summaryrefslogtreecommitdiffstats
path: root/lib/libncp/sap.c
diff options
context:
space:
mode:
authorbp <bp@FreeBSD.org>1999-10-12 11:56:41 +0000
committerbp <bp@FreeBSD.org>1999-10-12 11:56:41 +0000
commit41cfec2c0234181158049ff25ca1f5de503d7f1a (patch)
tree5df7367bbf1788b36cc6e008c01b2b3fe1e5f0e9 /lib/libncp/sap.c
parentc364c2223a670b6987002786543b211abaca4862 (diff)
downloadFreeBSD-src-41cfec2c0234181158049ff25ca1f5de503d7f1a.zip
FreeBSD-src-41cfec2c0234181158049ff25ca1f5de503d7f1a.tar.gz
Initial import of ncp library sources.
Reviewed by: jdp, mdodd
Diffstat (limited to 'lib/libncp/sap.c')
-rw-r--r--lib/libncp/sap.c302
1 files changed, 302 insertions, 0 deletions
diff --git a/lib/libncp/sap.c b/lib/libncp/sap.c
new file mode 100644
index 0000000..32e0ab2
--- /dev/null
+++ b/lib/libncp/sap.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright (c) 1999, Boris Popov
+ * 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. All advertising materials mentioning features or use of this software
+ * must display the following acknowledgement:
+ * This product includes software developed by Boris Popov.
+ * 4. Neither the name of the author nor the names of any co-contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``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 OR CONTRIBUTORS 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.
+ *
+ * $FreeBSD$
+ *
+ */
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+#include <sys/time.h>
+#include <netipx/ipx.h>
+#include <errno.h>
+#include <unistd.h>
+#include "ipxsap.h"
+
+/*
+ * TODO: These should go to ipx headers
+ */
+#define ipx_set_net(x,y) ((x).x_net.s_net[0] = (y).x_net.s_net[0]); \
+ ((x).x_net.s_net[1]=(y).x_net.s_net[1])
+#define ipx_set_nullnet(x) ((x).x_net.s_net[0]=0); ((x).x_net.s_net[1]=0)
+#define ipx_set_nullhost(x) ((x).x_host.s_host[0] = 0); \
+ ((x).x_host.s_host[1] = 0); ((x).x_host.s_host[2] = 0)
+#define ipx_set_wildnet(x) ((x).x_net.s_net[0] = 0xFFFF); \
+ ((x).x_net.s_net[1]=0xFFFF)
+#define ipx_set_wildhost(x) ((x).x_host.s_host[0] = 0xFFFF); \
+ ((x).x_host.s_host[1] = 0xFFFF); ((x).x_host.s_host[2] = 0xFFFF);
+
+
+static struct sap_packet* sap_packet_alloc(int entries);
+static int sap_size(int entries, u_short operation);
+int (*sap_sendto_func)(void*,int,struct sockaddr_ipx*,int sock)=NULL;
+
+static int
+sap_sendto(void* buffer, int size, struct sockaddr_ipx* daddr, int sock)
+{
+ if (sap_sendto_func)
+ return sap_sendto_func(buffer,size,daddr,sock);
+ return sendto(sock, (char*)buffer, size, 0,
+ (struct sockaddr*)daddr, sizeof(*daddr));
+}
+
+static struct sap_packet*
+sap_packet_alloc(int entries)
+{
+ if (entries > IPX_SAP_MAX_ENTRIES)
+ return NULL;
+ return
