I want to implement a C program that captures all the Precision-Time-Protocol (PTP) frames on the Ethernet, so I created a raw socket and attached a filter for PTP, I use recvmsg() to read data from the socket. The first issue is that I wasn't receiving any PTP frames so I commented out the filter but now I am not getting any Ethernet frames either.
#include <stdio.h>
#include <stdbool.h>
#include <unistd.h>
#include <errno.h>
#include <string.h>
#include <sys/ioctl.h>
#include <sys/socket.h>
#include <linux/if_packet.h>
#include <linux/filter.h>
#include <net/if.h>
#include <net/ethernet.h>
#include <arpa/inet.h>
#define PTP_IF_NAME "enx503eaa33fc9d"
#define PACKET_SIZE 300
#define NUM_OF_BLOCKS_PER_LINE 30
#define SIZE_OF_BYYE_PRINT_BLOCK 3
#define SIZE_OF_PRINT_BUFFER \
( ( NUM_OF_BLOCKS_PER_LINE * SIZE_OF_BYYE_PRINT_BLOCK ) + 1 )
#define PERROR(x, ...) printf( x "stderr:%m\n", ##__VA_ARGS__)
#define INFO(x, ...) printf( x "\n", ##__VA_ARGS__)
#define ERROR(x, ...) printf( x "\n", ##__VA_ARGS__)
int eventSock = -1;
void printBuffer(unsigned char* buffer, size_t len) {
unsigned char outBuffer[SIZE_OF_PRINT_BUFFER];
unsigned char* pOutBuffer = outBuffer;
int usedBytes = 0;
for (int i = 0; i < len; i++) {
if ((usedBytes + SIZE_OF_BYYE_PRINT_BLOCK) > SIZE_OF_PRINT_BUFFER) {
INFO("%s", outBuffer);
memset(outBuffer, 0, sizeof(outBuffer));
usedBytes = 0;
pOutBuffer = outBuffer;
}
sprintf(pOutBuffer, "%02x ", buffer[i]);
usedBytes += SIZE_OF_BYYE_PRINT_BLOCK;
pOutBuffer += SIZE_OF_BYYE_PRINT_BLOCK;
}
INFO("%s", outBuffer);
}
int getInterfaceIndex(char *ifaceName) {
int sockfd;
struct ifreq ifr;
sockfd = socket(AF_INET, SOCK_DGRAM, 0);
if (sockfd < 0) {
ERROR("Could not retrieve interface index for %s", ifaceName);
return -1;
}
memset(&ifr, 0, sizeof(ifr));
strncpy(ifr.ifr_name, ifaceName, IFNAMSIZ);
if (ioctl(sockfd, SIOCGIFINDEX, &ifr) < 0) {
PERROR("failed to request hardware address for %s", ifaceName);
close(sockfd);
return -1;
}
close(sockfd);
return ifr.ifr_ifindex;
}
bool InitSocket(char *ifaceName) {
int IfIndex;
struct sockaddr_ll sll;
IfIndex = getInterfaceIndex(ifaceName);
if (IfIndex < 0) {
return false;
}
/* create the socket */
if ((eventSock = socket(AF_PACKET, SOCK_RAW, htons(ETH_P_ALL))) < 0) {
PERROR("failed to initialize raw socket");
return false;
}
/* binding to the interface */
memset(&sll, 0, sizeof(sll));
sll.sll_family = AF_PACKET;
sll.sll_ifindex = IfIndex;
sll.sll_protocol = htons(ETH_P_ALL);
/* Bind the socket to the interface */
if (bind(eventSock, (struct sockaddr *) &sll, sizeof(sll)) < 0) {
PERROR("failed to bind raw event socket");
return false;
}
/* bpf bytecode to filter only PTP packets (ethertype 0x88f7)
* obtained via "tcpdump -dd ether proto 0x88f7" */
// struct sock_filter filter_ptp[] = {
// { 0x28, 0, 0, 0x0000000c }, // ldh [12]
// { 0x15, 0, 1, 0x000088f7 }, // jeq #0x88f7 jt 2 jf 3
// { 0x06, 0, 0, 0x00040000 }, // ret #262144
// { 0x06, 0, 0, 0x00000000 }, // ret #0
// };
//
// struct sock_fprog bpf_ptp = { .len = 4, .filter = filter_ptp, };
//
// if (setsockopt(eventSock, SOL_SOCKET, SO_ATTACH_FILTER, &bpf_ptp,
