When a OS signal/exception causes a panic a call to runtime.sigpanic is pushed onto the goroutine stack but a debugger has no way to recover the state of the CPU registers for the frame that caused the signal to happen. Having them would be useful for debugging optimized executables (where some variables may be registerized) as well as debugging problems with functions implemented in assembly.

Context: go-delve/delve#1838 (comment)

cc @heschik