Re: [patch 3/4] QEMU/KVM: non-virtualized ACPI PMTimer support

[Avi Kivity <avi@qumranet.com>]

Marcelo Tosatti wrote:
> On Sun, May 25, 2008 at 01:18:46PM +0300, Avi Kivity wrote:
>   
>>>         pm_io_base = PM_IO_BASE;
>>> +        pmtmr_base = cmos_readb(0x60);
>>> +        pmtmr_base |= cmos_readb(0x61) << 8;
>>> +        pmtmr_base |= cmos_readb(0x62) << 16;
>>> +        pmtmr_base |= cmos_readb(0x63) << 24;
>>> +        if (!pmtmr_base)
>>> +            pmtmr_base = pm_io_base + 0x08;
>>>  
>>>       
>> You're splitting the ACPI ioport range into two.  I think the correct 
>> fix here is to have qemu supply a PMBA hint to the BIOS.  If the hint is 
>> present, the bios should locate pm_io_base there, and should also avoid 
>> placing other pio resources there.
>>     
>
> PBLK_BASE (processor block) is statically defined in acpi-dsdt.dsl, and
> I don't see any easy way to change that dynamically.
>
> In practice I don't see any problem with not having the PMBA registers
> in a contiguous range, since I doubt any ACPI implementation will assume
> PIIX4 specific details (ACPI driver looks for the register addresses in
> FADT, so does not matter if they're contiguous or not).
>   

Right, it's a cleanliness issue, not correctness.

The only guest I can think of that will care about the actual layout is 
linuxbios/coreboot.

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.