On Wed, Feb 10, 2016 at 11:14:30AM -0500, John Snow wrote: > On 02/09/2016 01:48 PM, Michael S. Tsirkin wrote: > > On Tue, Feb 09, 2016 at 07:36:12PM +0100, Laszlo Ersek wrote: > >> Implementing this in QEMU would require: > >> - inventing virt-only registers for the FDC that provide the current > >> disk geometry, > > We do secretly have these registers, but of course there's no spec to > interpreting what they mean. For instance, what do they read when the > drive is empty? I am not sure that information is codified in the ACPI spec.
There are various possibilities, like returning False for the corresponding drive in _FDE (Floppy Disk Enumerate) object, or returning 0 aka unknown as drive type in _FDI. Anyway I'd hate needing to expose all of that in an ACPI-accessible form, this is going to become too fat. > Could be wrong, you seemed to indicate that the ACPI spec said that the > info matches what you get from a certain legacy bios routine, but I > don't know the specifics of that routine, either. Indeed, it's supposed to do the same as https://en.wikipedia.org/wiki/INT_13H#INT_13h_AH.3D08h:_Read_Drive_Parameters and, as I wrote in another mail, the SeaBIOS implementation here is rather simplistic. > Roman, does the 0xFF "empty disk geometry" hack appear to work for > Windows 10? > > Maybe it's fine enough as-is, as per Laszlo's good synopsis here. I'll test and let you know. Roman.
