RE: kernel bug? - routing table entry lost

From: Jeff Haran
Date: Wed Sep 14 2005 - 13:42:50 EST


Hmmm... Two days go by and not a single response to this (at least I
haven't seen any).

I would have thought that a mailing list named linux-net would be
interested in possible linux networking stack problems.

Is this the wrong mailing list to post something like this to?

Thanks,

Jeff Haran

> -----Original Message-----
> From: linux-net-owner@xxxxxxxxxxxxxxx
> [mailto:linux-net-owner@xxxxxxxxxxxxxxx]On Behalf Of Jeff Haran
> Sent: Monday, September 12, 2005 3:11 PM
> To: linux-net@xxxxxxxxxxxxxxx
> Subject: kernel bug? - routing table entry lost
>
>
> Hi all,
>
> I've been experimenting with the configuration of IP aliases
> on a 2.4.19 based kernel and have run across some behavior
> that I suspect is a bug, but am wondering if perhaps I am
> just doing something wrong.
>
> In summary, what I find is that if I start with an interface
> that is configured with a single IP address, then add a
> secondary IP address to that interface and then delete that
> secondary IP address so that I am now presumably back where I
> started, one of the internal routes generated by the kernel
> disappears.
>
> To get visibility into this, I wrote a program (adopted from
> some code I found on another site) that sends a RTM_GETROUTE
> message to a NETLINK_ROUTE socket and then reads and displays
> the response from the kernel. That program is called
> route_get in the logs that follow and I've copied the source
> to route_get to the end of email.
>
> The annotated console log that demonstrates the problem
> follows. I start with interfaces down and no IP addresses assigned:
>
> FD21:root> ip -s -s link show
> 1: lo: <LOOPBACK> mtu 16436 qdisc noop
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> RX: bytes packets errors dropped overrun mcast
> 0 0 0 0 0 0
> RX errors: length crc frame fifo missed
> 0 0 0 0 0
> TX: bytes packets errors dropped carrier collsns
> 0 0 0 0 0 0
> TX errors: aborted fifo window heartbeat
> 0 0 0 0
> 2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop qlen 100
> link/ether 00:60:69:90:02:7a brd ff:ff:ff:ff:ff:ff
> RX: bytes packets errors dropped overrun mcast
> 0 0 0 0 0 0
> RX errors: length crc frame fifo missed
> 0 0 0 0 0
> TX: bytes packets errors dropped carrier collsns
> 0 0 0 0 0 0
> TX errors: aborted fifo window heartbeat
> 0 0 0 0
> FD21:root> ip -s -s addr show
> 1: lo: <LOOPBACK> mtu 16436 qdisc noop
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> 2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop qlen 100
> link/ether 00:60:69:90:02:7a brd ff:ff:ff:ff:ff:ff
>
> So now I configure eth0 up, but still with no addresses:
>
> FD21:root> ip link set dev eth0 up
> FD21:root> eth0: Link status change: Link Up. 100 Mbps Half
> duplex Auto (autonegotiation complete).
>
> I add the first IP address and run my route_get program to
> see the resultant routes:
>
> FD21:root> ip address add 192.168.78.157/24 brd + dev eth0
> FD21:root> ./route_get
> nlmsg_type = 24 RTM_NEWROUTE
> nlmsg_flags = 0x0
> readSock() returned 208
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 24
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 254 RT_TABLE_MAIN
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 1 RTN_UNICAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.0
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 3 RTN_BROADCAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.255
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 254 RT_SCOPE_HOST
> rtmsg: rtm_type 2 RTN_LOCAL
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.157
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 3 RTN_BROADCAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.0
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> ============================================
> Destination Gateway Interface Source
> 192.168.78.0 *.*.*.* eth0 192.168.78.157
> 192.168.78.255 *.*.*.* eth0 192.168.78.157
> 192.168.78.157 *.*.*.* eth0 192.168.78.157
> 192.168.78.0 *.*.*.* eth0 192.168.78.157
>
> route_get displays each route in the response from the kernel
> in detail, followed by the short netstat style summary of the
> routes shown in the 4 above lines. The way I interpret the
> above routes, first to last, is as follows:
>
> 1) The unicast route to the directly connected subnet.
