Given the state of hardware and software support throughout the graphics stack on Linux we didn't think we could make this reliable," he explained
"It doesn't probe and reattach the controllers to the same point in the device tree that they were in when the system was suspended. Since those are the device nodes that SteamOS has open at the time of the suspend, and they route to The Noplace(tm), the controllers become unresponsive," user tlambert wrote. "This is a general problem in the Linux device model, and you can see problem in the device model poke their heads up in various places.
The issue is rooted, according to a Valve engineer posting on GitHub as JohnV, not in the suspend part of the equation. That works fine. It’s in the resume part. When a Linux session is resumed following suspend, the way it rediscovers hardware doesn’t work in a way that works with resuming games. In this case, that applies to the controllers plugged in.
Because Steam Machines use SteamOS, which is based on Linux and meant to be kept open in nature, this probably doesn’t look like something that will get fixed unless it gets fixed at a level more basic than SteamOS.