+ (struct sap_packet*)malloc(sap_size(entries, IPX_SAP_GENERAL_RESPONSE));
+}
+
+static int
+sap_size(int entries, u_short operation)
+{
+ if (entries <= 0)
+ return 0;
+ switch (operation) {
+ case IPX_SAP_GENERAL_QUERY:
+ return entries == 1 ? IPX_SAP_REQUEST_LEN : 0;
+ case IPX_SAP_GENERAL_RESPONSE:
+ if (entries > IPX_SAP_MAX_ENTRIES)
+ return 0;
+ return sizeof(struct sap_packet) + (entries - 1) * sizeof(struct sap_entry);
+ case IPX_SAP_NEAREST_QUERY:
+ return entries == 1 ? IPX_SAP_REQUEST_LEN : 0;
+ case IPX_SAP_NEAREST_RESPONSE:
+ return entries == 1 ? sizeof(struct sap_packet) : 0;
+ default:
+ return 0;
+ }
+}
+
+void
+sap_copyname(char *dest, const char *src)
+{
+ bzero(dest, IPX_SAP_SERVER_NAME_LEN);
+ strncpy(dest, src, IPX_SAP_SERVER_NAME_LEN - 1);
+}
+
+int
+sap_rq_init(struct sap_rq* rq, int sock)
+{
+ rq->buffer = sap_packet_alloc(IPX_SAP_MAX_ENTRIES);
+ if (rq->buffer == NULL)
+ return 0;
+ rq->entries = 0;
+ rq->buffer->operation = htons(IPX_SAP_GENERAL_QUERY);
+ rq->dest_addr.sipx_family = AF_IPX;
+ rq->dest_addr.sipx_len = sizeof(struct sockaddr_ipx);
+ rq->sock = sock;
+ return 1;
+}
+
+int
+sap_rq_flush(struct sap_rq* rq)
+{
+ int result;
+
+ if (rq->entries == 0)
+ return 0;
+ result = sap_sendto(rq->buffer,
+ sap_size(rq->entries, ntohs(rq->buffer->operation)),
+ &rq->dest_addr, rq->sock);
+ rq->entries = 0;
+ return result;
+}
+
+void
+sap_rq_general_query(struct sap_rq* rq, u_short ser_type)
+{
+ struct sap_entry* sep;
+
+ sap_rq_flush(rq);
+ rq->buffer->operation = htons(IPX_SAP_GENERAL_QUERY);
+ sep = rq->buffer->sap_entries + rq->entries++;
+ sep->server_type = htons(ser_type);
+}
+
+void
+sap_rq_gns_request(struct sap_rq* rq, u_short ser_type)
+{
+ struct sap_entry* sep;
+
+ sap_rq_flush(rq);
+ rq->buffer->operation = htons(IPX_SAP_NEAREST_QUERY);
+ sep = rq->buffer->sap_entries + rq->entries++;
+ sep->server_type = htons(ser_type);
+}
+
+void
+sap_rq_general_response(struct sap_rq* rq,u_short type,char *name,struct sockaddr_ipx* addr, u_short hops,int down_allow)
+{
+ struct sap_entry* sep;
+
+ if (hops >= IPX_SAP_SERVER_DOWN && !down_allow) return;
+ if (rq->entries >= IPX_SAP_MAX_ENTRIES)
+ sap_rq_flush(rq);
+ if (rq->buffer->operation != htons(IPX_SAP_GENERAL_RESPONSE)){
+ sap_rq_flush(rq);
+ rq->buffer->operation = htons(IPX_SAP_GENERAL_RESPONSE);
+ }
+ sep = rq->buffer->sap_entries + rq->entries;
+ sep->server_type = htons(type);
+ sap_copyname(sep->server_name, name);
+ memcpy(&sep->ipx, &addr->sipx_addr, sizeof(struct ipx_addr));
+ sep->hops = htons(hops);
+ rq->entries++;
+}
+
+void
+sap_rq_gns_response(struct sap_rq* rq,u_short type,char *name,struct sockaddr_ipx* addr,u_short hops)
+{
+ struct sap_entry* sep;
+
+ if (hops >= IPX_SAP_SERVER_DOWN) return;