> 2) A broadcast route to the subnet where the broadcast
> address is the all 1s in the host number form.
> 3) The unicast route to the host's own IP address.
> 4) A broadcast route to the subnet where the broadcast
> address is the all 0s in the host number form.
>
> So now I add the alias, an IP address on the same subnet on
> the same device:
>
> FD21:root> ip address add 192.168.78.158/24 brd + dev eth0
> label eth0:0
> FD21:root> ip -s -s addr show
> 1: lo: <LOOPBACK> mtu 16436 qdisc noop
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> 2: eth0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 100
> link/ether 00:60:69:90:02:7a brd ff:ff:ff:ff:ff:ff
> inet 192.168.78.157/24 brd 192.168.78.255 scope global eth0
> inet 192.168.78.158/24 brd 192.168.78.255 scope global
> secondary eth0:0
>
> Running route_get, I see one more route has been created:
>
> FD21:root> ./route_get
> nlmsg_type = 24 RTM_NEWROUTE
> nlmsg_flags = 0x0
> readSock() returned 260
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 24
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 254 RT_TABLE_MAIN
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 1 RTN_UNICAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.0
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 254 RT_SCOPE_HOST
> rtmsg: rtm_type 2 RTN_LOCAL
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.158
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 3 RTN_BROADCAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.255
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 254 RT_SCOPE_HOST
> rtmsg: rtm_type 2 RTN_LOCAL
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.157
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 3 RTN_BROADCAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.0
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> ============================================
> Destination Gateway Interface Source
> 192.168.78.0 *.*.*.* eth0 192.168.78.157
> 192.168.78.158 *.*.*.* eth0 192.168.78.157
> 192.168.78.255 *.*.*.* eth0 192.168.78.157
> 192.168.78.157 *.*.*.* eth0 192.168.78.157
> 192.168.78.0 *.*.*.* eth0 192.168.78.157
>
> The new route (the second listed) is the unicast route to the
> additional IP address. All the other routes are the same.
> Near as I can tell so far, all this is correct.
>
> Now I delete the secondary IP address (the alias). So now I
> am back to the single IP address and I would think should be
> in the same state as before I added the alias.
>
> FD21:root> ip address delete 192.168.78.158 dev eth0 label eth0:0
> FD21:root> ip -s -s addr show
> 1: lo: <LOOPBACK> mtu 16436 qdisc noop
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> 2: eth0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 100
> link/ether 00:60:69:90:02:7a brd ff:ff:ff:ff:ff:ff
> inet 192.168.78.157/24 brd 192.168.78.255 scope global eth0
>
> So now I run route_get to see my routes:
>
> FD21:root> ./route_get
> nlmsg_type = 24 RTM_NEWROUTE
> nlmsg_flags = 0x0
> readSock() returned 156
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 24
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 254 RT_TABLE_MAIN
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 1 RTN_UNICAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.0
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 253 RT_SCOPE_LINK
> rtmsg: rtm_type 3 RTN_BROADCAST
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.255
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> =============================================
> rtmsg: rtm_family 2 AF_INET
> rtmsg: rtm_dst_len 32
> rtmsg: rtm_src_len 0
> rtmsg: rtm_tos 0
> rtmsg: rtm_table 255 RT_TABLE_LOCAL
> rtmsg: rtm_protocol 2 RTPROT_KERNEL
> rtmsg: rtm_scope 254 RT_SCOPE_HOST
> rtmsg: rtm_type 2 RTN_LOCAL
> rtmsg: rtm_flags 0
> RTA_DST: 192.168.78.157
> RTA_PREFSRC: 192.168.78.157
> RTA_OIF: eth0
> ============================================
> Destination Gateway Interface Source
> 192.168.78.0 *.*.*.* eth0 192.168.78.157
> 192.168.78.255 *.*.*.* eth0 192.168.78.157
> 192.168.78.157 *.*.*.* eth0 192.168.78.157
> FD21:root>
>
> It seems like I am now missing a route, specifically number
> (4) from the list above:
>
> 4) A broadcast route to the subnet where the broadcast
> address is the all 0s in the host number form.
>
> Is this a feature or a bug (or is this a figment of route_get)?