// sizeof(bpf_ptp))) {
// PERROR("failed attaching filter to raw event socket");
// return false;
// }
}
int waitForData(struct timeval * timeout, fd_set *readfds) {
int ret, nfds;
struct timeval tv, *tv_ptr;
if (timeout) {
tv.tv_sec = timeout->tv_sec;
tv.tv_usec = timeout->tv_usec;
tv_ptr = &tv;
} else {
tv_ptr = NULL;
}
FD_ZERO(readfds);
nfds = 0;
FD_SET(eventSock, readfds);
nfds = eventSock;
nfds++;
ret = select(nfds, readfds, 0, 0, tv_ptr);
return ret;
}
void netr(void) {
ssize_t ret = 0;
char buf[1024];
struct msghdr msg;
struct iovec vec[1];
struct sockaddr_in from_addr;
union {
struct cmsghdr cm;
char control[256];
} cmsg_un;
struct cmsghdr *cmsg;
vec[0].iov_base = buf;
vec[0].iov_len = PACKET_SIZE;
memset(&msg, 0, sizeof(msg));
memset(&from_addr, 0, sizeof(from_addr));
memset(buf, 0, PACKET_SIZE);
memset(&cmsg_un, 0, sizeof(cmsg_un));
msg.msg_name = (caddr_t) & from_addr;
msg.msg_namelen = sizeof(from_addr);
msg.msg_iov = vec;
msg.msg_iovlen = 1;
msg.msg_control = cmsg_un.control;
msg.msg_controllen = sizeof(cmsg_un.control);
msg.msg_flags = 0;
ret = recvmsg(eventSock, &msg, MSG_DONTWAIT);
if (ret < 0) {
PERROR("failed to receive message!");
}
INFO("received Bytes: %ld", ret);
printBuffer(msg.msg_iov->iov_base, PACKET_SIZE);
}
int main(void) {
int ret;
fd_set readfds;
struct timeval timeout = { .tv_sec = 1, .tv_usec = 0 };
INFO("main started!!");
InitSocket(PTP_IF_NAME);
while (1) {
INFO("waiting for event!!");
ret = waitForData(&timeout, &readfds);
INFO("select return = %d", ret);
if (ret > 0) {
if (FD_ISSET(eventSock, &readfds)) {
netr();
}
}
}
}
It seems that nothing is wrong with your code itself. This can be verified by running your program on the localhost interface, for example, by changing only this line:
#define PTP_IF_NAME "lo"
And by sending frames, ping 127.0.0.1.
Based on the discussion in comments, you can apparently see the frames when Wireshark is running and dumping packets on the interface, but it doesn't work the rest of the time. This is likely a sign that you need to set your interface to “promiscuous mode”.
Promiscuous mode tells your card to capture all packets, even those that are not addressed to the interface (those for which the destination MAC address is not the one of the interface). When the mode is off, the card drops packets that it is not supposed to receive, no matter what BPF filter you use or whether you use a filter at all. In your case, it simply looks like you are running your test with packets that are not addressed to the NIC. When Wireshark runs it sets the interface to promiscuous, which also reflects with your program and allows you to see the frames. When you stop it, it restores the interface into non-promiscuous.
You don't have to run Wireshark to set the interface to promiscuous mode, you can do it with:
$ sudo ip link set enx503eaa33fc9d promisc on
Then disable with off once you are done. Alternatively, just sending packets with the correct destination MAC address should make them visible without changing the NIC configuration :).