odhcpd: fix infinite loop on recvmsg error

[pandax381]

Description

This PR fixes an issue where odhcpd can enter an infinite loop and consume 100% CPU when recvmsg() encounters an error other than EAGAIN (such as EBADF).

Background

We observed that the odhcpd process was pinning a specific CPU core at 100% utilization.

Investigating the running process with strace revealed that recvmsg() was being called continuously in a tight loop, relentlessly returning EBADF (Bad file descriptor) because the file descriptor had become invalid (-1).

strace -f -c output:

% time     seconds  usecs/call     calls    errors syscall
------ ----------- ----------- --------- --------- ----------------
100.00    1.069867           1    762638    762638 recvmsg
------ ----------- ----------- --------- --------- ----------------
100.00    1.069867           1    762638    762638 total

strace -f -e trace=recvmsg output:

recvmsg(-1, {msg_namelen=28}, MSG_DONTWAIT) = -1 EBADF (Bad file descriptor)
recvmsg(-1, {msg_namelen=28}, MSG_DONTWAIT) = -1 EBADF (Bad file descriptor)
recvmsg(-1, {msg_namelen=28}, MSG_DONTWAIT) = -1 EBADF (Bad file descriptor)
...

The root cause was that when recvmsg() failed with a non-EAGAIN error, the loop inside odhcpd_receive_packets() would continue indefinitely instead of breaking out.

Changes

• Modify the error handling in odhcpd_receive_packets() to also break out of the loop when a non-EAGAIN error occurs.
• Add unexpected recvmsg failure logging via error().
• Handle EINTR by continuing the loop to prevent premature termination on signal interrupts.

[pandax381]

@Noltari Sorry to ping you directly — would you have a moment to review this when you get a chance? Happy to make any changes if needed. Thanks!

[pandax381]

@hauke Sorry to reach out directly. Would you have a moment to take a look at this, or point me to the right person to review it? Happy to make any changes if needed. Thanks!