> If this is a kernel bug, has it been previously reported?
>
> Thanks,
>
> Jeff Haran
> Brocade Communications Systems
>
> p.s. The source to route_get is copied below:
>
> /* 8-04-2005, jharan,
>
> stolen from:
> http://mail.nl.linux.org/kernelnewbies/2003-12/msg00139.html
>
> Hi all,
> i am including another piece of code which i have written to
> print the
> routes
> in the kernel routing table using rtnetlink. Please do let me
> know if there
> is a better way of doing this.
>
> */
>
> #include <asm/types.h>
> #include <netinet/ether.h>
> #include <netinet/in.h>
> #include <net/if.h>
> #include <stdio.h>
> #include <sys/socket.h>
> #include <sys/ioctl.h>
> #include <linux/netlink.h>
> #include <linux/rtnetlink.h>
> #include <sys/types.h>
> #include <errno.h>
> #include <unistd.h>
> #define _GNU_SOURCE
> #include <getopt.h>
> #include <string.h>
> #include <arpa/inet.h>
>
> #define NL_BUFSIZE 8192
>
> #ifndef IF_NAMESIZE
> #define IF_NAMESIZE 16
> #endif
>
> struct route_info{
> struct route_info *next;
> struct in_addr dstAddr;
> struct in_addr srcAddr;
> struct in_addr gateWay;
> char ifName[IF_NAMESIZE + 50];
> };
>
> struct route_info *route_info_head = 0;
> struct route_info *route_info_tail = 0;
>
> typedef struct {
> __u16 nlmsg_type;
> char *name;
> } nlmsg_type_name;
>
> #define NLMSG_NAME_INIT(name) { name, #name }
>
> nlmsg_type_name nlmsg_type_table[] = {
> NLMSG_NAME_INIT(NLMSG_NOOP),
> NLMSG_NAME_INIT(NLMSG_ERROR),
> NLMSG_NAME_INIT(NLMSG_DONE),
> NLMSG_NAME_INIT(NLMSG_OVERRUN),
> NLMSG_NAME_INIT(RTM_NEWLINK),
> NLMSG_NAME_INIT(RTM_DELLINK),
> NLMSG_NAME_INIT(RTM_GETLINK),
> NLMSG_NAME_INIT(RTM_NEWADDR),
> NLMSG_NAME_INIT(RTM_DELADDR),
> NLMSG_NAME_INIT(RTM_GETADDR),
> NLMSG_NAME_INIT(RTM_NEWROUTE),
> NLMSG_NAME_INIT(RTM_DELROUTE),
> NLMSG_NAME_INIT(RTM_GETROUTE),
> NLMSG_NAME_INIT(RTM_NEWNEIGH),
> NLMSG_NAME_INIT(RTM_DELNEIGH),
> NLMSG_NAME_INIT(RTM_GETNEIGH),
> NLMSG_NAME_INIT(RTM_NEWRULE),
> NLMSG_NAME_INIT(RTM_DELRULE),
> NLMSG_NAME_INIT(RTM_GETRULE),
> NLMSG_NAME_INIT(RTM_NEWQDISC),
> NLMSG_NAME_INIT(RTM_DELQDISC),
> NLMSG_NAME_INIT(RTM_GETQDISC),
> NLMSG_NAME_INIT(RTM_NEWTCLASS),
> NLMSG_NAME_INIT(RTM_DELTCLASS),
> NLMSG_NAME_INIT(RTM_GETTCLASS),
> NLMSG_NAME_INIT(RTM_NEWTFILTER),
> NLMSG_NAME_INIT(RTM_DELTFILTER),
> NLMSG_NAME_INIT(RTM_GETTFILTER),
> NLMSG_NAME_INIT(RTM_MAX),
> { 0, 0 } /* end of table marker */
> };
>
> char *lookup_nlmsg_type(__u16 nlmsg_type)
> {
> int i;
>
> for (i = 0; nlmsg_type_table[i].name; ++i) {
> if (nlmsg_type_table[i].nlmsg_type == nlmsg_type)
> return nlmsg_type_table[i].name;
> }
> return ("unknown nlmsg_type");
> }
>
> typedef struct {
> int nlmsg_flag;
> char *name;