+ sap_rq_flush(rq);
+ rq->buffer->operation = htons(IPX_SAP_NEAREST_RESPONSE);
+ sep = rq->buffer->sap_entries + rq->entries;
+ sep->server_type = htons(type);
+ sap_copyname(sep->server_name, name);
+ memcpy(&sep->ipx, &addr->sipx_addr, sizeof(struct ipx_addr));
+ sep->hops = htons(hops);
+ rq->entries++;
+}
+
+void
+sap_rq_set_destination(struct sap_rq* rq,struct ipx_addr *dest)
+{
+ sap_rq_flush(rq);
+ memcpy(&rq->dest_addr.sipx_addr,dest,sizeof(struct ipx_addr));
+}
+
+int
+sap_getsock(int *rsock) {
+ struct sockaddr_ipx sap_addr;
+ int opt, sock, slen;
+
+ sock = socket(AF_IPX, SOCK_DGRAM, 0);
+ if (sock < 0)
+ return (errno);
+ slen = sizeof(sap_addr);
+ bzero(&sap_addr, slen);
+ sap_addr.sipx_family = AF_IPX;
+ sap_addr.sipx_len = slen;
+ if (bind(sock, (struct sockaddr*)&sap_addr, slen) == -1) {
+ close(sock);
+ return(errno);
+ }
+ opt = 1;
+ if (setsockopt(sock, SOL_SOCKET, SO_BROADCAST, &opt, sizeof(opt)) != 0){
+ close(sock);
+ return(errno);
+ }
+ *rsock = sock;
+ return(0);
+}
+
+static int
+sap_recv(int sock,void *buf,int len,int flags, int timeout){
+ fd_set rd, wr, ex;
+ struct timeval tv;
+ int result;
+
+ FD_ZERO(&rd);
+ FD_ZERO(&wr);
+ FD_ZERO(&ex);
+ FD_SET(sock, &rd);
+
+ tv.tv_sec = timeout;
+ tv.tv_usec = 0;
+
+ if ((result = select(sock + 1, &rd, &wr, &ex, &tv)) == -1) {
+ return result;
+ }
+ if (FD_ISSET(sock, &rd)) {
+ result = recv(sock, buf, len, flags);
+ } else {
+ errno = ETIMEDOUT;
+ result = -1;
+ }
+ return result;
+}
+
+int
+sap_find_nearest(int server_type, struct sockaddr_ipx *daddr, char *server_name)
+{
+ struct ipx_addr addr;
+ char data[1024];
+ int sock, error, packets, len;
+ struct sap_packet *reply = (struct sap_packet*)&data;
+ struct sap_rq sap_rq;
+
+ error = sap_getsock(&sock);
+ if (error)
+ return error;
+ bzero(&addr, sizeof(addr));
+ /* BAD: we should enum all ifs (and nets ?) */
+ if (ipx_iffind(NULL, &addr) != 0) {
+ return (EPROTONOSUPPORT);
+ }
+ ipx_set_wildhost(addr);
+ addr.x_port = htons(IPXPORT_SAP);
+
+ if (!sap_rq_init(&sap_rq, sock)) {
+ close(sock);
+ return(ENOMEM);
+ }
+ sap_rq_set_destination(&sap_rq, &addr);
+ sap_rq_gns_request(&sap_rq, server_type);
+ sap_rq_flush(&sap_rq);
+ packets = 5;
+ do {
+ len = sap_recv(sock, data, sizeof(data), 0, 1);
+ if (len < 66) {
+ packets++;
+ continue;
+ }
+ } while (ntohs(reply->operation) != IPX_SAP_NEAREST_RESPONSE &&
+ packets > 0);
+
+ if (packets == 0) {
+ close(sock);
+ return ENETDOWN;
+ }
+
+ daddr->sipx_addr = reply->sap_entries[0].ipx;
+ daddr->sipx_family = AF_IPX;
+ daddr->sipx_len = sizeof(struct sockaddr_ipx);
+ sap_copyname(server_name, reply->sap_entries[0].server_name);
+ errno = 0;
+ close(sock);
+ return 0;
+}
OpenPOWER on IntegriCloud