Are you building on arm64 host? I remember I had issues with compiling the xdp related packages. Since I’m running as dumb AP, I just disabled them all.
I have VMWare workstation on Windows 11, and i installed Ubuntu 22.04 to build the firmware, using pesa's configuration file.
I will try your suggestion and disable the package, maybe it will work.
Thanks!
I had to disable three packages for it go forward:
# CONFIG_PACKAGE_xdp-filter is not set
# CONFIG_PACKAGE_xdp-loader is not set
# CONFIG_PACKAGE_xdpdump is not set
Anyone seeing this error:
daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
I get it on reboot in the system logs. Running MT-6000 Pesa build "OpenWrt SNAPSHOT r32593-3500ccc9f3" (Nov-2025).
abridged systemlog for hostapd
Mon Feb 2 16:18:28 2026 daemon.notice hostapd: Set MLD config:
Mon Feb 2 16:18:28 2026 daemon.notice hostapd: Reload all interfaces
Mon Feb 2 16:18:28 2026 daemon.notice hostapd: Set new config for phy phy0:
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: Set new config for phy phy0: /var/run/hostapd-phy0.conf
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: Restart interface for phy phy0
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: Configuration file: (phy phy0) --> new PHY
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: Configuration file: Reading configuration file ''
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: Configuration file: Reading configuration file ''
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: ACS: Automatic channel selection started, this may take a bit
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: phy0-ap0: interface state UNINITIALIZED->ACS
Mon Feb 2 16:18:29 2026 daemon.notice hostapd: phy0-ap0: ACS-STARTED
Mon Feb 2 16:18:30 2026 daemon.notice hostapd: Set new config for phy phy1:
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: Set new config for phy phy1: /var/run/hostapd-phy1.conf
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: Restart interface for phy phy1
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: Configuration file: (phy phy1) --> new PHY
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: Configuration file: Reading configuration file ''
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: Configuration file: Reading configuration file ''
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: ACS: Automatic channel selection started, this may take a bit
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: phy1-ap0: interface state UNINITIALIZED->ACS
Mon Feb 2 16:18:31 2026 daemon.notice hostapd: phy1-ap0: ACS-STARTED
Mon Feb 2 16:18:34 2026 daemon.notice hostapd: phy1-ap0: ACS-COMPLETED freq=5180 channel=36
Mon Feb 2 16:18:34 2026 daemon.notice hostapd: phy1-ap0: interface state ACS->HT_SCAN
Mon Feb 2 16:18:35 2026 daemon.notice hostapd: phy1-ap0: interface state HT_SCAN->ENABLED
Mon Feb 2 16:18:35 2026 daemon.notice hostapd: phy1-ap0: AP-ENABLED
Mon Feb 2 16:18:35 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:35 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:37 2026 daemon.notice hostapd: phy0-ap0: ACS-COMPLETED freq=2447 channel=8
Mon Feb 2 16:18:38 2026 daemon.notice hostapd: phy0-ap0: interface state ACS->ENABLED
Mon Feb 2 16:18:38 2026 daemon.notice hostapd: phy0-ap0: AP-ENABLED
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:38 2026 daemon.err hostapd: VLAN: vlan_event_receive: recvfrom failed: No buffer space available
Mon Feb 2 16:18:39 2026 daemon.info hostapd: phy0-ap3: STA REDACTED IEEE 802.11: authenticated
Mon Feb 2 16:18:39 2026 daemon.info hostapd: phy0-ap3: STA REDACTED IEEE 802.11: associated (aid 1)
Mon Feb 2 16:18:39 2026 daemon.notice hostapd: phy0-ap3: AP-STA-CONNECTED REDACTED auth_alg=open
Mon Feb 2 16:18:39 2026 daemon.info hostapd: phy0-ap3: STA REDACTED RADIUS: starting accounting session A4407CE6127518AA
Mon Feb 2 16:18:39 2026 daemon.info hostapd: phy0-ap3: STA REDACTED WPA: pairwise key handshake completed (RSN)
Mon Feb 2 16:18:39 2026 daemon.notice hostapd: phy0-ap3: EAPOL-4WAY-HS-COMPLETED REDACTED
Mon Feb 2 16:18:40 2026 daemon.info hostapd: phy0-ap4: STA e8:db:84:c5:b0:08 IEEE 802.11: authenticated
Mon Feb 2 16:18:40 2026 daemon.info hostapd: phy0-ap4: STA e8:db:84:c5:b0:08 IEEE 802.11: associated (aid 1)
Not on mine.. although I do my own builds but it is still based almost 100% from pesa's repo.. just minor changes on very very few addon packages
I'm using VMWare Fusion with macos as host and ubuntu 22.04 as the guest.. never had issues based on the config from pesa
Has Anyone else had to completely disable ipv6 to avoid their internet dropping out constantly? My ISP doesn't support IPv6. Since disabling it, not even a glitch.