> } nlmsg_flags_name;
>
> #define NLMSG_FLAGS_INIT(name) { name, #name }
>
> nlmsg_flags_name nlmsg_flags_table[] = {
> NLMSG_FLAGS_INIT(NLM_F_REQUEST),
> NLMSG_FLAGS_INIT(NLM_F_MULTI),
> NLMSG_FLAGS_INIT(NLM_F_ACK),
> NLMSG_FLAGS_INIT(NLM_F_ECHO),
> NLMSG_FLAGS_INIT(NLM_F_ROOT),
> NLMSG_FLAGS_INIT(NLM_F_MATCH),
> NLMSG_FLAGS_INIT(NLM_F_ATOMIC),
> NLMSG_FLAGS_INIT(NLM_F_REPLACE),
> NLMSG_FLAGS_INIT(NLM_F_EXCL),
> NLMSG_FLAGS_INIT(NLM_F_CREATE),
> NLMSG_FLAGS_INIT(NLM_F_APPEND),
> { 0, 0 } /* end of table marker */
> };
>
> __u16 print_nlmsg_flags(__u16 nlmsg_flags)
> {
> int i;
>
> /* while some flag bits are still set */
> while (nlmsg_flags) {
> /* search table for a flag where we know what it is */
> for (i = 0; nlmsg_flags_table[i].name; ++i) {
> /* known flag? */
> if (nlmsg_flags_table[i].nlmsg_flag &
> nlmsg_flags) {
> /* print its name */
> fprintf(stdout, "%s ",
> nlmsg_flags_table[i].name);
> /* reset the flag */
> nlmsg_flags &=
> ~nlmsg_flags_table[i].nlmsg_flag;
> break;
> }
> }
> /* if we couldn't find any matching flags, bail */
> if (!nlmsg_flags_table[i].name)
> break;
> }
> /* if any unknown flags left, display them numerically */
> if (nlmsg_flags)
> fprintf(stdout, "mystery flags 0x%x", nlmsg_flags);
> return nlmsg_flags;
> }
>
> typedef struct {
> int rtmsg_flag;
> char *name;
> } rtmsg_flags_name;
>
> #define RTMSG_FLAGS_INIT(name) { name, #name }
>
> rtmsg_flags_name rtmsg_flags_table[] = {
> RTMSG_FLAGS_INIT(RTM_F_NOTIFY),
> RTMSG_FLAGS_INIT(RTM_F_CLONED),
> RTMSG_FLAGS_INIT(RTM_F_EQUALIZE),
> { 0, 0 } /* end of table marker */
> };
>
> unsigned int print_rtmsg_flags(unsigned int rtmsg_flags)
> {
> int i;
>
> /* while some flag bits are still set */
> while (rtmsg_flags) {
> /* search table for a flag where we know what it is */
> for (i = 0; rtmsg_flags_table[i].name; ++i) {
> /* known flag? */
> if (rtmsg_flags_table[i].rtmsg_flag &
> rtmsg_flags) {
> /* print its name */
> fprintf(stdout, "%s ",
> rtmsg_flags_table[i].name);
> /* reset the flag */
> rtmsg_flags &=
> ~rtmsg_flags_table[i].rtmsg_flag;
> break;
> }
> }
> /* if we couldn't find any matching flags, bail */
> if (!rtmsg_flags_table[i].name)
> break;
> }
> /* if any unknown flags left, display them numerically */
> if (rtmsg_flags)
> fprintf(stdout, "mystery flags 0x%x", rtmsg_flags);
> return rtmsg_flags;
> }
>
> int readSock(int sockFd, char *bufPtr, int seqNum, int pId)
> {
> struct nlmsghdr *nlHdr;
> int readLen = 0, flag = 0, msgLen = 0;
>
> do{
> /* Recieve response from kernel */
> if ((readLen = recv(sockFd, bufPtr, NL_BUFSIZE
> - msgLen, 0)) < 0){
> perror("SOCK READ: ");
> return -1;
> }
> nlHdr = (struct nlmsghdr *)bufPtr;
>
