Belkin RT3200/Linksys E8450 WiFi AX discussion

For Developers
6789

Good point, and I left out a step for brevity. I actually ran the dump twice, then compared the log files. If they were the same, it's a good chance there were no transfer errors. If they're different, run it a third time and see if any two files match. It doesn't take a lot of extra time for the factory partition, but it did take a lot longer (multiple days) when dumping all mtds to get the whole OEM image.

The goal for this method was to avoid altering the OEM firmware in any way. The typical install process involves getting the openwrt recovery image installed, which alters the flash. Is there another way to get the initramfs on there? Maybe TFTP (i was avoiding this just because it was a whole other thing to setup)?

6790

Hi @cocorouter I also experienced what you described and it was due to missing Default package fitblk, after installing it and trying the upgrade again it worked as expected.

For context:

6791

EDIT: ignore this post, there is 0xf=0b1111 for 4 antennas and it is the correct value.

Old text

On recent OpenWrt SNAPSHOT r31193-605543bc7b I see following otput:

root@OpenWrt:~# iw list | grep -e phy -e Ante
Wiphy wl1
        wiphy index: 1
        Available Antennas: TX 0xf RX 0xf
        Configured Antennas: TX 0xf RX 0xf
Wiphy wl0
        wiphy index: 0
        Available Antennas: TX 0xf RX 0xf
        Configured Antennas: TX 0xf RX 0xf

Which seems wrong because we don’t have 8 antennas?

Is it because of blank values in EEPROM?

Ref to similar issues threads:

https://github.com/openwrt/mt76/issues/980#issuecomment-2977364759

https://github.com/openwrt/openwrt/issues/19323

6792

Same output for me on 24.10.3:

root@CharlieExt:~# iw list | grep -e phy -e Ante
Wiphy wl1
        wiphy index: 1
        Available Antennas: TX 0xf RX 0xf
        Configured Antennas: TX 0xf RX 0xf
Wiphy wl0
        wiphy index: 0
        Available Antennas: TX 0xf RX 0xf
        Configured Antennas: TX 0xf RX 0xf

What does the 0xf mean?

6793

0xf means that there are 4 antennas, which is what this device actually has.

6794

You are right, it’s bitmask, that was silly mistake from my side. It is indeed okay.

6795

I ran into the same issue. This should be put in FAQ or probably the TOH for linksys E8450 device.

6796

I am just now upgrading my RT3200 from 23.05 to 24.10. Per the device page’s instructions here, I attempted to apply the latest UBI installer, v1.1.4. Below is the image I used.

https://github.com/dangowrt/owrt-ubi-installer/releases/download/v1.1.4/openwrt-24.10.0-mediatek-mt7622-linksys_e8450-ubi-initramfs-recovery-installer.itb

After attempting to force flash the image via LuCi without retaining any prior configuration, I do not believe that the device booted afterward. Numerous attempts to hard cycle the device after unplugging have not helped. Trying a different 12v power supply didn’t help. The LEDs never come on. But a Kill-a-Watt does show that it is pulling 2.5 watts and isn’t completely dead.

Edit: I did wait at least ten minutes after flashing prior to any attempts to hard cycle

I was coming from the 1.0.3 installer and OpenWrt 23.05.3.

I have ordered a USB to TTL adapter and hope that I can recover it. But does anyone have any guidance? I do see that the likely dated serial recovery instructions on the main device page have different instructions for whether or not the v1.1.3 or later installer has been run. In the case of potentially failed or improperly done installation, how should I proceed?

6797

Connecting the serial console is the best and in this case probably only way to even find out what has happened. The SoC’s BootROM is quite verbose on power on, so that will already tell quite a bit about the situation, and you’ll see if ARM TrustedFirmware-A and U-Boot come up at all, and if not, what the BootROM got to say about it.

In case the BootROM doesn’t like to launch BL2 you can use mtk_uartboot to launch U-Boot and attempt to re-run the installer, this time watching the progress on the serial console.

Important: Use the *bl31-uboot.fip file of OpenWrt 23.05.x because it can be that the device still got the factory data written according to the flash layout before the switch to the v1.1.x installer (we don’t know why and in which state the installer went south at this point)

6798

Thank you.

I followed the recovery instructions step by step using mtk_uartboot utility with the custom BL2 and 23.05.3 release FIP. I proceeded to perform the following all within the same session.

