qemu + haxm does not work to boot and run Mac OS X High Sierra #149
Comments
|
We don't plan to provide official support for running macOS guests, but we're open to pull requests for this feature. "VCPU shutdown request" indicates a panic in the HAXM driver. Could you post the HAXM driver log? |
|
The HAXM driver log can be got from Windbg. Also, would you help to provide the MacOS iso file and other files you used, as well as the method how you made them. |
|
Thanks. You can find everything below (the mac os x high sierra iso file and the log file). PS : About how I have obtained the mac os x iso file : I got it following this tutorial : https://github.com/kholia/OSX-KVM/tree/master/HighSierra https://drive.google.com/open?id=1sYLvF8tgd3baCD-IVoAJ1kpx8SggyqOq |
|
Anyway,I have attached the log file below.. |
|
Everything is ok ? did u get everything you need ?
Il giorno gio 3 gen 2019 alle ore 05:10 Huang, Jun <notifications@github.com>
ha scritto:
… The HAXM driver log can be got from Windbg. Also, would you help to
provide the MacOS iso file, as well as the method how you made it.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AAuGixcQ_emJumj47PyZtq_dd2QXTQdmks5u_YKhgaJpZM4ZmoLk>
.
--
Mario.
|
Yes, thanks for providing all the details! It seems the error has nothing to do with the macOS guest image (
Without |
|
Yes, we got them, thanks. |
|
news ? |
|
Today we received an important tip from the QEMU community (many thanks to @AlexAltea) on why OVMF doesn't yet run on HAXM: http://lists.nongnu.org/archive/html/qemu-devel/2019-01/msg08482.html If we implement the proposed changes, we should be a big step closer to getting OVMF booting. |
|
nice. |
|
any progress on this? tnx |
|
I have the same issue any progress ? |
|
do u remember me ? any progress ?
Il giorno mar 1 ott 2019 alle ore 17:54 TIBIDI <notifications@github.com>
ha scritto:
… I have the same issue any progress ?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#149?email_source=notifications&email_token=AAFYNC5TUNO7S5ML5CZUQOLQMNXEZA5CNFSM4GM2QLSKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEABYWTA#issuecomment-537103180>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAFYNC4ZK6M2KHB4LKUMD4DQMNXEZANCNFSM4GM2QLSA>
.
--
Mario.
|
|
The guest of Darwin has not been supported so far. Unfortunately, we have not prioritized this feature in our development plans yet. If community contributors are willing to investigate this feature, it will be welcomed and we will provide corresponding supports. We will record your requirements in order to include it in the future development plans. Thank you. |
|
this is not a good behavior. i remember very well your enthusiasm the first
time that i ve asked some help by you. you had taken in serious
consideration to work on that as one of your priority. now you are coming
back to your steps. you dont seem to be mature guys.
Il lun 14 ott 2019, 08:41 Huang, Jun <notifications@github.com> ha scritto:
… The guest of Darwin has not been supported so far. Unfortunately, we have
not prioritized this feature in our development plans yet. If community
contributors are willing to investigate this feature, it will be welcomed
and we will provide corresponding supports. We will record your
requirements in order to include it in the future development plans. Thank
you.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#149?email_source=notifications&email_token=AAFYNC4V6S4RTCS7MYATZTTQOQIBFA5CNFSM4GM2QLSKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBDPFTA#issuecomment-541520588>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFYNC2QIBAKGFKUK4V5TRTQOQIBFANCNFSM4GM2QLSA>
.