> fprintf(stdout, "nlmsg_type = %d %s\n",
> nlHdr->nlmsg_type,
> lookup_nlmsg_type(nlHdr->nlmsg_type));
> fprintf(stdout, "nlmsg_flags = 0x%x ",
> nlHdr->nlmsg_flags);
> print_nlmsg_flags(nlHdr->nlmsg_flags);
> fprintf(stdout, "\n");
>
> /* Check if header is valid */
> if (NLMSG_OK(nlHdr, readLen) == 0) {
> fprintf(stdout, "NLMSG_OK == 0\n");
> return -1;
> }
>
> /* check for error packet */
> if (nlHdr->nlmsg_type == NLMSG_ERROR) {
> struct nlmsgerr *errHdr;
>
> errHdr = (struct nlmsgerr *) NLMSG_DATA(nlHdr);
> fprintf(stdout,
> "Received NLMSG_ERROR, error = %d %s\n",
> errHdr->error,
> strerror(-(errHdr->error)));
> return -1;
> }
>
> /* Check if its the last message */
> if (nlHdr->nlmsg_type == NLMSG_DONE) {
> flag = 1;
> break;
> } else {
> /* Move the buffer pointer appropriately */
> bufPtr += readLen;
> msgLen += readLen;
> }
> #if 1
> /* jharan, though it seems to work, not clear to me
> that the following
> is "correct". I think we should only be looking
> for the NLMSG_DONE
> to terminate, like is done above */
> if ((nlHdr->nlmsg_flags & NLM_F_MULTI) == 0){
> flag = 1;
> break;
> }
> #endif
> } while ((nlHdr->nlmsg_seq != seqNum) ||
> (nlHdr->nlmsg_pid != pId) || (flag == 0));
> return msgLen;
> }
>
> /* For printing the routes. */
> void printRoute(struct route_info *rtInfo)
> {
> char tempBuf[512];
>
> /* Print Destination address */
> if(rtInfo->dstAddr.s_addr != 0)
> strcpy(tempBuf, (char *)inet_ntoa((struct
> in_addr) (rtInfo->dstAddr)));
> else
> sprintf(tempBuf,"*.*.*.*\t");
> fprintf(stdout,"%s\t", tempBuf);
>
> /* Print Gateway address */
> if(rtInfo->gateWay.s_addr != 0)
> strcpy(tempBuf, (char *)inet_ntoa((struct
> in_addr) (rtInfo->gateWay)));
> else
> sprintf(tempBuf,"*.*.*.*\t");
> fprintf(stdout,"%s\t", tempBuf);
>
> /* Print Interface Name*/
> fprintf(stdout,"%s\t\t", rtInfo->ifName);
>
> /* Print Source address */
> if(rtInfo->srcAddr.s_addr != 0)
> strcpy(tempBuf, (char *)inet_ntoa((struct
> in_addr) (rtInfo->srcAddr)));
> else
> sprintf(tempBuf,"*.*.*.*\t");
> fprintf(stdout,"%s\n", tempBuf);
> }
>
> /* For parsing the route info returned */
> void parseRoutes(struct nlmsghdr *nlHdr, struct route_info *rtInfo)
> {
> struct rtmsg *rtMsg;
> struct rtattr *rtAttr;
> int rtLen;
> char tempBuf[100];
>
> rtMsg = (struct rtmsg *)NLMSG_DATA(nlHdr);
>
> fprintf(stdout,
> "=============================================\n");
> fprintf(stdout, "rtmsg: rtm_family %d\t", rtMsg->rtm_family);
> if (rtMsg->rtm_family == AF_INET)
> fprintf(stdout, "AF_INET\n");
> else
> fprintf(stdout, "mystery\n");
>
> /* from the data, rtm_dst_len appears the be the number of bits
> set in the subnet mask, e.g. the 8 in 10.0.0.1/8.
