_expire = curtime + SO_EXPIRE;
+
+    addr.sin_family = AF_INET;
+    addr.sin_addr = so->so_faddr;
+
+    insque(so, &so->slirp->icmp);
+
+    if (sendto(so->s, m->m_data + hlen, m->m_len - hlen, 0,
+               (struct sockaddr *)&addr, sizeof(addr)) == -1) {
+        DEBUG_MISC((dfd, "icmp_input icmp sendto tx errno = %d-%s\n",
+                    errno, strerror(errno)));
+        icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, 0, strerror(errno));
+        icmp_detach(so);
+    }
+
+    return 0;
+}
+
+void icmp_detach(struct socket *so)
+{
+    closesocket(so->s);
+    sofree(so);
+}
+
 /*
  * Process a received ICMP message.
  */
@@ -97,7 +143,6 @@ icmp_input(struct mbuf *m, int hlen)
   DEBUG_ARG("icmp_type = %d", icp->icmp_type);
   switch (icp->icmp_type) {
   case ICMP_ECHO:
-    icp->icmp_type = ICMP_ECHOREPLY;
     ip->ip_len += hlen;	             /* since ip_input subtracts this */
     if (ip->ip_dst.s_addr == slirp->vhost_addr.s_addr) {
       icmp_reflect(m);
@@ -107,6 +152,9 @@ icmp_input(struct mbuf *m, int hlen)
       struct socket *so;
       struct sockaddr_in addr;
       if ((so = socreate(slirp)) == NULL) goto freeit;
+      if (icmp_send(so, m, hlen) == 0) {
+        return;
+      }
       if(udp_attach(so) == -1) {
 	DEBUG_MISC((dfd,"icmp_input udp_attach errno = %d-%s\n",
 		    errno,strerror(errno)));
@@ -321,6 +369,7 @@ icmp_reflect(struct mbuf *m)
   m->m_len -= hlen;
   icp = mtod(m, struct icmp *);
 
+  icp->icmp_type = ICMP_ECHOREPLY;
   icp->icmp_cksum = 0;
   icp->icmp_cksum = cksum(m, ip->ip_len - hlen);
 
@@ -351,3 +400,39 @@ icmp_reflect(struct mbuf *m)
 
   (void ) ip_output((struct socket *)NULL, m);
 }
+
+void icmp_receive(struct socket *so)
+{
+    struct mbuf *m = so->so_m;
+    struct ip *ip = mtod(m, struct ip *);
+    int hlen = ip->ip_hl << 2;
+    u_char error_code;
+    struct icmp *icp;
+    int id, len;
+
+    m->m_data += hlen;
+    m->m_len -= hlen;
+    icp = mtod(m, struct icmp *);
+
+    id = icp->icmp_id;
+    len = recv(so->s, icp, m->m_len, 0);
+    icp->icmp_id = id;
+
+    m->m_data -= hlen;
+    m->m_len += hlen;
+
+    if (len == -1 || len == 0) {
+        if (errno == ENETUNREACH) {
+            error_code = ICMP_UNREACH_NET;
+        } else {
+            error_code = ICMP_UNREACH_HOST;
+        }
+        DEBUG_MISC((dfd, " udp icmp rx errno = %d-%s\n", errno,
+                    strerror(errno)));
+        icmp_error(so->so_m, ICMP_UNREACH, error_code, 0, strerror(errno));
+    } else {
+        icmp_reflect(so->so_m);
+        so->so_m = NULL; /* Don't m_free() it again! */
+    }
+    icmp_detach(so);
+}
diff --git a/slirp/ip_icmp.h b/slirp/ip_icmp.h
index 2692822..b3da1f2 100644
--- a/slirp/ip_icmp.h
+++ b/slirp/ip_icmp.h
@@ -153,9 +153,12 @@ struct icmp {
 	(type) == ICMP_IREQ || (type) == ICMP_IREQREPLY || \
 	(type) == ICMP_MASKREQ || (type) == ICMP_MASKREPLY)
 
+void icmp_init(Slirp *slirp);
 void icmp_input(struct mbuf *, int);
 void icmp_error(struct mbuf *msrc, u_char type, u_char code, int minsize,
                 const char *message);
 void icmp_reflect(struct mbuf *);
+void icmp_receive(struct socket *so);
+void icmp_detach(struct socket *so);
 
