In virtual network emulation environments such as PNETLab, Linux bridges (kernel bridge module) are dynamically generated and reconstructed to configure Layer-2 connectivity between virtual nodes. In compliance with the IEEE 802.1D standard, the default Linux kernel bridge implementation does not forward certain link-local control frames (such as LACP: IEEE 802.3ad, LLDP: IEEE 802.1AB) upon reception, processing them internally or dropping them instead.

Due to this behavior, when configuring Link Aggregation (LAG) or performing LLDP-based neighbor discovery between virtual routers or switches, control packets fail to reach the peer devices, resulting in an issue where links cannot be established.

sysfs Interface and Bitmask Specifications

The Linux kernel provides the group_fwd_mask interface under sysfs to control forwarding permissions for reserved multicast MAC addresses on a per-bridge-port basis.

• Configuration file path: /sys/class/net/<bridge_interface>/brif/<port_interface>/group_fwd_mask

The bit offsets and control mask values for each control protocol are as follows:

• LACP (IEEE 802.3ad): Bit offset 2 $\rightarrow$ 0x0004
• LLDP (IEEE 802.1AB): Bit offset 14 $\rightarrow$ 0x4000
• Composite forward mask (logical OR): 0x0004 | 0x4000 = 0x4004 (Decimal: 16388)

Because vnet* virtual interfaces are dynamically created and removed as labs are started and stopped in PNETLab, status monitoring and automatic application via a persistent script are required rather than a static one-shot configuration.

Creating the Automation Script

Using a shell script, periodically scan /sys/class/net/vnet*/brif/*/group_fwd_mask and write only when the current mask value does not contain 0x4004 (16388). A diff comparison against current configured values is performed to reduce unnecessary I/O writes.

cat > /usr/local/sbin/pnetlab-lacp-forward.sh <<'EOF'
#!/bin/bash

MASK=0x4004

while true; do
  for f in /sys/class/net/vnet*/brif/*/group_fwd_mask; do
    [ -f "$f" ] || continue

    old=$(cat "$f" 2>/dev/null) || continue

    # Validate hexadecimal or standard integer decimal strings
    [[ "$old" =~ ^(0[xX][0-9a-fA-F]+|[0-9]+)$ ]] || continue

    # Convert to base-10 integer and compute bitwise OR with target mask
    old_dec=$((old))
    new_dec=$((old_dec | MASK))

    # Apply configuration update only if current state diverges from target
    if (( old_dec != new_dec )); then
      echo "$new_dec" > "$f" 2>/dev/null
    fi
  done

  sleep 2
done
EOF

After creating the script, grant execution permissions.

chmod +x /usr/local/sbin/pnetlab-lacp-forward.sh

Registering the systemd Service Unit

Define a systemd unit to ensure process monitoring and automatic start at system boot.

cat > /etc/systemd/system/pnetlab-lacp-forward.service <<'EOF'
[Unit]
Description=PNETLab LACP and LLDP Auto Forwarding
After=network.target

[Service]
Type=simple
ExecStart=/usr/local/sbin/pnetlab-lacp-forward.sh
Restart=always
RestartSec=3

[Install]
WantedBy=multi-user.target
EOF

Reload and start the service unit.

systemctl daemon-reload
systemctl enable pnetlab-lacp-forward.service
systemctl restart pnetlab-lacp-forward.service

Troubleshooting

⚠️ Key error causes and countermeasures encountered during operation and configuration setup are as follows:

1. Permission denied when writing to sysfs (Permission denied)

• Cause: Occurs when the service runs under non-root privileges, or when security modules such as SELinux / AppArmor restrict unrestricted access under /sys.
• Countermeasure: Verify that the systemd execution user is root, and either allow sysfs write permissions in security profiles or exempt the corresponding process.

2. File not found error during node deletion (No such file or directory)

• Cause: If a virtual interface on PNETLab is destroyed mid-polling loop, error logs will occur during path references by cat or echo.
• Countermeasure: Prevent exception logs by checking existence via [ -f "$f" ] || continue within the script and redirecting standard error to /dev/null.

3. Increased CPU resource consumption

• Cause: Occurs when a sleep delay is not inserted into the polling loop, causing a busy loop where CPU utilization sticks at 100%.
• Countermeasure: Set an appropriate interval such as sleep 2 at the end of the script to suppress kernel I/O load.

Verification

🛠️ Command and output protocol log examples to verify service running status and normal operation.

# systemctl status pnetlab-lacp-forward.service --no-pager
● pnetlab-lacp-forward.service - PNETLab LACP and LLDP Auto Forwarding
     Loaded: loaded (/etc/systemd/system/pnetlab-lacp-forward.service; enabled; vendor preset: enabled)
     Active: active (running) since Sat 2026-10-17 10:15:32 KST; 5min ago
   Main PID: 12043 (pnetlab-lacp-f)
      Tasks: 2 (limit: 9452)
     Memory: 1.2M
     CGroup: /system.slice/pnetlab-lacp-forward.service
             ├─12043 /bin/bash /usr/local/sbin/pnetlab-lacp-forward.sh
             └─12089 sleep 2

# cat /sys/class/net/vnet0/brif/vnet0_1/group_fwd_mask
16388

When the group_fwd_mask value is automatically applied as 16388 (hexadecimal: 0x4004), pass-through processing for LACP and LLDP control frames is successfully enabled.

Operational Notes

💡 Operational Notes:

• When adding a new control protocol (e.g., EAPOL 0x0001, etc.) as a pass-through target, perform the bitmask operation 0x4004 | NEW_MASK and update the MASK variable in the script.
• When operating large-scale topologies, tuning the trade-off between I/O load and interface recognition responsiveness by adjusting the polling interval (sleep seconds) is recommended.</port_interface></bridge_interface>