> in case of rtm_dst_len == 32, looks like a
> host route */
> fprintf(stdout, "rtmsg: rtm_dst_len %d\n", rtMsg->rtm_dst_len);
> /* rtm_src_len always seems to be 0, not sure what it means */
> fprintf(stdout, "rtmsg: rtm_src_len %d\n", rtMsg->rtm_src_len);
> fprintf(stdout, "rtmsg: rtm_tos %d\n", rtMsg->rtm_tos);
> fprintf(stdout, "rtmsg: rtm_table %d\t", rtMsg->rtm_table);
>
> if (rtMsg->rtm_table == RT_TABLE_UNSPEC)
> fprintf(stdout, "RT_TABLE_UNSPEC\n");
> else if (rtMsg->rtm_table == RT_TABLE_DEFAULT)
> fprintf(stdout, "RT_TABLE_DEFAULT\n");
> else if (rtMsg->rtm_table == RT_TABLE_MAIN)
> fprintf(stdout, "RT_TABLE_MAIN\n");
> else if (rtMsg->rtm_table == RT_TABLE_LOCAL)
> fprintf(stdout, "RT_TABLE_LOCAL\n");
> else
> fprintf(stdout, "mystery\n");
>
> fprintf(stdout, "rtmsg: rtm_protocol %d\t",
> rtMsg->rtm_protocol);
>
> if (rtMsg->rtm_protocol == RTPROT_REDIRECT)
> fprintf(stdout, "RTPROT_REDIRECT\n");
> else if (rtMsg->rtm_protocol == RTPROT_KERNEL)
> fprintf(stdout, "RTPROT_KERNEL\n");
> else if (rtMsg->rtm_protocol == RTPROT_BOOT)
> fprintf(stdout, "RTPROT_BOOT\n");
> else if (rtMsg->rtm_protocol == RTPROT_STATIC)
> fprintf(stdout, "RTPROT_STATIC\n");
> else
> /* there's more of these in rtnetlink.h,
> just handling the common ones for now */
> fprintf(stdout, "mystery\n");
>
> fprintf(stdout, "rtmsg: rtm_scope %d\t", rtMsg->rtm_scope);
>
> if (rtMsg->rtm_scope == RT_SCOPE_UNIVERSE)
> fprintf(stdout, "RT_SCOPE_UNIVERSE\n");
> else if (rtMsg->rtm_scope == RT_SCOPE_SITE)
> fprintf(stdout, "RT_SCOPE_SITE\n");
> else if (rtMsg->rtm_scope == RT_SCOPE_LINK)
> fprintf(stdout, "RT_SCOPE_LINK\n");
> else if (rtMsg->rtm_scope == RT_SCOPE_HOST)
> fprintf(stdout, "RT_SCOPE_HOST\n");
> else if (rtMsg->rtm_scope == RT_SCOPE_NOWHERE)
> fprintf(stdout, "RT_SCOPE_NOWHERE\n");
> else
> fprintf(stdout, "mystery\n");
>
> fprintf(stdout, "rtmsg: rtm_type %d\t", rtMsg->rtm_type);
>
> if (rtMsg->rtm_type == RTN_UNSPEC)
> fprintf(stdout, "RTN_UNSPEC\n");
> else if (rtMsg->rtm_type == RTN_UNICAST)
> fprintf(stdout, "RTN_UNICAST\n");
> else if (rtMsg->rtm_type == RTN_LOCAL)
> fprintf(stdout, "RTN_LOCAL\n");
> else if (rtMsg->rtm_type == RTN_BROADCAST)
> fprintf(stdout, "RTN_BROADCAST\n");
> else if (rtMsg->rtm_type == RTN_ANYCAST)
> fprintf(stdout, "RTN_ANYCAST\n");
> else if (rtMsg->rtm_type == RTN_MULTICAST)
> fprintf(stdout, "RTN_MULTICAST\n");
> else if (rtMsg->rtm_type == RTN_BLACKHOLE)
> fprintf(stdout, "RTN_BLACKHOLE\n");
> else if (rtMsg->rtm_type == RTN_UNREACHABLE)
> fprintf(stdout, "RTN_UNREACHABLE\n");
> else if (rtMsg->rtm_type == RTN_PROHIBIT)
> fprintf(stdout, "RTN_PROHIBIT\n");
> else if (rtMsg->rtm_type == RTN_THROW)
> fprintf(stdout, "RTN_THROW\n");
> else if (rtMsg->rtm_type == RTN_NAT)
> fprintf(stdout, "RTN_NAT\n");
> else if (rtMsg->rtm_type == RTN_XRESOLVE)
> fprintf(stdout, "RTN_XRESOLVE\n");
> else
> fprintf(stdout, "mystery\n");
>
> fprintf(stdout, "rtmsg: rtm_flags %d ", rtMsg->rtm_flags);
> print_rtmsg_flags(rtMsg->rtm_flags);
> fprintf(stdout, "\n");
>
