Xen devel - Nov 2006 - [PATCH][HVM] Reserve MMIO a0000-c0000 in ACPI dsdt

Hi,
This patch reserves MMIO 0xa0000 to 0xc0000 in ACPI dsdt, it avoids
possible vga driver loading problem in HVM Windows guest.
Also fixes a Makefile bug in hvmloader directory.

Signed-off-by: Qing He <qing.he@intel.com>


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

On 16/11/06 7:53 am, "He, Qing" <qing.he@intel.com> wrote:
> This patch reserves MMIO 0xa0000 to 0xc0000 in ACPI dsdt, it avoids
> possible vga driver loading problem in HVM Windows guest.
> Also fixes a Makefile bug in hvmloader directory.
Is this *actually* a problem? I didn''t think BIOS tables had to report
0xA0000-0xBFFFF as special. Certainly e820 and EFI reporting mechanisms
should ignore it as the hole is a well-defined part of the PC memory layout.
Also OSes don''t treat or probe for standard vga as a pci device.

 -- Keir



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

>
>> This patch reserves MMIO 0xa0000 to 0xc0000 in ACPI dsdt, it avoids
>> possible vga driver loading problem in HVM Windows guest.
>> Also fixes a Makefile bug in hvmloader directory.
>
>Is this *actually* a problem? I didn''t think BIOS tables had to
report
>0xA0000-0xBFFFF as special. Certainly e820 and EFI reporting mechanisms
>should ignore it as the hole is a well-defined part of the PC 
>memory layout.
>Also OSes don''t treat or probe for standard vga as a pci device.
>
without this fix, 32bit Vista boots ok, but 64bit Vista can not load vga
driver.
-Xin

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

Actually, I think it matters.
This DSDT section is directly drawn from one native box. Without it, 64bit vista
cannot boot successfully, and 64bit Windows 2003 reports resource conflict due
to this region.

I don''t know why only 64bit Windows complains, but add the region in
dsdt would solve the problem. Per my experience, Windows has higher restrictions
on ACPI tables than Linux.

Thanks,
Qing
>-----Original Message-----
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk]
>Sent: 2006年11月16日 16:08
>To: He, Qing; xen-devel@lists.xensource.com
>Subject: Re: [Xen-devel] [PATCH][HVM] Reserve MMIO a0000-c0000 in ACPI dsdt
>
>On 16/11/06 7:53 am, "He, Qing" <qing.he@intel.com> wrote:
>
>> This patch reserves MMIO 0xa0000 to 0xc0000 in ACPI dsdt, it avoids
>> possible vga driver loading problem in HVM Windows guest.
>> Also fixes a Makefile bug in hvmloader directory.
>
>Is this *actually* a problem? I didn''t think BIOS tables had to
report
>0xA0000-0xBFFFF as special. Certainly e820 and EFI reporting mechanisms
>should ignore it as the hole is a well-defined part of the PC memory layout.
>Also OSes don''t treat or probe for standard vga as a pci device.
>
> -- Keir
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 16/11/06 9:32 am, "He, Qing" <qing.he@intel.com> wrote:
> Actually, I think it matters.
> This DSDT section is directly drawn from one native box. Without it, 64bit
> vista cannot boot successfully, and 64bit Windows 2003 reports resource
> conflict due to this region.
> 
> I don''t know why only 64bit Windows complains, but add the region
in dsdt
> would solve the problem. Per my experience, Windows has higher restrictions
on
> ACPI tables than Linux.
Ok, I did the same with my native ACPI tables and yes, that region is marked
as reserved. I also got scared by the size and complexity of modern DSDTs. I
hate AML!

 -- Keir



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