I tested a recent build from master (from last week) and the H/W offloading seems to work OK. iperf3 for WAN-LAN traffic results in near zero sirq load. I'm still using firewall3 for my build. Not sure if firewall4 will cause problem, but it should not, as it's just a front-end to netfilter if my understanding is correct.
I have not tested PPPoE yet tho.
Too lazy to setup a PPPoE server to test ... haha.
I have 2 RT3200s working perfectly with 802.11s mesh. I replaced the basic ssl with wpad-openssl and libustream-openssl and use WPA3 encryption for the mesh settings. I set a static IP on the non-routing mesh point and disabled dnsmasq, firewall and odchpd on that device.
I do recall having to make a couple of changes to the LAN interface in the network settings area. I think that was setting the default gateway to my router's internal IP address and maybe the net mask. I remember the network part being a bit of trial and error to finally get it working so maybe that's what's needed? DHCP on the mesh node with a static IP assignment on the router side might work just as well as the static IP on the non-router.
I really didn't do much else as far as I can remember. I recall setting it up with a cable connected between the two before placing them in their final locations.
Hello,
I'm trying to install kmod-usb-net-cdc-ncm but I have a kernel version issue.
Collected errors:
* pkg_hash_check_unresolved: cannot find dependency kernel (= 5.10.113-1-d3a5d2af68dbbf43c631958a252f7270) for kmod-usb-core
* pkg_hash_check_unresolved: cannot find dependency kernel (= 5.10.113-1-d3a5d2af68dbbf43c631958a252f7270) for kmod-usb-net
* pkg_hash_check_unresolved: cannot find dependency kernel (= 5.10.113-1-d3a5d2af68dbbf43c631958a252f7270) for kmod-usb-net-cdc-ncm
* satisfy_dependencies_for: Cannot satisfy the following dependencies for kmod-usb-net-cdc-ncm:
* kernel (= 5.10.113-1-d3a5d2af68dbbf43c631958a252f7270)
* opkg_install_cmd: Cannot install package kmod-usb-net-cdc-ncm.
I'm running the last firmware version "OpenWrt 22.03-SNAPSHOT r19228-0877953235 / LuCI openwrt-22.03 branch git-22.119.37115-66994a5" found on released owrt-ubi-installer.
I tried to install the last upgrade from OpenWRT website (https://downloads.openwrt.org/snapshots/targets/mediatek/mt7622/openwrt-mediatek-mt7622-linksys_e8450-squashfs-sysupgrade.bin) but 5Ghz wifi disappears for available devices.
I've a lot of packages that are outdated, and impossible to upgrade them, with the same kernel version error.
Does anybody knows how I can do to upgrade kernel version?
Thanks
It's probably best to just install the latest snapshot, or you can try 22.03.0-rc1, it's the release candidate for the next stable version and the packages will be persistent so no worry about kernel mismatch.
Curious if there is any benefit of enabling "Generic receive offload (GRO)" (rx-gro-list / rx-udp-gro-forwarding) and if it is even implemented/supported in the mt* drivers?
From what I am reading from the code when the patches were posted to the kernel mailing lists, it doesn't necessarily need special support from the chipset drivers involved.
Are any packages known to set UDP_GRO on sockets?
I haven't used the attended sysupgrade in a while and now see this:
{
"url": "https://sysupgrade.openwrt.org",
"revision": "r19315-d4053d2e8e",
"advanced_mode": "0",
"app_version": "git-22.068.37885-65266c4",
"branch": "22.03",
"efi": null,
"rootfs_type": "squashfs",
"profile": "linksys,e8450-ubi",
"target": "mediatek/mt7622",
"version": "22.03-SNAPSHOT",
"packages": [
"auc",
"base-files",
"blockd",
"busybox",
"ca-bundle",
"dnsmasq",
"dropbear",
"firewall4",
"fstools",
"iperf3",
"kernel",
"kmod-fs-exfat",
"kmod-fs-ext4",
"kmod-fs-f2fs",
"kmod-fs-vfat",
"kmod-gpio-button-hotplug",
"kmod-leds-gpio",
"kmod-mt7615-firmware",
"kmod-mt7615e",
"kmod-mt7915e",
"kmod-nft-offload",
"kmod-nls-base",
"kmod-nls-cp437",
"kmod-nls-iso8859-1",
"kmod-nls-utf8",
"kmod-usb-storage",
"kmod-usb-storage-uas",
"kmod-usb3",
"libc",
"libustream-openssl",
"logd",
"luci",
"luci-app-attendedsysupgrade",
"luci-ssl-openssl",
"luci-theme-openwrt-2020",
"mtd",
"netifd",
"nftables-json",
"odhcp6c",
"odhcpd-ipv6only",
"opkg",
"ppp",
"ppp-mod-pppoe",
"procd",
"procd-seccomp",
"procd-ujail",
"tcpdump",
"uboot-envtools",
"uci",
"uclient-fetch",
"urandom-seed",
"urngd",
"wpad-openssl"
],
"diff_packages": true
}
root@OpenWrt:~# auc
auc (0.2.5-2)
Server: https://sysupgrade.openwrt.org
Running: 22.03-SNAPSHOT r19315-d4053d2e8e on mediatek/mt7622 (linksys,e8450-ubi)
No data available (61)
What's the issue here?
