833 replies to this topic

[maleorderbride]

mrmojorisin17, on Apr 14 2011, 09:57 AM, said:

Thanks for explanations maleorderbride
What kexts are you "waiting" for?

AppleLPC, SMBusController, and AppleAHCIPort from MBP 2011 10.6.6 restore DVD or directly from a MBP 2011.  Really, the LPC is the most important I think.  I might find I need more as this progresses.

My semi-rant at DHP was a bit out of place as well.  Too much drinking and not enough sleep last night!  Thanks for doing this DHP and I can't wait to try out the finished product.

[mrmojorisin17]

I respect DHP's opinion.
You can maybe go to an Apple store and take kexts you need

[Time2Retire]

maleorderbride, on Apr 14 2011, 07:51 PM, said:

Chameleon RC5 does not support Sandy Bridge yet.  The only widely available bootloader that does is AnVal.  ^Andy^ created a merged branch of AnVal and Kabyl's ATI RC5, but enabling C/P states is still not automatic.  So, we are waiting for AnVal or Chameleon/Mozodojo to add automatic P/C state generation for the latest generation still.

RevoBoot is 'widely available' and is currently the only, really, boot loader with real Sandy Bridge support. There's no need for any other boot loader, except for people who want a GUI to select a target drive. Which is not part of my mission, neither is C/P/T -state support. Can be done in the ACPI tables and that is what I use. Either use RevoBoot (for now) or wait.

maleorderbride, on Apr 14 2011, 07:51 PM, said:

I am attempting to manually add my own P/C states via DSDT or SSDT, but I still can't actually test native power management until I get the three kexts I requested of DHP.

There are different ways of obtaining the _PSS object data.

1.) Boot with RevoBoot with the PATCH_ACPI_TABLE_DATA directive set to 0 and to get them with IORegistryExplorer, which as you know is freely available from Apple's (get Xcode).

2.) Use any kind of Linux flavor to boot with and extract the ACPI tables with acpidump.

maleorderbride, on Apr 14 2011, 07:51 PM, said:

I assume that is the same approach that DHP is doing (or having revo do).  It would be nice if she would post the kexts so I could work on it as well.  My torrent is taking another 2 weeks! >_<

Please. Do not talk about torrents in threads where I work on stuff or I will be forced to quit.

maleorderbride, on Apr 14 2011, 07:51 PM, said:

I don't believe there is any part of the Apple EULA that forbids people from uploading a kext.  While it certainly does prohibit what we are all deriving from this website.  From here, her unwillingness to share 8MBs of files seems to be a moral contradiction at best.  Maybe I am missing something.

Here's another thing I will not do; Write installation guides. I also will not share information that I promised not to share (think NDA's here).

Please be grateful with the little things I do around here, because that is everything you ever get to see from me.

[mrmojorisin17]

DutchHockeyPro, on Apr 14 2011, 09:14 PM, said:

RevoBoot is 'widely available' and is currently the only, really, boot loader with real Sandy Bridge support. There's no need for any other boot loader, except for people who want a GUI to select a target drive. Which is not part of my mission, neither is C/P/T -state support. Can be done in the ACPI tables and that is what I use. Either use RevoBoot (for now) or wait.

I would like to try to use RevoBoot but is too much complicated for my knowledge.
If in the future RevoBoot will be more "approachable" for me I'll try it for sure
For the moment I'm compelled to use Chameleon.

Thanks for what are you doing here DHP.

Cheers

[maleorderbride]

Well, I can boot without NullCPUPowerManagement and have a few P states defined.  It clocks down to 1.7GHz and the temp drops accordingly.

Sleep and NPM still need more work though.

[Time2Retire]

maleorderbride, on Apr 15 2011, 02:49 AM, said:

Well, I can boot without NullCPUPowerManagement and have a few P states defined.  It clocks down to 1.7GHz and the temp drops accordingly.

Sleep and NPM still need more work though.

Awesome. Great news. What did you do?

p.s. My ssdt_usb.dsl need a bit of work as dgsga ran into an issue - booting stops a few lines into the process. I hope to be able to locate and fix this problem a.s.a.p.

[buoo]

maleorderbride, on Apr 15 2011, 01:49 AM, said:

Well, I can boot without NullCPUPowerManagement.

DutchHockeyPro, on Apr 15 2011, 02:27 PM, said:

Awesome. Great news. What did you do?

[maleorderbride]

DutchHockeyPro, on Apr 15 2011, 05:27 AM, said:

Awesome. Great news. What did you do?

p.s. My ssdt_usb.dsl need a bit of work as dgsga ran into an issue - booting stops a few lines into the process. I hope to be able to locate and fix this problem a.s.a.p.

I did a lot of things and I am afraid I did not take adequate notes.  Starting over with a fresh install this weekend so that I can pin down what was actually necessary and what was simply incidental.  I unfortunately have a busy weekend, so maybe you can get your build booting as you can probably make more of it then I can.

I think your DSDT is the most important piece so far.  With my own DSDT+my own SSDT (edited with the normal X58 edits from d00d's thread) I can boot without disabling NPM, but get no changes in clock or voltage.  WIth your DSDT+SSDTs, I see it downclock to 1.6GHz, 1.7GHz, and then 3.4GHz.  Nothing in between.  Voltage is either .7 or the full 1.3.

Also, the 10.6.6 MBP +10.6.7 MBP update result in a panic a few lines in, but my 10.6.3->10.6.7 combo update is the working booting volume.  Perhaps this means I am just sidestepping what we really should get working thorough.

With your DSDT+your two SSDTs I was able to boot without a disabler and see voltage/clock changes (with MSR Tools); however, I also started adding newer versions of each kext that I thought might help with AppleIntelCPUPowerManagement stalling out.  So I am not positive which kexts need to be with your DSDT.  I assume newer and more of them is probably the best.

Sorry, like I said, this is all a mess still.

Additionally, there is someone on the Tony Mac forums that was able to do this as well, but has not posted any information.  He actually reports that AppleIntelCPUPowerManagement+client were both successfully loaded.  Mine don't load, even though HPET and LPC do, they just don't cause a crash.

I sent a PM, but no response.

Also, I am still using AnVal.  I had to download the latest Xcode in order to try to use RevoBoot and haven't bothered rebooting my main computer yet and making it yet.

I am using the P8p67 as well.

[elitee]

I'm getting a sandybridge system, specifically the P8P67 (Rev 3.) Vanilla, and I was wondering if anyone's had any luck with this DSDT? I suspect it will work fine, as the boards are nearly identical, but I was curious in advance

[elitee]

maleorderbride, on Apr 15 2011, 08:08 PM, said:

With your DSDT+your two SSDTs I was able to boot without a disabler and see voltage/clock changes (with MSR Tools); however, I also started adding newer versions of each kext that I thought might help with AppleIntelCPUPowerManagement stalling out.  So I am not positive which kexts need to be with your DSDT.  I assume newer and more of them is probably the best.

Sorry, like I said, this is all a mess still.

Additionally, there is someone on the Tony Mac forums that was able to do this as well, but has not posted any information.  He actually reports that AppleIntelCPUPowerManagement+client were both successfully loaded.  Mine don't load, even though HPET and LPC do, they just don't cause a crash.

I sent a PM, but no response.

Also, I am still using AnVal.  I had to download the latest Xcode in order to try to use RevoBoot and haven't bothered rebooting my main computer yet and making it yet.

I run into the same issue,

com.apple.kextcache[314]	kxld[com.apple.driver.AppleIntelCPUPowerManagement]: The following symbols are unresolved for this kext:
com.apple.kextcache[314]	kxld[com.apple.driver.AppleIntelCPUPowerManagement]: 	_lapic_start

LPC isn't loading, but I haven't downloaded the updated SB kext's from the 10.6.7 update, so I will try that. I'm using AnVal as well, I haven't tried RevoBoot yet, but I will use RevoBuilder later to ease the process. Your DSDT seems to work fine for my board after a couple changes, and the SSDT is being loaded (after a couple changes for the i5-2500k), but it seems to not want to load CPUPM using MacPro5,1

[Time2Retire]

One of the moderators is currently testing the latest DSDT/SSDT mods with his 2500K and I will attach the files when he confirmed that they are working for him.

I also located a bug in RevoBoot that triggered a hang early in the boot process, and this patch will be pushed into the github repository later today (scrax need to do some testing with it).

[elitee]

DutchHockeyPro, on Apr 22 2011, 02:42 PM, said:

One of the moderators is currently testing the latest DSDT/SSDT mods with his 2500K and I will attach the files when he confirmed that they are working for him.

I also located a bug in RevoBoot that triggered a hang early in the boot process, and this patch will be pushed into the github repository later today (scrax need to do some testing with it).

Sounds great, in the meantime I will wait on RevoBoot and install it later and try to fix getting Vanilla sound to work with the DSDT + Info.plist + Vanilla kext's.

For RevoBoot, would you mind posting what settings you use for the header (.h) files? Or would using RevoBuilder take care of most of it fine for SB?

[mrmojorisin17]

Hi
I didn't have time to test my new dsdt and ssdt_pr (i5-2500K) without Null and also with a MacPro mac model identifier instead of MacBook Pro.
I'll make some test next week and post here the results.

[flAked]

Good day,

I'm testing a P8P67 with i5-2500K right now, so far pretty much every function in your DSDT matches the addresses in my factory DSDT.

ssdt_usb seems identical to my board, but I'm a bit concerned about the different PCI/E lanes, needs investigating.

Currently editing my specific p-states, is there a reason for zeroing out the other values under package (0x06)?


EDIT: not sure which combination of static injection I need to specify, is there a way to tell the /DSDT.aml and /SSDT_PR.aml were loaded? Or are they always loaded after the static injection?

The PerformanceStateArray is now filled with the right values and I'm using the kexts from MBP2011, but still AppleIntelCPUPM KPs on me.

[elitee]

polyglot, on Apr 23 2011, 11:31 AM, said:

Good day,

I'm testing a P8P67 with i5-2500K right now, so far pretty much every function in your DSDT matches the addresses in my factory DSDT.

ssdt_usb seems identical to my board, but I'm a bit concerned about the different PCI/E lanes, needs investigating.

Currently editing my specific p-states, is there a reason for zeroing out the other values under package (0x06)?


EDIT: not sure which combination of static injection I need to specify, is there a way to tell the /DSDT.aml and /SSDT_PR.aml were loaded? Or are they always loaded after the static injection?

The PerformanceStateArray is now filled with the right values and I'm using the kexts from MBP2011, but still AppleIntelCPUPM KPs on me.

From what she said it won't work without using the NullCPUPM kext. I have a P8P67 (vanilla rev3), is that what you have?

I tried to change hers to match my board, but still no luck with audio or details showing up in the system profiler, is there any chance you could share your DSDT? Or are you still using hers?

[flAked]

elitee, on Apr 23 2011, 06:18 PM, said:

From what she said it won't work without using the NullCPUPM kext. I have a P8P67 (vanilla rev3), is that what you have?

I tried to change hers to match my board, but still no luck with audio or details showing up in the system profiler, is there any chance you could share your DSDT? Or are you still using hers?

I've got the same board, BIOS version 1401. I did my install using tonymac's iBoot/MultiBeast approach and it worked very well to get the system running.

I did manage to get sound working with a special enabler for that chip, but not via DSDT. I have an additional CMI8738 sound-card, so I actually deactivated the whole thing.

Regarding the DSDT, I'm still experimenting and pretty much using hers at the moment, just added my specific P-states in SSDT_PR.

I'm on and off using RevoBoot or Chameleon, but I'm not able to pinpoint why exactly IntelCPUPM is crashing.

[Time2Retire]

I will add the file I talked about – for the i5-2500(K) – this evening.

Sorry. Didn't make it due to having a stupid eye infection. Anyway. Here it is. This one includes the _PSS object data aja P-States for your CPU. All that you will have to do, for now, is to remove the extra 4 processor block declarations (only keep CPU0-CPU3 and remove the others).

Have fun.

Attached Files

•  ssdt_pr.dsl.zip   1.69K   39 downloads

[flAked]

DutchHockeyPro, on Apr 23 2011, 10:45 PM, said:

I will add the file I talked about – for the i5-2500(K) – this evening.

Sorry. Didn't make it due to having a stupid eye infection. Anyway. Here it is. This one includes the _PSS object data aja P-States for your CPU. All that you will have to do, for now, is to remove the extra 4 processor block declarations (only keep CPU0-CPU3 and remove the others).

Have fun.

Cheers!

Hope your infection gets better soon, thanks for posting regardless.

The PSS data for my i5-2005K looks like this (extracted in Gentoo), not sure if those numbers need to be converted to work in OSX (another multiplier base maybe)?

PSS: 0x3B00, 0x3A00, 0x3900, 0x3800, 0x3700, 0x3600, 0x3500, 0x3400, 0x3300, 0x3200, 0x3100, 0x3000, 0x2F00, 0x2E00, 0x2D00, 0x1000

In my current configuration those are also populated in PerformanceStateArray without any DSDT (not sure if Chameleon passes them along or a BIOS config did it as they were not present a couple of days ago).

Using a legacy kernel I can see temperatures drop 3-4°C indicating some sort of throttling while IntelCPUPM loads fine.

Using your DSDT/SSDT the kernel still panics on IntelCPUPM.

Still no happy panda, then. Ideas on what I could try next?

[Time2Retire]

polyglot, on Apr 24 2011, 12:33 PM, said:

Using a legacy kernel I can see temperatures drop 3-4°C indicating some sort of throttling while IntelCPUPM loads fine.

Using your DSDT/SSDT the kernel still panics on IntelCPUPM.

Legacy kernel? Doesn't that just block AppleCPUPowermanagement.kext from getting loaded?

I am using the vanilla 10.6.7 kernel of a MacBookPro8,3 and I think (hope really) to have found something interesting (we're working on it) that could potentially lead to vanilla Intel SpeedStep support for our Sandy Bridge CPU's

However. Since this is easter weekend, and still very nice whether... don't wait for it as I am goggled up for a plunge in our pool

Happy Easter!

[flAked]

DutchHockeyPro, on Apr 24 2011, 11:25 AM, said:

Legacy kernel? Doesn't that just block AppleCPUPowermanagement.kext from getting loaded?

Spot on! Too much coffee yesterday... it just blocks it, yes.

DutchHockeyPro, on Apr 24 2011, 11:25 AM, said:

I am using the vanilla 10.6.7 kernel of a MacBookPro8,3 and I think (hope really) to have found something interesting (we're working on it) that could potentially lead to vanilla Intel SpeedStep support for our Sandy Bridge CPU's

However. Since this is easter weekend, and still very nice whether... don't wait for it as I am goggled up for a plunge in our pool

Happy Easter!

Great, looking forward to it! Need to get some real work done anyways  

Have a nice weekend!