|
|
Great thanks for your previous information from raising this issue to providing the detailed logs. All of these contributions helped us to narrow down the original issue to the OVMF booting support. We had already marked this issue as a feature in January of this year, and we are very open to discuss the implementation of this feature and welcome any pull request for it. Currently, we are busy in solving the issues reported by Android SDK customers and our focus is mainly targeted to Android Emulator, sorry for delayed response. Thanks for you understanding. |
|
Is OVMF_CODE.fd MacOS bios image? Can MacOs run with bioses shipped with qemu? |
|
With some patches in qemu and hax I'm able to run MacOS. It requires exactly Penryn CPU version in cpuid1.eax. I don't wan't to change the code to return exactly this version, instead qemu should set it through specific ioctls. But there are no ioctls for doing this yet. KVM API has an ioctl for setting values returned by cpuid. It would be good to make exactly this API in hax, but it does not allow to set only one register for cpuid. Any suggestions for solving this problem? Should hax move towards KVM-like ioctl API? |
|
I am working on a book developing for UEFI. I've been writing operating system drivers, and developing custom hardware for over 40 years. And I co-authored some very early PC applications and developer tools. I have had issues with UEFI in a virtual environment. First off, OVMF development isn't supported in VirtualBox, as the emulator is incomplete. They note that fact in the documentation. What does surprise me is the fact that HAXM doesn't support protected ROM areas when used with QEMU. The reason I need this is for screen captures for the book. My screen capture and screen recording software works on Windows, so I routinely run Linux and OVMF in qemu to capture these screens. This is just the OVMF_CODE.fd ROM image. The same happens for the OVMF_VARS.fd ROM image as well. If the right person reads this who is on the HAXM team, allocating a reserved area for ROM images is going to be a requirement anyway, since many hardware devices install ROM images during the POST phase of the BIOS. HaHoYou needs to stop making excuses for this and start finding the right people in the HAXM group at Intel to get this moving NOW. You can't finish developing UEFI without prioritizing this facet of development. If anyone at Intel who is involved with HAXM development reads reads this. Contact me directly through LinkedIn. Regards, |
|
The patch for supporting ROMD in qemu for hax is ready. Same for supporting MacOS in hax. These were ready at the time of my last post in this issue. They depend on some other patches in qemu and hax, which are not yet accepted. When these are accepted and merged, I will create a PR for solving this issue which includes ROMD support. |
|
Approximately how long do you anticipate approval will take? My book is a
few days from completion, so if this is going to take more than a few
weeks, I need to let the publisher know.
Thank you,
Oliver Bailey
…On Mon, Mar 16, 2020, 2:22 PM nevilad ***@***.***> wrote:
The patch for supporting ROMD in qemu for hax is ready. Same for
supporting MacOS in hax. These were ready at the time of my last post in
this issue. They depend on some other patches in qemu and hax, which are
not yet accepted. When these are accepted and merged, I will create a PR
for solving this issue which includes ROMD support.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBICDHM4WZPANTQIPCULRHZ37NANCNFSM4GM2QLSA>
.
|
|
HAXM has merged ROMD support. But needs another patch below in Qemu to be accepted. Let me ping Qemu community. |
|
Thank you.
Oliver
…On Tue, Mar 17, 2020, 2:15 AM hyuan3 ***@***.***> wrote:
HAXM has merged ROMD support. But needs another patch below in Qemu to be
accepted. Let me ping Qemu community.
https://lists.sr.ht/~philmd/qemu/patches/6470
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBIAT5KMTA5GRBN4G2BLRH4PPHANCNFSM4GM2QLSA>
.
|
At best two weeks, but I think it will take longer. How do yo want to test ROMD support? MacOS won't run with the uploaded qemu patch. |
|
Ok.
Send me the patch and I'll tessd's t on linux snd qemu for you. It will
probsbly be tomorrow morning before I test.
Oliver
…On Tue, Mar 17, 2020, 1:21 PM nevilad ***@***.***> wrote:
Approximately how long do you anticipate approval will take?
At best two weeks, but I think it will take longer. How do yo want to test
ROMD support? MacOS won't run with the uploaded qemu patch.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBIC33RQCVQ6DZHTNUV3RH65SJANCNFSM4GM2QLSA>
.
|
|
I've tried to run the patched version but it doesn't run anymore. I worked in the this branch on other features and it seems they affected the result. I don't have time now to debug it, so cant't help you now. |
|
Ok.
Thanks
…On Wed, Mar 18, 2020, 3:44 AM nevilad ***@***.***> wrote:
I've tried to run the patched version but it doesn't run anymore. I worked
in the this branch on other features and it seems they affected the result.
I don't have time now to debug it, so cant't help you now.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBIGSDDQARLERMNZM2G3RICCX3ANCNFSM4GM2QLSA>
.