Well, I just flashed vanilla OpenWRT v25.12.rc4 sysupgrade then re-applied my same config files and I'm not seeing the vlan_event_receive: recvfrom failed: No buffer space available error any more.
I'll run it a while and see if it stays stable before messing with recommended tweaks for MT6000 to see if it returns.
I'm back, sorry for delay, busy days at work.
4.7.2 with latest commit just flashed, if it is ok on my side on the weekend will be online. Sorry for delay
11 Likes
Welcome back! 
HI @pesa1234 sorry I know you're busy but can you update latest luci and packages if you have extra time? hehehe
stangri just released an updated PBR version.. thanks in advance! well unless your PBR version is more latest, I know you pull directly from his repo from time to time
Refer to this:
1 Like
You can always update yourself directly from Stangri's repo:
1 Like
upload new version 2026-02-07_r33756-49c9934b50_next-r4.7.2.rss.mtk:
changelog:
+1163 commits
- Linux 6.12 series bumped to 6.12.69.
- mac80211 updated from 6.16 to 6.18.7.
- mt76 synced/updated with extra MediaTek patching.
- hostapd gained ubus enhancements, DPP API support, and AP+STA/DFS stability fixes.
- linux-firmware updated to 20260204 in package/firmware/linux-firmware/Makefile.
- wireguard-tools reload path improved (no teardown), renew handling, and address-change detection.
- netifd improvements: loglevel default/fallback handling, stable DHCPv4 IAID, DHCP option behavior fixes.
- procd gained reload_config support for UCI overlay directory.
- apk package manager updated across several versions (including 3.0.x line updates and fixes).
- Many MediaTek Filogic device additions/fixes.
15 Likes
Thank you, Sir Pesa.
There appears to be a new? option asking to configure attended sysupgrade, which I set to off.
The new version is working great as usual for me!
Not sure if you ever identified or mitigated your issues. I have had performance/stability degradations when on VPN and downloading bigger files or when in an RDP session.
Noticed it only occurring on AX.
Via LAN or on AC no issues at all.
All good here! Wi-Fi rock solid and even those weird ipv6 storms disappeared. Great work and thank you very much!
this would be incredible to add
1 Like
This further reduces the load on cores while maintaining piece of cake performance. Caveat: cannot use dual-dsthost or triple-isolate keywords on ingress device.
Tested on r4.7.2
SFO enabled
make sure to run ./update-available-qdiscs so it appears in luci.
Should improve Wi-Fi bandwidth as well during a speedtest. Everything outside of this is stock sqm and no custom smp affinity or rps, no extra packages.
# Multi-Queue CAKE Shaper
# Distributes ingress shaping across multiple CPU cores using multi-queue IFB
# Attempts to reduce single-core bottleneck for high-speed connections
# Keyword "dual-dsthost" - Per-host fairness caused excessive flow isolation, limiting bufferbloat test to ~62% throughput.
# Keyword "flows" - Per-connection fairness allowed parallel streams to utilize full bandwidth while maintaining excellent bufferbloat control.
# Recommendation: - Use `flows ingress nat` for +300Mbit connections with bufferbloat control.
# Key insight: mq is just a multiplexer - it needs XPS to tell it which queue to use. Without XPS, there's no distribution mechanism!