Same issue with auc here. The online image builder is working and just used it to sysupgrade.
I think good news re. PPPoE + hardware offloading with this update. My litmus test that it works is Amazon Alexa can reliably control my Philips Hue hub / lights. I enabled it after updating and so far so good. Will keep it turned on and see how things go over the next day or two. Previously with it on other things such as a couple of cloud / IP cameras report a loss of connectivity after 12 hours or so. If that doesn't happen and Alexa continues to work with the lights then I would consider the issue resolved!
Same issue with auc here. The online image builder is working and just used it to sysupgrade.
Does anyone have a clever 'ubus' command or otherwise to dump out the list of currently installed packages, so that it is just a simple cut and paste into the online builder? I suspect I could strip off the version information dumped from 'opkg list-installed' but it seems like there is probably a much easier way I'm not thinking of.
LuCi attended sysupgrade still not working for me. Same error message as above.
My easy way is I maintain a text file with the string of packages I have installed. I then remove a couple from the default list since they are replaced then I copy / paste my string onto the end of the list and verify using the installed list once the image is built.
Does this support 2.4ghz 40mhz?
I've set mine to use 40 but it's only broadcasting 20
If there's any overlap with other detected AP for the secondary it'll refuse to unless they vacate or you set noscan. The latter is non-compliant.
You may not see better performance anyway, as you then need to have a compatible client, and a not very noisy environment.
Thanks a lot, it works!
My testing shows massive performance drop with ax 160mh. Ax 80mhz, ac 160mhz, ac 80mhz all performing within 10% of each other.
Running 22.03-rc1
Sorry, no idea.
(I have never used auc, as I build my own firmware from scratch.)
I found the issue: the default attended sysupgrade server is down. Replacing with asu.aparcar.org in LuCi (or /etc/ocnfig/attendedsyupgrade) allows update to go ahead.
Looks like the thermal fix for 5.10 and 5.15 will be landing soon, which should boost performance a bit with less throttling, at least, that's what I've seen so far. It definitely affects the 'schedutil' freq numbers before and after.
I'm using this device with today's snapshot in dumb AP mode. Everything's working great, except for IPv6 traffic for wireless clients. Router advertisements are working fine -- wireless clients are getting IPv6 addresses from the PD advertised by my ISP and delegated by my router. Wired clients on the same subnet as the wireless clients have full IPv6 connectivity. The WAP itself has a SLAAC IPv6 address on its management interface and I have full IPv6 connectivity from the WAP, as well. It's only the WAP's wireless clients that can't receive IPv6 traffic.
Here's a tcpdump running on my router (the next hop for my wireless clients), observing a single ping from a wireless client to ipv6.google.com:
20:49:15.755256 IP6 (flowlabel 0x00400, hlim 63, next-header ICMPv6 (58) payload length: 16) 2001:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx > mad07s10-in-x0e.1e100.net: [icmp6 sum ok] ICMP6, echo request, id 5572, seq 0
20:49:15.774956 IP6 (flowlabel 0x00400, hlim 115, next-header ICMPv6 (58) payload length: 16) mad07s10-in-x0e.1e100.net > 2001:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx: [icmp6 sum ok] ICMP6, echo reply, id 5572, seq 0
20:49:15.843021 IP6 (flowlabel 0x928da, hlim 64, next-header ICMPv6 (58) payload length: 64) 2001:8a0:75be:9600::1 > mad07s10-in-x0e.1e100.net: [icmp6 sum ok] ICMP6, destination unreachable, unreachable address 2001:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx
On the router, ip -6 neigh show shows:
2001:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx dev br0.100 FAILED
and it's the same story for all other wireless client IPv6 addresses.
I recalled that my R7800 had a similar issue with IPv6, and I dug up these old threads which I used to fix the issue on the R7800:
However, setting neither allmulti nor promisc on all of my bridge interfaces makes any difference on my RT3200. (Wireless clients are on a separate VLAN (100) from the WAP's management traffic (101).)
Is anyone able to pass IPv6 traffic to wireless clients through this device when configured as a dumb AP, and/or is anyone else seeing the same issue as I am?
I'm interested to your IPv6 issue but I couldn't not reproduce.
I have RT3200 as dumb AP as well. The IPv6 works fine. My AP doesn't have IPv6 configured normally.
I tried adding DHCPv6 interface AP, with dhcpv6 or SLAAC only (reqaddress='none'). Wireless clients' IPv6 still work fine, in different VLANs/VAPs.
Here's my test config.
root@RT3200-3:~# uci show network.lan6
network.lan6=interface
network.lan6.proto='dhcpv6'
network.lan6.device='@lan'
network.lan6.reqprefix='no'
network.lan6.reqaddress='none'
root@RT3200-3:~# uci show firewall.@zone[0]
firewall.cfg02dc81=zone
firewall.cfg02dc81.name='lan'
firewall.cfg02dc81.input='ACCEPT'
firewall.cfg02dc81.output='ACCEPT'
firewall.cfg02dc81.forward='ACCEPT'
firewall.cfg02dc81.network='lan' 'lan6'
Could you provide more info on your VAP setup and AP SLAAC setup?
A tcpdump from your AP may be helpful too.