> #if 0
> /* If the route is not for AF_INET or does not belong
> to main routing table
> then return. */
> if((rtMsg->rtm_family != AF_INET) || (rtMsg->rtm_table
> != RT_TABLE_MAIN))
> return;
> #endif
>
> /* get the rtattr field */
> rtAttr = (struct rtattr *)RTM_RTA(rtMsg);
> rtLen = RTM_PAYLOAD(nlHdr);
> for(;RTA_OK(rtAttr,rtLen);rtAttr = RTA_NEXT(rtAttr,rtLen)){
> switch(rtAttr->rta_type) {
> case RTA_DST:
> inet_ntop(AF_INET, RTA_DATA(rtAttr),
> tempBuf, sizeof(tempBuf));
> fprintf(stdout, "RTA_DST: %s\n",tempBuf);
> rtInfo->dstAddr.s_addr= *(u_int
> *)RTA_DATA(rtAttr);
> break;
> case RTA_SRC:
> inet_ntop(AF_INET, RTA_DATA(rtAttr),
> tempBuf, sizeof(tempBuf));
> fprintf(stdout, "RTA_SRC: %s\n",tempBuf);
> break;
> case RTA_IIF:
> if (!if_indextoname(*(int
> *)RTA_DATA(rtAttr), tempBuf))
> sprintf(tempBuf, "errno %d", errno);
> fprintf(stdout, "RTA_IIF: %s\n", tempBuf);
> break;
> case RTA_OIF:
> if (!if_indextoname(*(int
> *)RTA_DATA(rtAttr), tempBuf))
> sprintf(tempBuf, "errno %d", errno);
> fprintf(stdout, "RTA_OIF: %s\n", tempBuf);
> strcpy(rtInfo->ifName, tempBuf);
> break;
> case RTA_GATEWAY:
> inet_ntop(AF_INET, RTA_DATA(rtAttr),
> tempBuf, sizeof(tempBuf));
> fprintf(stdout, "RTA_GATEWAY: %s\n",tempBuf);
> rtInfo->gateWay.s_addr = *(u_int
> *)RTA_DATA(rtAttr);
> break;
> case RTA_PREFSRC:
> inet_ntop(AF_INET, RTA_DATA(rtAttr),
> tempBuf, sizeof(tempBuf));
> fprintf(stdout, "RTA_PREFSRC: %s\n",tempBuf);
> rtInfo->srcAddr.s_addr = *(u_int
> *)RTA_DATA(rtAttr);
> break;
> default:
> /* there's more RTA_s under
> rtattr_type_t in rtnetlink.h,
> just covering some common ones above */
> fprintf(stdout, "RTA mystery: %d len %d\n",
> rtAttr->rta_type, rtAttr->rta_len);
> break;
> }
> }
> return;
> }
>
> unsigned char command_rtm_table;
> unsigned char command_rtm_protocol;
> unsigned char command_rtm_family;
> int command_ack;
> int command_any_args;
>
> struct option long_options[] = {
> { "table", required_argument, 0, 't'},
> { "protocol", required_argument, 0, 'p'},
> { "family", required_argument, 0, 'f' },
> { "ack", no_argument, 0, 'a'},
> { 0, 0, 0, 0}
> };
>
> char short_options[] = "t:p:f:a";
>
> int parse_command_line(int argc, char *argv[])
> {
> int rc;
> int option_index;
> unsigned int x;
>
> while (1)
> {
> rc = getopt_long(argc, argv, short_options,
> long_options, &option_index);
> switch (rc) {
> case 't' :
> sscanf(optarg, "%u", &x);
> command_rtm_table = (unsigned char) x;
> printf("command_rtm_table =
> %u\n", command_rtm_table);
> command_any_args++;
> break;
> case 'p' :
> sscanf(optarg, "%u", &x);
> command_rtm_protocol =
> (unsigned char) x;
> printf("command_rtm_protcol =
> %u\n", command_rtm_protocol);
> command_any_args++;
> break;
> case 'f' :
> sscanf(optarg, "%u", &x);
> command_rtm_family = (unsigned char) x;
> printf("command_rtm_family =
> %u\n", command_rtm_family);
> command_any_args++;
> break;
> case 'a' :
> command_ack = 1;
> printf("acknowledgement enabled\n");
> break;
> case -1 :
> break;
> default :
> printf("unrecognized option %c
