It seems a C6 state is an individual core sleeping. The intersection of people who don't reboot for 3 years and people who have sleep states enabled must be pretty small. It's an interesting bug though!
I had a very similar issue with some AMD-based servers (bulldozer, I think) about ten years ago. There was a bug where Xen-based virtual machines could set a C-state on cores it was assigned, but for whatever reason it wasn't able to wake them up. It was fun trying to figure out what the heck was going on.
Do you mean a bug in an old version of linux that is now fixed? Because I have been using Zen3 and Zen3+ on linux since their release and never had to mess with C-states.
EPYCs have many cores, and most applications (including those with long uptime requirements) use only a subset of the cores continuously. So it is totally normal for some of the cores to go to deep sleep C6 during phases of lower load. It will cause server operators headaches, when those cores don't come back eventually. Reboots help, disabling C6 in the (already running) OS also helps.
Please note, that we are not talking about a core sleeping for three years. We are talking about a core going to deep sleep, when the system has been up for three years or longer.