  1. The device booted
  2. I logged into LuCi
  3. I force flashed the 1.1.4 installer without retaining config
  4. The device booted into recovery mode
  5. I flashed the latest 24.10 release.
  6. The device booted into what I believe was recovery mode 24.10.0 recovery mode. I do know that the status page showed version 24.10.0.

Logging out and logging in, LuCi still showed 24.10.0. Power cycling the device resulted in the expected 24.10.3 version.

Is this expected? I will report back if I encounter unexpected behavior.

Whatever the case, I should note that prior to the device being bricked, I had not rebooted the device for approximately 18 months. It was always stable and performed well for my family. Many thanks to the devs and community for maintaining and enhancing this wonderful platform.

6799

Yes, that sounds right. The installer will setup the recovery partition with version 24.10.0, and then the actual live system would be running the most recent patch release.

Here's a way to tell the difference (using SSH):

ubus call system board | grep rootfs_type

And how to interpret the result:

  • squashfs = normal
  • initramfs = recovery
6800

There were two interesting commits to mt76 recently that affect our router:

mt76: mt7915: add bf backoff limit table support
mt76: Fix DTS power-limits on little endian systems

From my reading, they correct longstanding bugs relating to misconfigured power limits and beamforming settings, and could improve wireless performance.

6801

Is it relevant to 24.10.x branch?

6802

Yesterday when I updated the firmware to the then-latest snapshot (“kernel: ksmbd: revert upstream limit repeated connection”), WiFI wouldn’t come on-line. The “wireless” page in luci would show the interfaces and networks, but I couldn’t enable any of the networks. I think there was something along the lines of “missing radio”. Wired network and WAN worked fine, as did channel analysis on all 4 radios.

Apologies for not having more debugging info; I had to restore an older back-up to make the device usable again. It’s now at OpenWrt SNAPSHOT r31056-2503e5bc37, which seems a bit older than I bargained for (14th of September). I remember having a kernel 6.12.49 running before yesterday, must have forgotten to back up that specific itb.

I’m not sure whether this is an actual bug in said snapshot, or whether something (e.g. interface name) has renamed in a way that the configuration on the device is no longer valid. But hope the heads-up helps.

6803

What did you use to update the firmware?

I had the same experience using owut and it turned out the build was missing iwinfo.

6804

I use a fork of and old version of the owrt-ubi-installer to create a sysupgrade itb. Not the perfect solution, but it usually does what I want it to. Appreciate the hint! I guess iwinfo used to be a dependency for one of the other packages, but now isn’t anymore? There definitely seem to have been some iwinfo changes in the git repo from a week ago. I’ll go and play with this next time I have some free time (will take a few days).

6805

It still is, however a couple of commits changed things and back.

6806

After various distractions, I finally got some time to try to install OpenWRT on my RT3200.

I’ve got an odd issue with the OEM firmware (can’t actually get OpenWRT installed yet) where, after getting past the wizard (I clicked exactly inside the circle and everything), I’m presented with the password screen, but after entering ‘admin’ it loads the dashboard for about a quarter second before kicking me back to the same login screen.

I tried SSH, it’s not enabled, and I’ve used both Firefox (turned of uBlock too) and Chromium. Also tried setting the PC date to 2023, but that didn’t help either. I can’t read the console because somehow the kick process resets the page load information and that vanishes too.

Am I going to have to find a serial adapter?

6807

got the same problem with E8450 only very old firmware.perhaps 1.0.0 or 1.0.01 .And must update firmware to solve this.
I have 1-2 time successful login and update firmware to 1.1unsigned by using some very old browser(very old os also worth trying). this method only work few time at first before these browsers not work,too.
Further more I also got losing of wireless 5Ghz on some E8450 with non-ubi layout but ubi is ok and EA8500 devices as well.
Using serial adapter is better.

6808

Sadly, this is indeed an issue with modern browsers vs. old firmware. The whole thing is due to the firmware using older coding standards that are directly incompatible with modern browser security requirements. It’s a situation that exists on many older embedded devices, not just the E8450/RT3200. To get around the problem, you can use an old version of a browser (Look for a 2020-2022-era release of Firefox; there are portable and pre-packaged versions around) or you can indeed upload the UBI installer via serial console.