On 19/11/2014 15:16, Dr. David Alan Gilbert wrote: >> > Consider the similar case on real hardware: >> > >> > boot >> > update microcode RPM >> > s4 >> > turn on >> > >> > CPU microcode is installed early by the kernel, before looking for a >> > hibernation image to resume from, so the CPU microcode after resume from >> > S4 is different from the microcode at the time you suspended to disk. >> > This probably would work. > You mean, unless for example, someone had disabled a CPU feature in the > new microcode?
A random example, right? :) I think it reinforces my point---just like it would work almost always on real hardware, but can fails if the stars align right, it should work almost always on QEMU. It doesn't have to be _perfect_. Bugs are always possible, of course, but things can be tested. Interested people/downstreams (hint, hint!) can try doing S4 on a release and restarting on the next, with and without machine type changes. If it breaks, it can be fixed or just documented. Paolo