|
|
Anyway,let's move our attention to the reasons why the vm that I have created with esxi 7 does not work great. I have a lot of problems to fix in my situation and I don't know how. Every time that the windows 10 VM restarts,the nvidia driver stops working because the error 43,the audio device does not work....check these errors : https://drive.google.com/drive/folders/1Rg9e0F3agJAnoRgEcussqAzxe6HjCTxx?usp=sharing Actually I'm using these parameters : hypervisor.cpuid.v0 = FALSE Anyway I've just found something that's really interesting. A reddit user gave me important informations about the graphic card that he is using and that's similar to mine :
and now check my 2080 ti parameters :
inside these different parameters there are the reasons why he does not see the error 43 and his audio works correctly. For example,why his bios is UEFI enabled,but mine isn't ? |
|
DDA script said that it does not support passthrough of all your devices, so windows is not able to dismount them. It may be not a bug of gigabyte, but missing functionality, that hyper-v requires. Some BIOS functions, that it requires, but other hypervisors not. If you are able to get the exact functions, which hyper-v/DDA script uses/checks, you can ask gigabyte if they support these. |
Too many differences to say what's exact missing. But it's a good idea to upgrade your GPU BIOS. Ask Nvidia for an update. |
So after installation but before reboot you are able to work with the GPU from VM? |
|
when I do the reinstallation of the drivers it works until the next reboot. I could use a script to make an automatic task,that tells to windows to reinstall automatically the nvidia driver as soon as it starts. could be good. but I'm finicky. In addition to this,the audio does not work. And I suspect that I could use an external monitor,when I will be able to fix the error 43. Read at the end of this article 👍 https://www.igorslab.de/community/threads/threadripper-x399-esxi-gpu-passthrough.1637/ it says : Cosmetics If you are sure everything works (i.e. you see the windows desktop on the monitor directly connected to the GPU and the core USB-Devices work as well) & still works after rebooting the VM, feel free to disable the vmware virtual gpu in the advanced settings, i.e. svga.present = FALSE. In case something goes wrong with your passthrough-GPU, you need to set the parameter to "TRUE" again. does not seem that an external monitor could be used ? |
|
This sounds interesting. I have installed an old version of the microsoft driver for the RTX 2080 ti and it seems to work better.
I have installed the nvidia control panel because I wanted to change the primary graphic card,from the vmware svga to the rtx,but it says : actually I'm not using a monitor attached to the nvidia GPU. |
|
no way. both the screens are tied to the vmware display adapter even if I have another monitor and graphic card. If I disable it from the beginning from the esxi options,the screen will be white.
|
The monitor connected to GPU? He writes that it should work. |
|
yeah,not for me.
Il lun 15 giu 2020, 15:36 nevilad <notifications@github.com> ha scritto:
… does not seem that an external monitor could be used ?
The monitor connected to GPU? He writes that it should work.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFYNC6H2ZBSYGZSVKNYSOLRWYPV3ANCNFSM4GM2QLSA>
.
|
|
do u remember the old connectivity problem with WSL 2 ? maybe the fix arrived...this could fix that : https://github.com/aaronlks/wsl2-19041-fixDNS/blob/master/fixdns.sh so,now could be nice to see if esxi get the proper ip address inside WSL2 :P |
|
This is about DNS. Under WSL you was unable to ping by IP and didn't used names. |
|
I've reinstalled windows 10. This time from an ISO image and it will create a virtual installation. Before I was using the physical installation of Windows 10 that I usually use everydays. it does not work....check the image below. https://drive.google.com/file/d/1GmoFu3Ro7YHOdh_s_cgi3dCBAXib6bnF/view?usp=sharing |
|
I've virtualized xen on top of qemu / kvm and It seems that the graphic card is seen by the client and can be used on a guest vm. isn't cool ?
|
|
@nevilad : are u alive ? |
|
Yep. |
|
what do you think about xen ?
Il gio 18 giu 2020, 09:37 nevilad <notifications@github.com> ha scritto:
… Yep.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFYNC3ABBAR35VGFVPEG5LRXG74RANCNFSM4GM2QLSA>
.
|
|
I've never worked with xen. |
|
its time to learn together ?
Il gio 18 giu 2020, 12:30 nevilad <notifications@github.com> ha scritto:
… I've never worked with xen.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFYNC65FV4OCJ2P7EGZAJTRXHUDHANCNFSM4GM2QLSA>
.
