Jump to content

ESXi 5 Mac OS X Unlocker

Donk

By Donk,
in Multi-booting and Virtualisation

 Share
817 posts in this topic

Recommended Posts

  • 3 weeks later...
  • 1 month later...
Docop
Docop
Posted
Posted

Hi

 first thanks for all the details and great work on that project. I might perhaps had a small problem in regards of the version or process to do the install...

 

I'm over esxi5.5u1, using the esxi windows client, I cannot boot directly on maverick 10.9 dmg, as i got the apple logo out because of the check...
I've install the unlock 1.3 and have a scratchdisk label as .locker on my datastore, being an SSD.

 What I've found is I install OsX Lion without any problem. Do all the software update, enable vnc logging and install the VMtools patched.

 All is ok. But if I boot again on the Maverick dmg directly, with vmware boot option, it still not boot on. I try to change the boot option
during the loading of Lion, but I can't got it. Supposed to be apple+C. See in esxi as Windows keyboard button+c.

 So was it possible to install the update offline ? Or i need to create an Apple id and re download maverick and install it live from
Lion ?

Thanks

 

-My sys: Intel VT-d enable, VM as osx10.7, VM machine ver.8, 2Vcpu, hdd 40gb thick eager, net100 inet, 2gb ram, esxi 5.5u1 full fresh inst

Link to comment
Share on other sites
MSoK
MSoK
Posted
Posted

Hi

 

 first thanks for all the details and great work on that project. I might perhaps had a small problem in regards of the version or process to do the install...

 

I'm over esxi5.5u1, using the esxi windows client, I cannot boot directly on maverick 10.9 dmg, as i got the apple logo out because of the check...

I've install the unlock 1.3 and have a scratchdisk label as .locker on my datastore, being an SSD.

 What I've found is I install OsX Lion without any problem. Do all the software update, enable vnc logging and install the VMtools patched.

 

 All is ok. But if I boot again on the Maverick dmg directly, with vmware boot option, it still not boot on. I try to change the boot option

during the loading of Lion, but I can't got it. Supposed to be apple+C. See in esxi as Windows keyboard button+c.

 

 So was it possible to install the update offline ? Or i need to create an Apple id and re download maverick and install it live from

Lion ?

 

Thanks

 

-My sys: Intel VT-d enable, VM as osx10.7, VM machine ver.8, 2Vcpu, hdd 40gb thick eager, net100 inet, 2gb ram, esxi 5.5u1 full fresh inst

Docop,

 

Regardless of whether you want to upgrade Lion or do a clean install of Mavericks you will need to download the Mavericks Install files from Apple, it should be an option when you do a Software Update. Once downloaded you can then just install over your existing Lion, or create a Mavericks DMG file to do a clean install. However creating a bootable DMG from the installation files stored on your Lion instance is not straight forward, see post 1 and post 2 in the topic below, they should answer all or most of your questions:

 

http://www.insanelymac.com/forum/topic/290949-how-to-install-os-x-10x-snow-leopard-lion-mountain-lion-and-mavericks-in-vmware-workstation-910-or-player-56/?p=1936143

 

P.S. I assume you meant VT-x is enabled in BIOS, VT-d is nice to have but not a requirement to run OS X as a VM under ESXi.

Link to comment
Share on other sites
Docop
Docop
Posted
Posted

Hi

 

 thanks a lot for the missing details. Basically i was still trying with only dmg... so just boot Mav with the iso,, and Voila. all run fine. I need to check further, but after the vmtools installed it's look slower. Anyway

 

Thanks again

Link to comment
Share on other sites
gergito
gergito
Posted
Posted

Hello there,

please excuse for my bad English as Im not a native speaker.

 

I need help to passthrough Quadro 6000 into Mavericks 10.9.4 VM.

 

My setup is:
CPU: Intel Xeon E3 1230v2
MB: Asrock Z77 Extreme4
Primary Video Card: GTX770 hard modded to GRID K2
Secondary Video Card: GTX480 soft modded to Quadro 6000

Hypervisor: VMWare ESXi 5.5 U1
Guest OS: Mavericks 10.9.4

 

First i'm testing this video cards and they both are passthrough capable when running into Windows VM guest.

 

Now i try to pass them with Mavericks VM guest.
First i successful install ESXi 5 Mac OS X Unlocker 1.3.0.
Next i download Mavericks 10.9.3 app and and convert it to bootable iso.

 

I tried several installations.
From vCenter i create Apple Mac OS X 10.9 VM (i tried with VM versions: 10,9 and 8) and attach bootable iso from datastore.
In Video Card section on Virtual Hardware 3D Graphics checkbox are greyed!
Also i adding PCI devices (passthrough 02:00.0 | NVIDIA Corporation GF100GL [Quadro 6000] and 02:00.1 | NVIDIA Corporation GF100 High Definition Audio Controller)
If PCI device are added i cannot boot and stuck on grey screen with reboot loop.
Trying to start with -v, PCIRootUID 0,1 GraphicsEnabler no,yes  npci... all combinations without success.
Booting stuck on [iOBluetoothHCIController][start] -- completed, and VM self rebooted.
After removing PCI devices, installation start and i can complete setup process.
Next i shutdown VM and again attach PCI Devices and boot, but without success.
Removing PCI Devices and booting into Mavericks, and create upgrade to 10.9.4.
Shutdown and attach PCI, not booting.
Detach PCI and boot.
Next install <<censored>> Mavericks Edition, and select DSDT Free, GraphicsEnabler=Yes and PCI Root ID Fix -> Rebuild and install.
After booting not loading <<censored>> screen, loading default grey screen with apple logo.
At this step if attach video card, i can boot normally and in Hardware -> Graphics/Displays show "NVIDIA Chip Model" without signal to my monitor, card not working properly.
VM cant properly initialize GPU:
Chipset Model: NVIDIA Chip Model
Type: GPU
Bus: PCIe
PCIe Lane Width: x32
VRAM (Total): 128 MB
Vendor: NVIDIA (0x10de)
Device ID: 0x06d8
Revision Id: 0x00a3
ROM Revision: preset 1.0.0
At next step i install Nvidia quadro drivers (Quadro & GeForce Mac OS X Driver Release 334.01.02) for Quadro 4000 for Mac. No signal to monitor.
Install and Cuda last version, no signal to monitor.
Installing and VMWare tools .. no signal.
Installing and vmsvga2 drivers for VMWare video card.
Card remains to be initialized as NVIDIA Chip Model with 128 MB and no signal to attached monitor.
Also tried to attach GTX770 (GRID K2) to VM, but will be recognized in the same manner - NVIDIA Chip Model with 128 MB (with different Device Id...)

 

Can you give me an idea what to do to run it?
I'm desperate :(

 

Thank you in advance!!!

Link to comment
Share on other sites
MSoK
MSoK
Posted
Posted

gergito,

 

A couple of things, if you are using the Unlocker do not use vCenter, always attach directly to the unlocker host using the vSphere client.

 

Also I thing you will struggle to get the Quadro 6000 to work as it is not supported by Apple, have a look at the compatibility list mentioned in the following post:

 

http://www.insanelymac.com/forum/topic/267296-esxi-5-mac-os-x-unlocker/?p=2006727

 

Finally have a look at this topic it may help:

 

http://www.insanelymac.com/forum/topic/285733-graphics-acceleration-in-esxi-51-and-os-x-mountain-lion-vm/?p=1883583

Link to comment
Share on other sites
  • 2 weeks later...
keithl
keithl
Posted
Posted

I just successfully got 1.2.0 working with ESXi5.5.0U2 (build 2068190) and OSX 10.9.4 (build 13E28).  I tried 1.3.0 and it failed (specifically, stuck on Apple logo with spinning "wait" wheel).

 

Info:

New ESXi box (i7 4790) is replacement for outdated one (Core 2 Duo)

Completely fresh install to 4GB USB on new diskless h/w.

Datastore is NFS and was previously attached to the former ESXi box (5.5.0)

New ESXi box has different hostname and newer H/W but otherwise same.

Followed instructions (especially persistent scratch)

Installed 1.3.0 and verified with ssh looking at /etc/rc.local.c/local.sh

Added VM to inventory (was shutdown when old ESXi was decommissioned)

Powered up VM

Got warning from ESXi (was this moved or copied) and said moved.

Waited at Apple logo screen for a few hours

 

Uninstalled 1.3.0 rebooted and verified (local.sh)

Reinstalled 1.3.0 rebooted and verified

No change

 

Uninstalled 1.3.0 rebooted and verified

Installed 1.2.0 rebooted and verified

Worked

 

Note one other very interesting thing.  On my old ESXi box, I always saw the VM "Resource Allocation" CPU as 0 and Memory as -1.  Never caused a problem, so I ignored it.  When I made the new box with 1.3.0, it showed believable values (even though I was stuck at the logo).  When I got it working with 1.2.0, I was back to 0 and -1.

Link to comment
Share on other sites
Donk
Donk
Posted
  • Author
Posted

I just successfully got 1.2.0 working with ESXi5.5.0U2 (build 2068190) and OSX 10.9.4 (build 13E28).  I tried 1.3.0 and it failed (specifically, stuck on Apple logo with spinning "wait" wheel).

 

Info:

New ESXi box (i7 4790) is replacement for outdated one (Core 2 Duo)

Completely fresh install to 4GB USB on new diskless h/w.

Datastore is NFS and was previously attached to the former ESXi box (5.5.0)

New ESXi box has different hostname and newer H/W but otherwise same.

Followed instructions (especially persistent scratch)

Installed 1.3.0 and verified with ssh looking at /etc/rc.local.c/local.sh

Added VM to inventory (was shutdown when old ESXi was decommissioned)

Powered up VM

Got warning from ESXi (was this moved or copied) and said moved.

Waited at Apple logo screen for a few hours

 

Uninstalled 1.3.0 rebooted and verified (local.sh)

Reinstalled 1.3.0 rebooted and verified

No change

 

Uninstalled 1.3.0 rebooted and verified

Installed 1.2.0 rebooted and verified

Worked

 

Note one other very interesting thing.  On my old ESXi box, I always saw the VM "Resource Allocation" CPU as 0 and Memory as -1.  Never caused a problem, so I ignored it.  When I made the new box with 1.3.0, it showed believable values (even though I was stuck at the logo).  When I got it working with 1.2.0, I was back to 0 and -1.

 

Can you list the contents of the /scratch folder please when 1.3.0 is installed?

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

Will have that for you in a couple of hours.  What do you want, just an ls -l?

 

Any insight into the Resource Allocation stuff.  Not a big problem, but it would be nice to have.  Note that the usage graphs show correctly.  Very odd.

 

Also, why would I want to use 1.3.0 instead of 1.2.0?  I saw msgs above about fitting into available space better with some builds?


Quick note, I remember seeing a folder "osx" in the /scratch recently and it had some unlocker files and I see now that they're still there!  Don't know if those were from 1.3.0 or if they are from the later install (and run) of 1.2.0  Do you still want me to uninstall 1.2.0 and install 1.3.0?  If so, should I delete that folder between installs?


Oh yeah, one other, perhaps trivial, thing.  When I had 1.3.0 running, ESXi claimed the VMware tools were outdated (unsupported) yet they worked even though I was stuck at the logo screen (I could shut down host).  When I ran 1.2.0, ESXi said they were current.  Slim possiblility that an "upgrade" of the tools took place automatically, but I did turn that off part way through my first 1.3.0 testing.

 

Just giving you every piece of the puzzle...

 

P.S. Thank you and all others who helped (Zenith432, MSoK etc.) for an outstanding tool!

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

OK.

 

Uninstalled 1.2.0 (now no unlocker installed).  VMware tools say current but not running.  Stuck on logo (as expected).  Resource allocation ("consumed") showing correctly.

 

Installed 1.3.0.  VMware tools say current and running.  Stuck on logo.  Resource allocation ("consumed") showing correctly. Looks like it gets a bit farther (running tools) but otherwise looks like unlocker not working.

