Re: [ACPI] Fw: Re: 2.6.15-rc5-mm1

[Voluspa <li...@te...>]

On 2005-12-06 15:12:28 Matthew Garrett wrote:
>On Tue, Dec 06, 2005 at 10:17:38AM +0300, Roman I Khimov wrote:
>>
>>    *C1:                  type[C1] promotion[--] demotion[--]
>>    latency[000] usage[00000000]
>
>Yup, that looks like the problem. I've no idea what changed in this 
>area, but without C2 state support heat output will be higher.

Observe that C1 claims to be unused as well. I just compiled the
2.6.15-rc5 proper and have lost my one and only C-state:

root@sleipner:~# diff -Nur dmesg-2.6.14 dmesg-2.6.15-rc5 |grep "power "
-ACPI: CPU0 (power states: C1[C1])

root@sleipner:~# cat /proc/acpi/processor/CPU0/power
active state:            C1
max_cstate:              C8
bus master activity:     00000000
states:
   *C1:                  type[C1] promotion[--] demotion[--]
latency[000] usage[00000000]

The usage field really does get used in 2.6.14 while staying zero in
15-rc5. I suspect this is the culprit (and no, I haven't tried the nocst
boot param):

--quote--
commit 6d93c64803a5fea84839789aae13290419c62d92
Author: Venkatesh Pallipadi <ven...@in...>
Date:   Thu Sep 15 12:19:00 2005 -0400

    [ACPI] Prefer _CST over FADT for C-state capabilities
    
    Note: This ACPI standard compliance may cause regression
    on some system, if they have _CST present, but _CST value
    is bogus. "nocst" module parameter should workaround
    that regression.
    
    http://bugzilla.kernel.org/show_bug.cgi?id=5165
    
    Signed-off-by: Venkatesh Pallipadi<ven...@in...>
    Signed-off-by: Len Brown <len...@in...>
    (cherry picked from 883baf7f7e81cca26f4683ae0d25ba48f094cc08 commit)
--unquote--

Mvh
Mats Johannesson
--