Re: [Xen-devel] [PATCH 3/4] x86-64: EFI runtime code

["Jan Beulich" <JBeulich@xxxxxxxxxx>]

>>> On 28.06.11 at 18:05, Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx> wrote:
> What kind of minimal "stuff" is required from EFI for Dom0 consumption?

The minimal set of things is obtaining the various pointers to
configuration tables (ACPI, MPS, SMBIOS).

> I see the need for the VGA console, and that looks easy enough.

There's no VGA console in modern UEFI, you only get a graphical
one (with no guarantee that it's VGA).

> The E820 is parsed by the Xen hypervisor, so XENMEM_machine_memory_map from 
> Dom0
> will take care of that.
> 
> The other information Dom0 needs is the ACPI DSDT to figure out the polarity
> and trigger for the interrupts. Is the ACPI DSDT "lost" when it is running 
> from EFI?

No, but it can't be retrieved the "traditional" way.

> In the future we will need the _CST, _PSS (and its friends) to shuffle
> the contents of that up to the hypervisor for CPU Cx/Px states. Does the 
> ACPI
> runtime code need to be aware of EFI?

No - ACPI is all the same except for the mechanism to retrieve the
root pointer.

> What about ACPI S3? The contents of ACPI FADT table is required for that. 
> Will
> EFI make that obsolete?

Same again.

> I am just wondering whether Dom0 actually needs to know that EFI exists.

Yes, it needs to, due to the table pointer retrieval differences. And
then there's the runtime services (RTC access, EFI variables) that
the OS ought to use. On the RTC side of things, a legacy free system
might not have a CMOS clock, and hence even native Linux is currently
broken in that respect (because the driver can be built only for IA64).
Access to the variables is necessary so you can control the EFI
environment (most importantly configure the boot loader) from within
the OS.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel