Voodooprojects

VoodooPower => Bug Reports => Topic started by: Azimutz on April 16, 2010, 03:23:23 AM

Title: VoodooPowerMini Snow feedback
Post by: Azimutz on April 16, 2010, 03:23:23 AM
Hi Superhai,
machine, patches and system are on the sig; kernel is patched just with cpu id, bcopy stuff for emt64 and kext blacklist.
Loading the kext results in kp (pics attached), any arch, type of install/load, either normal or debug version. Nothing gets to logs or from debug version. 10.5.x version works fine.
Nothing more worth of mention. If you need any other feedback, i'm glad to provide.

Thanks
Stay safe...
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 16, 2010, 07:35:29 AM
Ok, I know why it panics now. I will see if I can make some kind of fix.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 16, 2010, 08:43:02 AM
Nice :) will stay tuned...
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 16, 2010, 10:23:49 PM
Forgot to mention something..
The personalities blocked by kext blacklist:
Code: [Select]
    { "com.apple.driver.AppleIntelMeromProfile",    0 },
    { "com.apple.driver.AppleIntelNehalemProfile",    0 },
    { "com.apple.driver.AppleIntelPenrynProfile",    0 },
    { "com.apple.driver.AppleIntelYonahProfile",    0 },
    { "com.apple.driver.AppleIntelCPUPowerManagement",    0 },
the cpu family Profile ones are blocked because they cause a huge hang at boot (right before getting to login window) when booting with "64 bit mode" enabled or k64. Obviously, they don't need to be blocked for Core cpus.
Don't know if these are relevant for power management?!
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 18, 2010, 01:23:30 AM
Try the version here and see if it works

http://superhai.com/beta-darwin.html

Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 18, 2010, 03:10:51 AM
Ok.. this one loads but it seems there's no stepping. From kernel log:

i386 loaded from E/E.mkext (64 bit mode enabled):
Code: [Select]
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB6B8E]           probe:   95: CPU Family 15 Model 6
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB73C7]           probe:  235: Netburst calculations
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB73F9]           probe:  236: Minimum fid: 0F vid: 27
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB743B]           probe:  237: Maximum fid: 0F vid: 27
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB7495]           probe:  241: Using preset values
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB74C7]           probe:  242: Minimum fid: 0E vid: 1A
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB75A5]           probe:  252: P-state 0 - 0F27
Apr 18 00:44:21 localhost kernel[0]: Voodoo[0x41CB75A5]           probe:  252: P-state 1 - 0E1A
Apr 18 00:44:21 localhost kernel[0]: VoodooPowerMini 1.2 (C) 2009-10 Superhai, All Rights Reserved. Apr 18 2010 01:12:59 DEBUG 32 bit
x86_64 loaded from S/L/E:
Code: [Select]
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0ECB1]           probe:   95: CPU Family 15 Model 6
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F517]           probe:  235: Netburst calculations
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F548]           probe:  236: Minimum fid: 0F vid: 27
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F587]           probe:  237: Maximum fid: 0F vid: 27
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F5DF]           probe:  241: Using preset values
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F610]           probe:  242: Minimum fid: 0E vid: 1A
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F6F3]           probe:  252: P-state 0 - 0F27
Apr 18 01:02:47 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80E0F6F3]           probe:  252: P-state 1 - 0E1A
Apr 18 01:02:47 AziLandz kernel[0]: DSMOS has arrived
Apr 18 01:02:47 AziLandz kernel[0]: VoodooPowerMini 1.2 (C) 2009-10 Superhai, All Rights Reserved. Apr 18 2010 01:12:11 DEBUG 64 bit
The main difference between loading from S/L/E versus E/E.mkext is that it loads later when from S/L/E. A while ago i managed to get pstate info from OS X with PstateChanger, etc.. after a ton of kp's:

pstate,  MHz,  mV,    FID, VID
0=        3000, 1262, 0F,   27
1=        2800, 950,   0E,  1A
2=        2600, 1400, 0D,  0D
3=        2400, 1075, 0C,  00

Don't know if it's accurate (voltages seem a bit weird to me) but it's all i can show for now.
Still looking at sysctl, etc.. stuff. If you want i can roll up a archive with all the stuff you mention on the bug report template and send you a link, or just tell me what you need specifically and i'll get it :)
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 18, 2010, 01:13:31 PM
Don't know if it's accurate (voltages seem a bit weird to me) but it's all i can show for now.
Still looking at sysctl, etc.. stuff. If you want i can roll up a archive with all the stuff you mention on the bug report template and send you a link, or just tell me what you need specifically and i'll get it :)

What would be helpful would be the BIOS writers guide for netburst CPU's, but I am sure you don't have that.

I have some data however, so I think I will be able to make it work. I am curious which P-states you get with the older VoodooPower you said worked. At least the lowest FID should be 0x0C/12x.

The Netburst architecture has an odd vid-table, it seem to start with 0.8375 V and advancing in 0.1250 V steps. But it splits at 1.1000 V.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 18, 2010, 02:05:26 PM
Your assurance is right ;D googlin' for Netburst... ooh, it's a Intel CPU microarchitecture.
Yeah, voltages may seem strange to me but i know that can be normal.
Ok, let me check pstates with the old PowerMini.. that crossed my mind.

Update:
On Leo atm, VoodooPower & vpower app:
Quote
Kernel Extension Name: VoodooPower.kext
Version: 1.2.3 Release Feb 15 2009 17:34:22
CPU: 2x Intel(R) Pentium(R) D CPU 3.00GHz
Technology: Intel Netburst New Desktop/Server [00000f65]
P-States : P00 FID 0f DID 00 VID 27
             : P01 FID 0e DID 00 VID 20
             : P02 FID 0d DID 00 VID 13
             : P03 FID 0c DID 00 VID 06
.. long time since i last used vpower. Back to Snow.. VoodooPower is being a bitch there...

Update again:
I was assuming VoodooPower would work on Snow but, no. I don't even get kp, machine just freezes!
Is the info from Leo useful? If not i'm in the dark :P
Title: Re: VoodooPowerMini Snow feedback
Post by: rockingturtle on April 19, 2010, 12:43:09 PM
Hey Superhai,

I am unable to download the beta release, it says "Go back to main".

-Rock
Title: Re: VoodooPowerMini Snow feedback
Post by: Blackosx on April 19, 2010, 01:17:26 PM
Try it again, all the downloads are working from here.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 19, 2010, 05:09:34 PM
Funny this Netburst stuff... i read about Pentium D several times over the past and i don't remember the word at all!?
Must be getting old...
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 27, 2010, 09:56:11 AM
Made some changes. New download - > beta 2.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 27, 2010, 12:59:27 PM
Ok, still no stepping but, there's some life on beta3... cpu frequency shows 3000 instead of 3001 (the usual), based on sysctl output (and cpu-x.. same source). Can't trust Sys Profiler anymore on this as Chameleon is now reporting the correct frequency (it used to show 3001). Your voodoo always "corrected" this.
Report attached (just x86_64), all betas + VoodooPower output on Leo.
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 27, 2010, 02:22:29 PM
the kext do not update the visible frequency anymore, as it is largely unnecessary code. I uploaded a beta 4 which have a minor change and added a line for the log which show info for each load iteration (as it is a debug version it is updating every 5 second). Which should give some hints to what really happens and not happens.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 27, 2010, 04:03:50 PM
hum.. it's not giving that much output. All i got from it so far was:
Code: [Select]
Apr 27 14:41:08 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C26 Idle (440/500) P-State 1 Trend Down Fail 0
on previous reboot and
Code: [Select]
Apr 27 14:55:46 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C26 Idle (439/500) P-State 1 Trend Down Fail 0
Apr 27 14:55:51 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C21 Idle (79/500) P-State 0 Trend Up Fail 0
Apr 27 14:56:01 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C24 Idle (406/500) P-State 1 Trend Down Fail 0
on this last. Been trying to poke it but, it's offering resistance :)
Going for lunch.. brb

Edit.. last boot:
Code: [Select]
Apr 27 18:11:44 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C27 Idle (393/500) P-State 1 Trend Down Fail 0
Apr 27 18:11:49 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C21 Idle (100/500) P-State 0 Trend Up Fail 0
Apr 27 18:11:59 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C24 Idle (449/500) P-State 1 Trend Down Fail 0
Apr 27 21:17:44 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C21 Idle (83/500) P-State 0 Trend Up Fail 0
Apr 27 21:24:19 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C25 Idle (405/500) P-State 1 Trend Down Fail 0
Apr 27 21:26:44 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C21 Idle (111/500) P-State 0 Trend Up Fail 0
Apr 27 21:32:59 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C24 Idle (418/500) P-State 1 Trend Down Fail 0
Apr 27 22:32:29 AziLandz kernel[0]: Voodoo[0xFFFFFF7F80D2E9D3] event:  405: State: 0C27 Idle (403/500) P-State 1 Trend Down Fail 0
it's trying to do it...

Same stuff on i386 kernel:
Code: [Select]
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x42129B8E]           probe:   96: CPU Family 15 Model 6
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x4212A4B0]           probe:  246: Netburst calculations
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x4212A4E2]           probe:  247: Maximum fid: 0F vid: 27
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x4212A524]           probe:  248: Minimum fid: 0C vid: 20
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x4212A612]           probe:  259: P-state 0 - 0F27
Apr 27 22:20:05 AziLandz kernel[0]: Voodoo[0x4212A612]           probe:  259: P-state 1 - 0C20
Apr 27 22:20:05 AziLandz kernel[0]: DSMOS has arrived
Apr 27 22:20:05 AziLandz kernel[0]: VoodooPowerMini 1.2 (C) 2009-10 Superhai, All Rights Reserved. Apr 27 2010 14:17:15 DEBUG 32 bit

Apr 27 22:20:15 AziLandz kernel[0]: Voodoo[0x4212AA3E] event:  405: State: 0C25 Idle (391/500) P-State 1 Trend Down Fail 0
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 28, 2010, 09:28:54 AM
Try the new (same beta 4)
Title: Re: VoodooPowerMini Snow feedback
Post by: mitchde on April 28, 2010, 12:46:18 PM
I always get error tryinf dl of latest b4.
I used official beta dl main link.
http://superhai.com/beta-darwin.html (http://superhai.com/beta-darwin.html)
Downloading file?
Please do not link directly to my binary files. Use the corresponding download page as link.

Thanks

Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 28, 2010, 02:11:21 PM
Hello there...
Superhai, this one is reporting fine.
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 28, 2010, 03:23:24 PM
Hello there...
Superhai, this one is reporting fine.

Good. But it seem to not change the fid properly, it look like you something that is interfering there.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 28, 2010, 03:41:11 PM
yeah... What can it be? another kext? Kextstat attached.
Those are the usual kexts loaded, except for VoodooHDA, removed it from S/L/E for test purposes. Using a bin patched AppleHDA for now.

Edit: i have a Snow test install. It's a clean install, just OS X installed and updated. The driver does the same there so, if something is interfering it must be on the system or on my patches!?
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 28, 2010, 08:57:53 PM
yeah... What can it be? another kext? Kextstat attached.


No idea. It could be something in bios.
Title: Re: VoodooPowerMini Snow feedback
Post by: Azimutz on April 29, 2010, 06:19:34 PM
Ok.. i got back to Snow test system, this time with minimal patches; just booter, kernel, fakesmc and dsdt. On dsdt the only permanent patch is the one for cmos reset; when using AppleHDA i add the generic   injection; occasionally i also use graphics injection there; none of the usual hpet/rtc patches are applied.
The driver behaves the same way so, i think we can exclude the patches from the problem too!?

I gave a look at the bios setup; i found a while ago i can get into setup also with F11 (not mentioned by HP any were) instead of F1 and   that brings up some more options, mostly on the Power pane. These are the options and default values:
  - After AC Power Failure: Auto
  - WOL (wake on LAN) in S4: Enabled
  - APIC: Enabled (not changeable)
  - S3: Enabled
  - XD (Execute Disable): Enabled
  - C1E (Enhanced Halt State): Enabled
  - EIST: Enabled
  - Intel(r) 64: Enabled (not changeable)
  - Quick Resume Technology: Enabled
 
I have just minimal knowledge about this stuff and i understand most of it has nothing to do with EIST!? but, i do need to do some reading.. that's what i'm doing atm, checking your stuff about EIST and reading about the other stuff all over the net. I don't use sleep/hibernation (in fact it's not working properly on OS X) or any other related stuff like Quick Resume; the machine is either On or Off, no "gray areas" here; that's why EIST is important to me and why i'd like to have it working properly. 90% of the time, i don't need the CPU running at "full throttle" so the power consumption reduction and consequent lower temperatures are welcome!
For now i disabled Quick Resume and C1E. Will post results later but, FID still seems screwed...
 
  Two questions (for now):
  - are you sure this is caused by something on the system/machine?
  - are those P-States i got from Leo with vpower the correct ones? (post #7)
 
Another thing... the only reason why i started using VoodooPowerMini instead of VoodooPower was because, i don't really need the extra functionality present on VoodooPower "all the time" and i still could easily check CPU freq. Now, if future PowerMini will "not update the visible frequency anymore", i'm gonna start "hating" it ;)
Title: Re: VoodooPowerMini Snow feedback
Post by: Superhai on April 29, 2010, 07:17:57 PM
 
  Two questions (for now):
  - are you sure this is caused by something on the system/machine?
  - are those P-States i got from Leo with vpower the correct ones? (post #7)
 
Another thing... the only reason why i started using VoodooPowerMini instead of VoodooPower was because, i don't really need the extra functionality present on VoodooPower "all the time" and i still could easily check CPU freq. Now, if future PowerMini will "not update the visible frequency anymore", i'm gonna start "hating" it ;)

I cannot be sure what is the cause, but I can't see why there should be something else. It could also be that there is no BIOS option for some low-level power management. I wouldn't disable C1E unless it breaks something.

And the P-States are not so very interesting, the fid value is correct, but the vid values are more arbitrary.

I guess you should prepare yourself for some hating then. VoodooPowerMini is supposed to be as small as possible, and not really visible in any way. That is also the reasoning for no configuration. No bells -- no whistles -- but just work.