One potential reality is, the BIOS populate the CSDT definition based on the
same condition as the evaluatioin of the _OSC method. (i.e. Initially all is
0x80000000, BIOS initialization code check the same condition, and only malloc
memory and update it if the condition meet, later the _OSC method check the
same condition also). It is said BIOS does not change the DSDT/SSDT table
dynamically, but just create it, and if needed, update several field.
But yes, it will be helpful to get confirmation from OEM.
BTW, Jan, where did you get the native kernel complaining? I didn't find it in
Carsten's mail.
Thanks
--jyh
>-----Original Message-----
>From: Jan Beulich [mailto:JBeulich@xxxxxxxxxx]
>Sent: Wednesday, September 15, 2010 4:15 PM
>To: Carsten Schiers
>Cc: jeremy; Wei, Gang; Wang, Winston L; Jiang, Yunhong; xen-devel
>Subject: AW: RE: AW: Re: [Xen-devel] ACPI problem, was Xen BUG in mm / Xen
>4.0.1
>with 2.6.32.18/21 pvops Kernel?
>
>>>> On 15.09.10 at 09:22, "Carsten Schiers" <carsten@xxxxxxxxxx> wrote:
>> As I am not such an expert (although through this experience, I now know
>> much more about ACPI),
>> can we now assume that the BIOS is ok? It's because I mailed the Asrock
>> guys already and either
>> need to give them the latest info, or I would explain them everything is
>> settled.
>
>Based on even the native kernel complaining (just not dying) the
>BIOS certainly isn't fully correct. This particularly includes (but isn't
>necessarily limited to)
>
> Name (CSDT, Package (0x18)
> {
> "CPU0CST ",
> 0x80000000,
> 0x80000000,
> "CPU1CST ",
> 0x80000000,
> 0x80000000,
> "CPU2CST ",
> 0x80000000,
> 0x80000000,
> "CPU3CST ",
> 0x80000000,
> 0x80000000,
> "CPU4CST ",
> 0x80000000,
> 0x80000000,
> "CPU5CST ",
> 0x80000000,
> 0x80000000,
> "CPU6CST ",
> 0x80000000,
> 0x80000000,
> "CPU7CST ",
> 0x80000000,
> 0x80000000
> })
>
>listing invalid addressed for *all* CPUs (not just the ones actually
>not present). Another point is that currently you may be luck in that
>they may not get evaluated. If they can guarantee that this will
>never get accessed, imo they should just leave it (and the dead
>access) out.
>
>Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home