 

/scratch listing attached.

 

 

 

scratch.txt

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

Uninstalled 1.3.0 (now no unlocker installed).  VMware tools say current and running (I think I didn't wait long enough the time before).  Stuck on logo (as expected).  Resource allocation ("consumed") showing correctly.

 

Installed 1.2.0 VMware tools say current and running.  OSX boots fine.  Resource allocation ("consumed") showing 0 and -1

 

Totally back to where I was this morning, as expected.  Interesting possibility, though. 

 

My 1.2.0 install folder contains only 4 files: Unlocker.ESXi, install.sh, uninstall.sh and readme.txt.  No tools files from 1.2.0.

 

My 1.3.0 is a tree as follows: tree root has readme.txt and folders esxi and tools.  Folder esxi has install.sh, uninstall.sh and local.sh.  Folder tools has darwin.iso and darwin.iso.sig  The relevant difference (other than readme) is that 1.3.0 has the tools folder as well and 1.2.0 doesn't or, more likely, the tools are NOT in the install folder, they are in a sister folder (../tools/)


Typical.  Just rebooted again and, for the first time in a year, I have Resource Allocation working.  Not quite exactly back to where I was, but the rest still stands.  (note I rebooted a few times when I first tried 1.3.0).

Link to comment
Share on other sites
Donk
Donk
Posted
  • Author
Posted

OK so let's just check this out one step at a time. The expanded zip files should look like this when run ls -R:

 

Unlocker 1.2.0

 

.:
esxi  linux  osx  readme.txt  src  tools  windows
./esxi:
install.sh  uninstall.sh  Unlocker.ESXi
./linux:
install.sh  uninstall.sh  Unlocker.Linux32  Unlocker.Linux64
./osx:
install.sh  uninstall.sh  Unlocker.OSX
./src:
build.sh  clean.sh  Makefile  Unlocker.cpp  Unlocker.sln  Unlocker.vcxproj
./tools:
darwin.iso  darwin.iso.sig
./windows:
install.cmd  uninstall.cmd  Unlocker.exe

Unlocker 1.3.0

 

.:
esxi  linux  osx  readme.txt  src  tools  windows
./esxi:
install.sh  local.sh  uninstall.sh
./linux:
install.sh  uninstall.sh  Unlocker.Linux32  Unlocker.Linux64
./osx:
install.sh  uninstall.sh  Unlocker.OSX
./src:
build.sh  clean.sh  Makefile  Unlocker.cpp  Unlocker.sln  Unlocker.vcxproj
./tools:
darwin.iso  darwin.iso.sig
./windows:
install.cmd  uninstall.cmd  Unlocker.exe

First confirm that these are the same folders you see on your system.

 

Next the tools do not make any difference as the scripts do nothing with them on ESXi. ESXi comes with the darwn.iso and they are only there for Workstation which does not have them shipped with the installer.

 

The resource details coming through is probably a symptom of the 1.2.0 way of patching where a new overlay VisorFS tar disk was added to replace the default boot vmx files. This caused memory errors which could be manifested in different way especially if connecting via vCenter Server. This is why the new way of patching was developed.

 

So next steps are:

 

1. Ensure 1.2.0 and 1.3.0 are no longer on the system by running both uninstall.sh commands.

2. Reboot

3. Check for any 1.2.0 remnants - if  /bootbank/darwin.vgz is still present then delete it

4. Check for 1.3.0 remnants - delete /etc/local.d/local.sh and delete /scratch/osx folder

5. Reboot

 

Now expand unlocker 1.3.0 zip and check matches the folder structure above. Run the installer and reboot and check to see if the OS X VM starts. If does not boot then the next step is to post the VMX and vmware.log files as attachments. 

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

1. ran 1.2 uninstall.sh then 1.3 uninstall.h

2. rebooted

3. bootbank.Darwin.vgz not present

4. /etc/local.d/local.sh not present.  /etc/rc.local.d/local.sh was original (comments and exit(0)), deleted /scratch/osx folder

4a. deleted 1.2.0 and 1.3.0 previously uploaded unzipped files

5. rebooted

6. uploaded unlock-all-v130.zip to /vmfs/volumes/MainDatastore/HostFiles/unlock-all-v130.zip

7. from ssh cd /vmfs/volumes/MainDatastore/HostFiles/

8. unzip unlock-all-v130.zip and verify files as described above

9 cd /unlock-all-v130/esxi then ./install.sh

10. rebooted

 

Still get logo screen only.  vmx and VMware.log files attached.

TestFiles.zip

Link to comment
Share on other sites
Donk
Donk
Posted
  • Author
Posted

1. ran 1.2 uninstall.sh then 1.3 uninstall.h

2. rebooted

3. bootbank.Darwin.vgz not present

4. /etc/local.d/local.sh not present.  /etc/rc.local.d/local.sh was original (comments and exit(0)), deleted /scratch/osx folder

4a. deleted 1.2.0 and 1.3.0 previously uploaded unzipped files

5. rebooted

6. uploaded unlock-all-v130.zip to /vmfs/volumes/MainDatastore/HostFiles/unlock-all-v130.zip

7. from ssh cd /vmfs/volumes/MainDatastore/HostFiles/

8. unzip unlock-all-v130.zip and verify files as described above

9 cd /unlock-all-v130/esxi then ./install.sh

10. rebooted

 

Still get logo screen only.  vmx and VMware.log files attached.

 

Can you change the guestos value as below and try again?

 

guestOS = "darwin13-64"

Link to comment
Share on other sites
Donk
Donk
Posted
  • Author
Posted

tried again with 13-64 instead of 11-64.  no change.

 

log attached

 

Can you try adding this line to the VMX file please? It may be the CPU is not being identified by OS X. This masks it to a known compatible Xeon CPU. Does not affect performance just ensures some of the internal CPUID checking code is passed in the xnu kernel.

 

cpuid.1.eax = "0000:0000:0000:0001:0000:0110:1010:0101"

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

Added cpuid.1.eax.  First tried with darwin13-64 (complains about newer version of OS with older version of VM hardware but still tries) then with darwin11-64 (doesn't complain at all).  Both stop at logo.

 

Note I couldn't add the line cpuid.1.eax in VSphere client (no complaint, just doesn't save it).  I had to use ssh and vi.  Note I only use the windows VSphere client, no vmotion, FT, HA, vsphere web client etc.

 

Currently setup to use the cpuid line and darwin11-64.  If you want me to try something else, let me know what you want Darwin and cpuid to be.

 

Reminder: it works under 1.2.0 (so I'm not dead in the water!)

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

As before, using 1.3.0 unlocker, guestOS = "darwin11-64" and cpuid.1.eax = "0000:0000:0000:0001:0000:0110:1010:0101", but with Mavericks 10.9.5 (13F34) and boot logging.

 

I have captured most of the boot logging (boot -v) by doing overlapping screen shots of the VSphere Client vm console.  The missed the very first part (too fast) and hang at the very last line. Without the -v, they hang with the apple logo and wait ball.

 

Any suggestions as to what to try next?

 

PS, is there a better way to capture the boot log?

Link to comment
Share on other sites
keithl
keithl
Posted
Posted

Hmm.. My setup is pretty plain.  I don't even use vCenter.  Could you post screenshots of your boot.  I'm really not sure of what matters in the boot, but I could compare and focus on the differences.

 

I'm using a 4790 CPU, GPU on processor, a Z97 mobo, no disk (just an external NFS datastore and booting from USB).  The only extra card is an Intel PCIx NIC (CT).  I don't have any passthru hardware configured in esxi (but I do have vt-d enabled in bios - maybe I will try disabling).  Maybe some other bios thing (APM? mobo hardware enabled/disabled? processor embedded graphics?).  Something simple I'm sure, but I don't know how to find it.

 

Also, 1.2.0 works.

Link to comment
Share on other sites
 Share
Go to topic listing
×
×
  • Create New...