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License version 2 as
# published by the Free Software Foundation.
. ${SQM_LIB_DIR}/defaults.sh
INGRESS_CAKE_OPTS="besteffort"
EGRESS_CAKE_OPTS="besteffort"
QDISC=cake
NUM_QUEUES=$(grep -c "CPU" /proc/interrupts)
sqm_debug "Detected ${NUM_QUEUES} CPU cores"
egress() {
sqm_debug "egress (standard single-queue)"
SILENT=1 $TC qdisc del dev $IFACE root
$TC qdisc add dev $IFACE root $( get_stab_string ) cake \
bandwidth ${UPLINK}kbit $( get_cake_lla_string ) ${EGRESS_CAKE_OPTS} ${EQDISC_OPTS}
}
ingress() {
sqm_debug "ingress (multi-queue IFB with ${NUM_QUEUES} queues)"
# Clean up
SILENT=1 $TC qdisc del dev $IFACE handle ffff: ingress
SILENT=1 $IP link del dev $DEV 2>/dev/null
# Enable RPS on physical interface RX (distribute to CPU1,2,3)
local rps_mask="e" # 0xe = CPU1+CPU2+CPU3 (avoid CPU0)
sqm_debug "Enabling RPS on $IFACE RX queue (mask: ${rps_mask})"
echo ${rps_mask} > /sys/class/net/$IFACE/queues/rx-0/rps_cpus 2>/dev/null || \
sqm_warn "Could not set RPS for $IFACE rx-0"
# Create multi-queue IFB
$IP link add $DEV numtxqueues ${NUM_QUEUES} type ifb || {
sqm_error "Failed to create multi-queue IFB"
return 1
}
$IP link set dev $DEV up
# Enable RPS on IFB RX for pre-distribution
local ifb_rps_mask="f" # All CPUs for IFB
sqm_debug "Enabling RPS on $DEV RX queue (mask: ${ifb_rps_mask})"
echo ${ifb_rps_mask} > /sys/class/net/$DEV/queues/rx-0/rps_cpus 2>/dev/null || \
sqm_warn "Could not set RPS for $DEV rx-0"
# Calculate bandwidth per queue
local BW_PER_QUEUE=$(( ${DOWNLINK} / ${NUM_QUEUES} ))
sqm_debug "Bandwidth per queue: ${BW_PER_QUEUE}kbit"
# Set up mq root qdisc on the IFB
$TC qdisc add dev $DEV root handle 1: mq || {
sqm_error "Failed to add mq qdisc - multi-queue IFB not supported by kernel"
return 1
}
# Add CAKE to each queue
local i
for i in $(seq 1 ${NUM_QUEUES}); do
sqm_debug "Adding CAKE to queue 1:${i}"
$TC qdisc add dev $DEV parent 1:${i} $( get_stab_string ) cake \
bandwidth ${BW_PER_QUEUE}kbit $( get_cake_lla_string ) ${INGRESS_CAKE_OPTS} ${IQDISC_OPTS} || {
sqm_error "Failed to add CAKE to queue 1:${i}"
return 1
}
done
# Enable XPS on IFB TX queues (all CPUs)
local xps_mask="f" # All CPUs
sqm_debug "Enabling XPS on $DEV TX queues (mask: ${xps_mask})"
for i in $(seq 0 $((${NUM_QUEUES} - 1))); do
echo ${xps_mask} > /sys/class/net/$DEV/queues/tx-${i}/xps_cpus 2>/dev/null || \
sqm_warn "Could not set XPS for $DEV tx-${i}"
done
# Set up ingress redirection
$TC qdisc add dev $IFACE handle ffff: ingress
$TC filter add dev $IFACE parent ffff: protocol all prio 10 u32 \
match u32 0 0 flowid 1:1 action mirred egress redirect dev $DEV
}
sqm_prepare_script() {
do_modules
verify_qdisc $QDISC "cake" || return 1
# mq verification skipped - tested during ingress()
}
root@OpenWrt:~# uci show | grep steering
advanced.@advanced_packet_steering[0]=advanced_packet_steering
advanced.@advanced_packet_steering[0].advanced_packet_steering_enable='2'
advanced.@advanced_packet_steering[0].advanced_packet_steering_delay='30'
network.globals.packet_steering='0'
3 Likes
Hello, I bought the MT6000. I uploaded the latest PESA build with the openwrt-snapshot-r33756-49c9934b50-mediatek-filogic-glinet_gl-mt6000-squashfs-sysupgrade.bin extension. Is that all I need to do? Will everything install automatically, or do I still need to do something else? Thanks