|
|
I'm not interested in xen, but can answer some questions. Hypervisors are similar. |
|
why you aren't interested ? it seems the perfect product for the virtualization. if it works,it should generate interest. Right now I have installed xen orchestra (similar to the vsphere client),using this tutorial : https://openschoolsolutions.org/how-to-install-xen-orchestra-with-all-features-and-update-it/ it says that I can install all the features of the professional versions,but I tried and I found that its not true. or it is not true anymore. Now I'm trying to understand how to create a virtual machine using the physical installation of windows 10. |
|
Let's come back to esxi. there is a new hope that it works like it should,thanks to the help of another reddit user,who says : It took me a while to get back to my system, but here's the information. You do need to have the VMWare video driver enabled, so do reinstall VMWare tools, but you can disable the VMWare display later. The VMWare driver does NOT interfere with the NVIDIA driver. I have both installed, and each handles its own monitor. However, be aware that the VMWare display will be blank if the NVIDIA driver is functioning (You won't be able to manage it through ESXi VM console, so you need a remote access software like parsec or remote utilities installed). I did NOT mess with the 64bit mmio settings. I did change the following hypervisor.cpuid.v0 FALSE pciPassthru0.msiEnabled FALSE pciHole.start 2048 pciHole.dynStart 3072 I also edited /etc/vmware/passthru.map and added NVIDIA10de ffff d3d0 false Those are the only changes I made. I DO NOT pass through the NVIDIA High Definition Audio device, JUST the video because I have seen issues with NVIDIA HDA and I don't need it since I use a cheap USB sound card. I have had this configuration for years and have never had issues with driver problems (barring some bad driver update that was quickly fixed). After I found the "magic sauce" I just left it alone and never had any more problems. I'm currently running one of my VMs as a gaming system with a 4k monitor as well as a secondary 1080p. For the sake of testing, I would remove all the customizations you already made, and try these options to see whether they will work for you. After the NVIDIA driver is working, you will see a small "monitor" in display settings that is the VMWare device. Just disable it. If there is some crash with the NVIDIA driver then it will re-activate the VMWare monitor and control, which you can then access through ESXi's VM console. Since I am also passing through a USB device with keyboard and mouse, and gaming performance is important to me, I also set the latency sensitivity to high. |
|
it didn't work. I asked more help from him and he asked me a lot of money,so I let it go. |
|
the gpu passthrough is coming into wsl 2 👍 https://ubuntu.com/blog/new-gpu-and-gui-features-announced-for-wsl-at-build |
|
Finally I found the way to connect mac os x to internet inside the WSL2.
This is the magic recipe :
…-netdev user,id=net0 -device
vmxnet3,netdev=net0,id=net0,mac=52:54:00:c9:18:27
Il giorno gio 18 giu 2020 alle ore 13:10 nevilad <notifications@github.com>
ha scritto:
I'm not interested in xen, but can answer some questions. Hypervisors are
similar.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#149 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAFYNC7IO7FAEAXI6YAPXD3RXHYZ5ANCNFSM4GM2QLSA>
.
--
Mario.
|
|
why u don't reply anymore ? there are big changes in the WSL platform. |
|
Startergo has experience in WSL and answers your questions. I'm watching. |
|
any news with OVMF support + haxm ? |
and others
Hello,
I tried to boot Mac Os X High Sierra with Qemu (3.1.0) and HAXM,but it did not work. Maybe someone can help me to fix it in some way ? Would be nice to make it work,since HAXM does not support nested virtualization...
C:\Programmi\qemu\qemu-system-x86_64 -m 3072 -cpu Penryn,vendor=GenuineIntel,+invtsc,vmware-cpuid-freq=on -machine pc-q35-2.9 -smp 4,cores=2 -usb -device usb-kbd -device usb-tablet -device isa-applesmc,osk="ourhardworkbythesewordsguardedpleasedontsteal(c)AppleComputerInc" -drive if=pflash,format=raw,readonly,file=C:\Programmi\qemu\OVMF_CODE.fd -drive if=pflash,format=raw,file=C:\Programmi\qemu\OVMF_VARS-1024x768.fd -smbios type=2 -device ich9-intel-hda -device hda-duplex -device ide-drive,bus=ide.2,drive=Clover -drive id=Clover,if=none,snapshot=on,format=qcow2,file=C:\Programmi\qemu\Clover.qcow2 -device ide-drive,bus=ide.1,drive=LinuxHDD -drive id=LinuxHDD,if=none,file=I:\OS\Images\mac_hdd.img,format=qcow2 -device ide-drive,bus=ide.0,drive=LinuxDVD -drive id=LinuxDVD,if=none,snapshot=on,media=cdrom,file=I:\OS\Images\HighSierra.iso -accel hax
HAX is working and emulator runs in fast virt mode.
C:\Programmi\qemu\qemu-system-x86_64: warning: Ignoring ROMD region 0x00000000ffc84000->0x0000000100000000
C:\Programmi\qemu\qemu-system-x86_64: warning: Ignoring ROMD region 0x00000000ffc00000->0x00000000ffc84000
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
VCPU shutdown request
The text was updated successfully, but these errors were encountered: