Haswell early reboot, Mavericks, locked MSRs, and HP Envy 15-J063CL (i7-4700MQ)

[stranger599]

I ran the pearl patch and managed to get the installer started, but I seem to be runnning into a small hickup. The installer itself seems to be overclocked or as if VSync was disabled... As in everything is operating way above normal speed. The loading cursers and status bars are moving so fast they are almost flickering. Is this normal? Currently the installer is running properly.

 

OS X 10.9

Lenovo Yoga 2 Pro

Intel core i5 4200u

Intel HD4400

ssd 128GB

 

The computer is maintaining proper temps. As in, its not burning up to the touch.

[RehabMan]

I ran the pearl patch and managed to get the installer started, but I seem to be runnning into a small hickup. The installer itself seems to be overclocked or as if VSync was disabled... As in everything is operating way above normal speed. The loading cursers and status bars are moving so fast they are almost flickering. Is this normal? Currently the installer is running properly.

 

OS X 10.9

Lenovo Yoga 2 Pro

Intel core i5 4200u

Intel HD4400

ssd 128GB

 

The computer is maintaining proper temps. As in, its not burning up to the touch.

This seems to be a problem with some Haswell laptops. Mostly Lenovo but also Dell. I don't have a clue about that one and didn't run into it with my HP. It is as if the system clock used for time intervals has too high resolution.

[Simonej]

Most Gigabyte boards are not MSR locked, so no need for replacing the kernel or KernelPm.

 

No states between idle and nominal appears to be normal for xcpm. It is that way on my MacBookAir6,2. It is that way on real iMacs (haswell or ivy), so it appears it is intentional. On a desktop, jumping from idle -> nominal is not an issue. And it appears Apple doesn't believe it is an issue on laptops either.

 

Yes I agree, but why ameris cyning had:

"11/29/13, 1:14:43 PM, P States: 8, 10, 19, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37"

 

ameris cyning system: i7 4770, Z87-D3HP, MacPro3,1 4,1 iMac13,2 14,2 , Clover with KernelPm

my system:                  i5 4670, Z87-N, MacPro5,1 + same test, Clover with KernelPm

 

Similar set-up but i haven't P states from 8 to 34, my system is working fine but I don't know why no PStates like Ameris

[RehabMan]

Yes I agree, but why ameris cyning had:

"11/29/13, 1:14:43 PM, P States: 8, 10, 19, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37"

 

ameris cyning system: i7 4770, Z87-D3HP, MacPro3,1 4,1 iMac13,2 14,2 , Clover with KernelPm

my system:                  i5 4670, Z87-N, MacPro5,1 + same test, Clover with KernelPm

 

Similar set-up but i haven't P states from 8 to 34, my system is working fine but I don't know why no PStates like Ameris

Try using MacPro3,1 to see if you can reproduce the same results. Note that was using xcpm (patched kernel, not built kernel).

 

I haven't tried yet, and running MacPro3,1 is not an option for me because I want AirPlay mirroring...

[Simonej]

Tried MacPro3,1 but nothing... sorry may I have lost a piece...xcpm is with your perl script? isn't the same of vanilla kernel+KernelPm true in config? I'm novice in this case, where can I find information about it? (xcpm)

 

Thank you for you patience and very hard work!

[RehabMan]

Tried MacPro3,1 but nothing... sorry may I have lost a piece...xcpm is with your perl script? isn't the same of vanilla kernel+KernelPm true in config? I'm novice in this case, where can I find information about it? (xcpm)

 

Thank you for you patience and very hard work!

KernelPm is same as patching vanilla kernel and placing patched version on disk.

 

But you should also verify other parts of configuration:

- verify using xcpm (AppleIntelCPU* not loaded)

- custom SSDT + Generate C/PStates from Clover (that is a bit strange, but that is what ameris cyning was doing)

- drop OEM tables

[Simonej]

But you should also verify other parts of configuration:
√ verify using xcpm (AppleIntelCPU* not loaded) - not loaded + removed, no problem
√ custom SSDT + Generate C/PStates from Clover - tested with and without generete states from clover, SSDT is from pike's script
√ drop OEM tables 

√ SSDT.aml in Clover "patched" folder

 

tested SMbios MacPro and iMac, bios setting (EIST and more), KernelPm, nothing to do... but without SSDT works better, I have 8>34>35>36>37>38>..42 with SSDT 8>34>36>38>42

Tomorrow I need to do more tests with EIST and CPU advanced options, i thinks works better Auto or Disabled.

[RehabMan]

But you should also verify other parts of configuration:

√ verify using xcpm (AppleIntelCPU* not loaded) - not loaded + removed, no problem

√ custom SSDT + Generate C/PStates from Clover - tested with and without generete states from clover, SSDT is from pike's script

√ drop OEM tables 

√ SSDT.aml in Clover "patched" folder

 

tested SMbios MacPro and iMac, bios setting (EIST and more), KernelPm, nothing to do... but without SSDT works better, I have 8>34>35>36>37>38>..42 with SSDT 8>34>36>38>42

Tomorrow I need to do more tests with EIST and CPU advanced options, i thinks works better Auto or Disabled.

That has pretty much been my results whenever I attempted to repro a claim of states between idle and nominal. So far, I've only been able to do that using non-xcpm kernel, Sandy SSDT, Sandy smbios, and 10.8.5 (patched) AppleIntelCPUPowerManagement.

 

I haven't tried the scenario above... yet (busy with a few other things lately...)

[useletters]

Hi RehabMan, in theory, if I use the patch for the vanilla kernel, having locked MSR 0xE2 is alright? My understanding is, AICPM woll not be used and it writes to 0xE2. Correct me if I'm wrong. Thank you

[RehabMan]

Hi RehabMan, in theory, if I use the patch for the vanilla kernel, having locked MSR 0xE2 is alright? My understanding is, AICPM woll not be used and it writes to 0xE2. Correct me if I'm wrong. Thank you

That is correct. The patches keep the kernel from writing to MSR 0xE2. And AppleIntelCPUPowerManagement.kext will not load, so you don't have to worry about patching it.

[RehabMan]

That has pretty much been my results whenever I attempted to repro a claim of states between idle and nominal. So far, I've only been able to do that using non-xcpm kernel, Sandy SSDT, Sandy smbios, and 10.8.5 (patched) AppleIntelCPUPowerManagement.

 

I haven't tried the scenario above... yet (busy with a few other things lately...)

Just a little update/tidbit of information. If I boot my ProBook 4540s (i5-3230m, MacBookPro9,2, custom SSDT w/ plugin-type=1, Clover, no-generate SSDT) with xcpm enabled (using patched kernel, -xcpm flag), I get pstates between idle and nominal. More testing later with the Haswell machine...

[Simonej]

Can I help with tests? (Haswell CPU)

SMBios(MacPro,iMac)+SSDT+DropOem=True+P/CStates=False,KernelPM=False+Kernel from first post, right?

[RehabMan]

Can I help with tests? (Haswell CPU)

SMBios(MacPro,iMac)+SSDT+DropOem=True+P/CStates=False,KernelPM=False+Kernel from first post, right?

I'm using patched kernel (Clover KernelPm). Now that we can patch the kernel with either KernelPm or the perl patches I created, there is little reason to use the kernel attached to post #1.

[ameris_cyning]

I'm using patched kernel (Clover KernelPm). Now that we can patch the kernel with either KernelPm or the perl patches I created, there is little reason to use the kernel attached to post #1.

 

There is still one good use for it: it is an alternative for power management concerning Haswell systems on OS X. An AppleIntelCPUPowerManagement.kext from 10.8.5 and this kernel provides a very good alternative for power management.

[lisai9093]

hi rehabman, will the old-time pmpatch for BIOS still work in Haswell? So ppl can forget about patching kernel in the future. I do see patching BIOS can solve reboot issue in Haswell, but not sure about p-state, etc.

[RehabMan]

hi rehabman, will the old-time pmpatch for BIOS still work in Haswell? So ppl can forget about patching kernel in the future. I do see patching BIOS can solve reboot issue in Haswell, but not sure about p-state, etc.

If you can pmpatch your BIOS, and are comfortable doing that, then you do not need patched kernel.

Getting power management properly configured is then a matter of getting all the right pieces into place (smbios, ssdt, etc)

[Onixs]

@RehabMan

 

What geekbench score are you getting on your i3 laptop. I also have the i3, but score is too low (3580). I have patched AppleIntelCPUPowerManagement, created SSDT from ssdtpgen.

 

Am i missing something in here?

 

Thanks

[Demonbane]

Hi Rehabman, i have used Clover's key "KernelPM" to patch my kernel. But now the AppleIntelCPU* cant not be loaded, even if i use the 10.8.5 or 10.9 version.  How should i do exactly? Should i replace the kernel that you provided and cancle the Clover's KernelPM function?

[pokenguyen]

Hi Rehabman, i have used Clover's key "KernelPM" to patch my kernel. But now the AppleIntelCPU* cant not be loaded, even if i use the 10.8.5 or 10.9 version.  How should i do exactly? Should i replace the kernel that you provided and cancle the Clover's KernelPM function?

It's normal behavior. Kernel in OP does not have XPCM so it will use AICPM, while "KernelPM" patches your kernel to avoid KP, and system can use XCPM in kernel to manage power management instead of AppleIntelCPU. Are there any reasons that you want to use AICPM instead of XCPM?

[RehabMan]

@RehabMan

 

What geekbench score are you getting on your i3 laptop. I also have the i3, but score is too low (3580). I have patched AppleIntelCPUPowerManagement, created SSDT from ssdtpgen.

 

Am i missing something in here?

 

Thanks

Whether 3580 is lower than expected depends on specifically what i3 you have, which you have not specified.

[Demonbane]

It's normal behavior. Kernel in OP does not have XPCM so it will use AICPM, while "KernelPM" patches your kernel to avoid KP, and system can use XCPM in kernel to manage power management instead of AppleIntelCPU. Are there any reasons that you want to use AICPM instead of XCPM?

 

Sorry that i'm new and don't know the stuff concretely about Hackintosh what's the difference between XCPM and AICPM? how can i figur out if my XCPM is loaded or not?

[RehabMan]

Sorry that i'm new and don't know the stuff concretely about Hackintosh what's the difference between XCPM and AICPM? how can i figur out if my XCPM is loaded or not?

XCPM is CPU PM in the kernel, mach_kernel. It is used on Macs with Hasewell. It can also be used with Ivy CPUs by using the -xcpm kernel flag. You are using XCPM if AppleIntelCPUPowerManagement is not loaded.

[Onixs]

Whether 3580 is lower than expected depends on specifically what i3 you have, which you have not specified.

Pardon me if this is off-topic in regards with my processor... I am on i3-23370M, HD3000, 4GB ram

[RehabMan]

Pardon me if this is off-topic in regards with my processor... I am on i3-23370M, HD3000, 4GB ram

i3-23370M is not a valid model number for an Intel CPU. In addition, your query is off-topic. Any CPU with HD3000 is Sandy Bridge. This thread is about Haswell (and some Ivy) XCPM only.

[kehabouter]

Hi!

 

When i use your kernel there are no instant reboots anymore, but i get stuck at this line:

 

hfs: mounted "name of disk" on device_root

 

I tried so many things, but without luck..

 

My build:

Asrock H81M-DGS mainboard

Intel G3220 Pentium

2x2GB DDR3 1333

8800GTS 512MB (Saving for 650GTX)

64GB SSD

1TB HDD

 

I've got an Macmini 5.1 smbios.plist, no dsdt and no ssdt

 

i tried all flags available (-v -f -x GraphicsEnabler=Yes PCIRootUID=1/0 USBBusFix=Yes npci=0x2000/3000 cpus=1 etc etc)

 

What can i else try?