 #endif
diff --git a/slirp/ip_input.c b/slirp/ip_input.c
index 46c60b0..5e67631 100644
--- a/slirp/ip_input.c
+++ b/slirp/ip_input.c
@@ -58,6 +58,7 @@ ip_init(Slirp *slirp)
     slirp->ipq.ip_link.next = slirp->ipq.ip_link.prev = &slirp->ipq.ip_link;
     udp_init(slirp);
     tcp_init(slirp);
+    icmp_init(slirp);
 }
 
 /*
diff --git a/slirp/misc.c b/slirp/misc.c
index 34179e2..6002550 100644
--- a/slirp/misc.c
+++ b/slirp/misc.c
@@ -407,4 +407,17 @@ void slirp_connection_info(Slirp *slirp, Monitor *mon)
                        inet_ntoa(dst_addr), ntohs(dst_port),
                        so->so_rcv.sb_cc, so->so_snd.sb_cc);
     }
+
+    for (so = slirp->icmp.so_next; so != &slirp->icmp; so = so->so_next) {
+        n = snprintf(buf, sizeof(buf), "  ICMP[%d sec]",
+                     (so->so_expire - curtime) / 1000);
+        src.sin_addr = so->so_laddr;
+        dst_addr = so->so_faddr;
+        memset(&buf[n], ' ', 19 - n);
+        buf[19] = 0;
+        monitor_printf(mon, "%s %3d %15s  -    ", buf, so->s,
+                       src.sin_addr.s_addr ? inet_ntoa(src.sin_addr) : "*");
+        monitor_printf(mon, "%15s  -    %5d %5d\n", inet_ntoa(dst_addr),
+                       so->so_rcv.sb_cc, so->so_snd.sb_cc);
+    }
 }
diff --git a/slirp/slirp.c b/slirp/slirp.c
index 1593be1..faaa2f3 100644
--- a/slirp/slirp.c
+++ b/slirp/slirp.c
@@ -373,6 +373,31 @@ void slirp_select_fill(int *pnfds,
 				UPD_NFDS(so->s);
 			}
 		}
+
+                /*
+                 * ICMP sockets
+                 */
+                for (so = slirp->icmp.so_next; so != &slirp->icmp;
+                     so = so_next) {
+                    so_next = so->so_next;
+
+                    /*
+                     * See if it's timed out
+                     */
+                    if (so->so_expire) {
+                        if (so->so_expire <= curtime) {
+                            icmp_detach(so);
+                            continue;
+                        } else {
+                            do_slowtimo = 1; /* Let socket expire */
+                        }
+                    }
+
+                    if (so->so_state & SS_ISFCONNECTED) {
+                        FD_SET(so->s, readfds);
+                        UPD_NFDS(so->s);
+                    }
+                }
 	}
 
         *pnfds = nfds;
@@ -542,6 +567,18 @@ void slirp_select_poll(fd_set *readfds, fd_set *writefds, fd_set *xfds,
                             sorecvfrom(so);
                         }
 		}
+
+                /*
+                 * Check incoming ICMP relies.
+                 */
+                for (so = slirp->icmp.so_next; so != &slirp->icmp;
+                     so = so_next) {
+                     so_next = so->so_next;
+
+                    if (so->s != -1 && FD_ISSET(so->s, readfds)) {
+                        icmp_receive(so);
+                    }
+                }
 	}
 
 	/*
diff --git a/slirp/slirp.h b/slirp/slirp.h
index 954289a..16bb6ba 100644
--- a/slirp/slirp.h
+++ b/slirp/slirp.h
@@ -152,6 +152,7 @@ int inet_aton(const char *cp, struct in_addr *ia);
 #include "tcp_var.h"
 #include "tcpip.h"
 #include "udp.h"
+#include "ip_icmp.h"
 #include "mbuf.h"
 #include "sbuf.h"
 #include "socket.h"
@@ -218,6 +219,10 @@ struct Slirp {
     struct socket udb;
     struct socket *udp_last_so;
 
+    /* icmp states */
+    struct socket icmp;
+    struct socket *icmp_last_so;
+
     /* tftp states */
     char *tftp_prefix;
     struct tftp_session tftp_sessions[TFTP_SESSIONS_MAX];
diff --git a/slirp/socket.c b/slirp/socket.c
index 6119234..9b8ae13 100644
--- a/slirp/socket.c
+++ b/slirp/socket.c
@@ -71,6 +71,8 @@ sofree(struct socket *so)
       slirp->tcp_last_so = &slirp->tcb;
   } else if (so == slirp->udp_last_so) {
       slirp->udp_last_so = &slirp->udb;
+  } else if (so == slirp->icmp_last_so) {
+      slirp->icmp_last_so = &slirp->icmp;
   }
   m_free(so->so_m);