I've been troubleshooting Wi-Fi 5GHz issues on my Belkin RT3200 (running OpenWrt 23.05.5), and I discovered that the EEPROM for the MT7915 seems to be missing or invalid on my unit. The driver loads with the default calibration, power is limited to 6 dBm, and a random MAC is assigned at every boot.
Here’s what I see in dmesg:
mt7915e 0000:01:00.0: eeprom load fail, use default bin
mt7915e 0000:01:00.0: Invalid MAC address, using random address a6:02:f1:2d:a2:51
I’ve also verified the current contents of /lib/firmware/mediatek/mt7915_eeprom.bin and it’s completely filled with 0xFF, meaning it's empty or corrupted.
I’d like to ask if anyone with a working E8450 (or compatible MT7915D/E device) can help by extracting their EEPROM and sharing a generic version (with MAC removed). This is perfectly safe and does not expose any personal data as long as the MAC is stripped beforehand.
What I need:
A working 4KB file of the mt7915_eeprom.bin
Ideally with MAC blanked or zeroed out
From a unit where dmesg shows EEPROM loaded successfully
(Optional) To remove the MAC, you can overwrite the first 6 bytes with 00 or random values, or I can do that myself.
Share the file (e.g. upload to file.io, paste.rs, or any temp host)
I’ll use it just to restore calibration and inject my own MAC. If you’ve restored Wi-Fi performance on this device before, your help would mean a lot.
Thanks in advance
“Although I’m using a Belkin RT3200, it is hardware-identical to the Linksys E8450 — they share the same SoC, Wi-Fi chipset (MT7915D/E), and partition layout. So a valid EEPROM from either model would be compatible.”
Just FYI, calibration data is unique to the indvidual device - but restoring some other unit's calibration may help. The equipment and software to restore your calibration is at the OEM, who mass ordered the WiFi chips in your device (i.e., Link$ys).
Caldata represents factory calibration, eg 5 or 10 dbm LNA. Check "coverage", much better signal than before means you are likely out of spec and legal limits. Though depends on location if black helicopters arrive next morning or next year. Show first 4 octets of MAC, matching caldata would be more likely from same factory using same parts batch.
Thanks! The first 8 hex digits of my MAC are D8:EC:5E:79.
I believe the caldata might be partially corrupted—I'm trying to restore Wi-Fi functionality (especially 5 GHz) on a Belkin RT3200 running OpenWrt. If anyone happens to have a factory dump or known-good EEPROM with a matching or nearby MAC prefix, it’d be incredibly helpful to compare or test.
I appreciate any help—and if you need specific output like hexdump, mtd layout, or dmesg excerpts, let me know. Thanks again.
The RT3200/e8450 doesn't actually use individual calibration data, so that's one less thing to worry about. However, if you're getting that error, your factory partition is likely damaged and so you probably also have a broken or missing EEPROM from the mt7622 radio. The mt7622 will still run without the EEPROM, but you'll be missing features and be limited to 6dBm output, exactly as you described. The factory partition as a whole should be replaced if missing, since it's what OpenWRT will try to use if possible, and it will only resort to trying to load the EEPROM from the filesystem if the factory data cannot be found. The factory partition will also remain stable and unchanged across firmware updates, but individual filesystem files can be lost. To that end, I produced a surrogate partition back during the days when OKD was a problem.
That file contains rebuilt versions of both EEPROMs and, at least by my own testing and the reports from users that have used it, it still provides full or near-to-full functionality while containing only the data directly referenced by the drivers for each radio. There are no MAC addresses present in the file. Instead, it waits for OpenWRT to generate random MAC addresses upon boot.
I had a damaged factory partition on my RT3200 (running OpenWrt 23.05), which caused the 5GHz radio (MT7915) to fall back to default EEPROM with no calibration — limited to 6 dBm and random MACs. > > Thanks to a helpful link shared by @grauerfuchs, I used the e8450_factory_emergency.bin file as a full replacement for the factory partition. > > Here's how I fixed it: > > 1. Backed up my original factory partition: > > > dd if=/dev/mtd2 of=/tmp/factory_backup.bin > > > 2. Enabled write access using: > > > opkg install kmod-mtd-rw > insmod mtd-rw i_want_a_brick=1 > > > 3. Flashed the emergency bin: > > > mtd write /tmp/e8450_factory_emergency.bin factory > > > 4. Rebooted, and now 5GHz shows proper calibration, fixed MAC, and 20 dBm Tx power. > > Hope this helps anyone else with a wiped or broken factory partition — this method saved my router without needing to recompile firmware or fiddle with overlay tricks!
For those following along because of similar problems, the above posted steps for recovery are viable on OpenWRT 23.05.x and older. OpenWRT 24.10 and newer require a major data layout change for this device, so a different set of instructions must be used. At the time of this post, there is a link to complete recovery instructions for both versions (including how to modify and write the factory data) at the bottom of the Troubleshooting section of the wiki page for the Linksys e8450.
What I need:
How to extract it (safely):