I posted a little about this in the 24.10 release announcement thread as well but it would be good to get some developers to weigh in on if this is something that should be addressed or just carry on as normal.
In kernel 6.6.51 a commit was added to usbnet.c that changed the previous behavior of naming devices usb* to eth* in some instances. The mainline linux commit is the following. I have validated that applying just this commit to usbnet.c from 6.6.50 changes the device names from usb to eth.
This impacted at least my use case which is tethering from a phone/tablet. There was a follow on commit that supposedly fixed a "name regression" but this did not change the behavior in openwrt since the commit is in the 24.10 release kernel and my tethered connections were still named eth*.
The name issue was easy enough to fix in the network config file but I had forgot that I also had a nftable rule for altering the ttl that depended on the usb* names so it caused me some grief temporarily. I would think this could also impact others that haven't upgraded to 24.10 but are planning to do so.
There was discussion on the behavior change on the lore.kernel.org thread for the original commit by Dominique Martinet.
https://lore.kernel.org/all/20241017071849.389636-1-oneukum@suse.com/#t
He proposed the following patch to reinstate the previous behavior but it does not appear to have been excepted by GKH. I can confirm that applying this patch returns my tethered device names to usb*. A thought crossed my mind that if a patch such as this is accepted into the kernel at some point in the future it could catch people that already modified their configs to the eth naming convention unawares. I think at the very least a mention should be made in the 24.10 release notes about this change in behavior.
https://lore.kernel.org/all/a7b0b637-8ecb-4092-b6a9-162bafb95454@lunn.ch/T/#u