Cisco switch problem - port shuts down due to loopback.

I have a bizarre problem. I have a Cisco 2960X switch that started shutting its uplink port down due to loopbacks being received.

I can console into the switch and shut all 48 ports down, but when I bring the uplink port up, 10 seconds later I see a loopback error message and the port goes into err-disabled.

I have swapped uplink ports on both the affected switch and the upstream switch. I've swapped cables. Same thing.

If I do a "no loopback" on the 2960X uplink port, the problem goes away.

Last night I updated the 2960X and the 2960S to the latest starred IOS. The problem went away for a few hours but came back.

Here's the topology

S2 (2960X) <-> S1 (2960S) <-> S0 (2960G)

The S2 - S1 link is a singe 1000Base-T connection, and the S1 - S0 link is a single 1Gb MM fiber.

This setup has been in place, mostly untouched for about two years. The problem started yesterday and the local IT guy swears he made no topology changes or switch config changes. The local IT guy has verified there's no device between S2 and S1 - it is a cat6 patch cable.

If I understand this correctly, the loopback error message is telling me that S2 is sending S1 its own loopback packets back to it. What could make a switch do that?

The switch configs are pretty simple. Layer two only. A handful of VLANs. No QoS or inter-VLAN ACLs. One SVI for switch management.

Any ideas on how to troubleshoot this? Notions on what causes it?