> (%d)\n", rc, rc);
> break;
> }
>
> if (rc == -1)
> break;
> }
>
> return 0;
> }
>
> int main(int argc, char *argv[])
> {
> int sock, len, msgSeq = 0;
> struct nlmsghdr *nlMsg;
> struct rtmsg *rtMsg;
> struct route_info *rtInfo;
> char msgBuf[NL_BUFSIZE];
>
> if (parse_command_line(argc, argv))
> return -1;
>
> /* Create Socket */
> if((sock = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_ROUTE)) < 0)
> perror("Socket Creation: ");
>
> /* Initialize the buffer */
> memset(msgBuf, 0, NL_BUFSIZE);
>
> /* point the header and the msg structure pointers into
> the buffer */
> nlMsg = (struct nlmsghdr *)msgBuf;
> rtMsg = (struct rtmsg *)NLMSG_DATA(nlMsg);
>
> rtMsg->rtm_table = command_rtm_table;
> rtMsg->rtm_protocol = command_rtm_protocol;
> rtMsg->rtm_family = command_rtm_family;
>
> /* Fill in the nlmsg header*/
> // Length of message.
> nlMsg->nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
> // Get the routes from kernel routing table .
> nlMsg->nlmsg_type = RTM_GETROUTE;
>
>
> nlMsg->nlmsg_flags = NLM_F_REQUEST;
> #if 0
> /* it appears that only NLM_F_REQUEST | NLM_F_ROOT is supported
> by the RTM_GETROUTE service. when I attempt to setup the
> fields of ifinfomsg sent to the kernel as
> above, I either
> get all routes back when NLM_F_ROOT is specified or an
> error message when its not, jharan, 8-08-2005 */
> if (!command_any_args)
> #endif
> nlMsg->nlmsg_flags |= NLM_F_ROOT;
>
> /* it appears the NLM_F_ACK has no effect on gets. guess this
> makes sense given that the get response is sufficent
> acknowledgement, jharan, 8-08-2005 */
> if (command_ack)
> nlMsg->nlmsg_flags |= NLM_F_ACK;
>
> // Sequence of the message packet.
> nlMsg->nlmsg_seq = msgSeq++;
> // PID of process sending the request.
> nlMsg->nlmsg_pid = getpid();
>
> /* Send the request */
> if(send(sock, nlMsg, nlMsg->nlmsg_len, 0) < 0){
> printf("Write To Socket Failed...\n");
> return -1;
> }
>
> /* Read the response */
> if((len = readSock(sock, msgBuf, msgSeq, getpid())) <
> 0) {
> printf("Read From Socket Failed...\n");
> return -1;
> }
>
> printf("readSock() returned %d\n", len);
>
> /* Parse and print the response */
> for(;NLMSG_OK(nlMsg,len);nlMsg = NLMSG_NEXT(nlMsg,len)){
> rtInfo = (struct route_info *)
> malloc(sizeof(struct route_info));
> if (!rtInfo) {
> fprintf(stdout, "malloc() failed\n");
> return -1;
> }
> memset(rtInfo, 0, sizeof(struct route_info));
> parseRoutes(nlMsg, rtInfo);
> if (!route_info_head) {
> route_info_head = rtInfo;
> route_info_tail = rtInfo;
> } else {
> route_info_tail->next = rtInfo;
> route_info_tail = rtInfo;
> }
> }
> close(sock);
> /* netstat style display of accumulated data */
> fprintf(stdout,
> "============================================\n");
> fprintf(stdout, "Destination\tGateway\t\tInterface\tSource\n");
> for (rtInfo = route_info_head; rtInfo; rtInfo = rtInfo->next)
> printRoute(rtInfo);
> return 0;
> }
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-net" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html