Re: RDMSR always returns 0?
Posted: Tue Jan 20, 2009 3:23 am
I don't think anyone is claiming that emulators should be a *replacement* for real hardware.
Emulators make progress faster by allowing the developers to debug issues much faster. When you test on real hardware in the early stages of the kernel (when there is inadequate/non-existent debugging code/traps), it is extremely time consuming to track down a bug that causes a triple-fault. In an emulator, you can freeze the virtual environment and get register dumps, memory dumps, single-stepping, and disassembly's of target areas. Also, if it works in an emulator well, then chances are it will work on real hardware, but like real hardware, they have to follow sketchy (non-)standards and such that need to be compensated for in the OS code.
bottom line: make the OS run in *all* environments.
Emulators make progress faster by allowing the developers to debug issues much faster. When you test on real hardware in the early stages of the kernel (when there is inadequate/non-existent debugging code/traps), it is extremely time consuming to track down a bug that causes a triple-fault. In an emulator, you can freeze the virtual environment and get register dumps, memory dumps, single-stepping, and disassembly's of target areas. Also, if it works in an emulator well, then chances are it will work on real hardware, but like real hardware, they have to follow sketchy (non-)standards and such that need to be compensated for in the OS code.
bottom line: make the OS run in *all* environments.