Keir Fraser wrote:
> On 01/02/2010 11:21, "Liu, Jinsong" <jinsong.liu@xxxxxxxxx> wrote:
>
>> 5. a simple way to solve the issue is, to make sure scan loop of
>> NTFY = scan loop of PRSC, and to make sure NTFY always scan 2^n
>> vcpus. However, NTFY scan loop may change in the future, not
>> necessary equal to 2^n, and not necessary equal to scan loop of
>> PRSC. That's the reason why I use for() loop inside of
>> decision_tree() in method NTFY at the patch I send Jan
>> 27. It will work correctly under whatever conditions, and keep
>> mk_dsdt.c easier to understand. Decision_tree indeed reduce scan
>> greatly, but it's not in key path.
>
> Ah, now I understand!
>
> Okay, how about the attached patch? This fixes the bug by inlining
> NTFY into PRSC, and also uses HVM_MAX_VCPUS as appropriate.
>
> Please take a look and also test it. I'll apply it if/when you Ack it.
>
> -- Keir
Keir,
It's really good to merge NTFY with PRSC, it's easy to understand and effective!
I just change a little, as attached 01-dsdt.
I test these 2 patches, it's OK.
Thanks,
Jinsong
00-dsdt
Description: 00-dsdt
01-dsdt
Description: 01-dsdt
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|