10.13 Failing installation; Kernel driver not loaded

Discussions related to using VirtualBox on Mac OS X hosts.
michaln
Oracle Corporation
Posts: 2973
Joined: 19. Dec 2007, 15:45
Primary OS: MS Windows 7
VBox Version: PUEL
Guest OSses: Any and all
Contact:

Re: 10.13 Failing installation; Kernel driver not loaded

Post by michaln »

"NVRAM" is an abstract concept here, especially if it's implemented on top of EFI (I believe it is). It could be stored in some EEPROM or flash, but it could be a file on disk. It's up to the firmware, which again makes it plausible that there would be model-specific misbehavior.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

Yes, these were absolutely clean installs where I booted into the recovery partition and deleted the HD with Disk Utility and reinstalled the OS. I did this for 10.13 and 10.13.1 and 10.13.2 and always got the same error. After the clean install, I did nothing other than try to install the latest version of VirtualBox, not even connect to an iCloud account. I have plenty of boxes to experiment with. We have over 1000 Macs in our district.

I knew about the NVRAM problem and tried clearing it with cmd-opt-p-r as mentioned in another post, but still no joy. Also tried sudo spctl --master-disable as mentioned in the same post.

Here are my results so far and I will post more that I am working on:

Model Identifier: MacPro5,1
OSX version: 10.13, 10.13.1, 10.13.2
Installation: Failed on clean install for each OS on 5 identical computers

Model Identifier: MacPro6,1
OSX version: 10.13, 10.13.1, 10.13.2
Installation: Failed on clean install for each OS

Model Identifier: MacBookPro11,3
OSX version: 10.13.1, 10.13.2
Installation: Success

Model Identifier: MacBookPro13,3
OSX version: 10.13.2
Installation: Success

If the problem is solely 10.13, then why is it working on some 10.13.1 and 10.13.2 machines, but not on others?

Bob Cassidy
IT Director
Somerton School District
Last edited by socratis on 3. Jan 2018, 21:39, edited 2 times in total.
Reason: Removed unnecessary verbatim quote of the whole previous message.
socratis
Site Moderator
Posts: 27330
Joined: 22. Oct 2010, 11:03
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Win(*>98), Linux*, OSX>10.5
Location: Greece

Re: 10.13 Failing installation; Kernel driver not loaded

Post by socratis »

@bcassid
Thank you for trying out so many systems!!! I've put a call to every other thread that deals with the 10.13 installer issue. Let's wait a couple of days. Most probably I will be editing the messages in this thread (it's the "sticky" one ;) ) to sum all the information.

So far it seems that all "MacBookPro11,#" don't exhibit the problem, while the "MacPro#,#" do. Let's see what other users are experiencing. Then you'll have a more solid case to go to Apple with.

Thanks for doing that! It's been a real headache for more than you can imagine... :?
Do NOT send me Personal Messages (PMs) for troubleshooting, they are simply deleted.
Do NOT reply with the "QUOTE" button, please use the "POST REPLY", at the bottom of the form.
If you obfuscate any information requested, I will obfuscate my response. These are virtual UUIDs, not real ones.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

More results:

Model Identifier: MacBookAir7,2
OSX version: 10.13.2
Installation: Success

Model Identifier: MacBookPro8,2
OSX version: 10.13.2
Installation: Success

I'm trying to locate models other than Mac Pros and MacBooks to try this on and will post results when I'm able. Maybe it's some interaction between High Sierra and the firmware? It must have something to do with permissions because the only difference I'm seeing between the successful and failed installations is that in the Security & Privacy pane, the Allow button does nothing for the "System software from developer "Oracle America, Inc." was blocked from loading." message for the failed installations. For the successful installations, the message goes away when you click the Allow button. That is the ONLY difference I'm seeing. If anyone knows of a workaround for allowing those permissions ...

Bob Cassidy
IT Director
Somerton School District
Last edited by socratis on 3. Jan 2018, 21:39, edited 2 times in total.
Reason: Removed unnecessary verbatim quote of the whole previous message.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

Here's another result:

Model Identifier: iMac12,1
OSX version: 10.13.2
Installation: Success

It's looking to me like the problem is related to 10.13.x running on a Mac Pro, so the question is, has anyone successfully installed VirtualBox on any Mac Pro running High Sierra?

Bob Cassidy
IT Director
Somerton School District
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

and another:

Model Identifier: MacBookPro6,2
OSX version: 10.13.2
Installation: Success

Bob Cassidy
IT Director
Somerton School District
socratis
Site Moderator
Posts: 27330
Joined: 22. Oct 2010, 11:03
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Win(*>98), Linux*, OSX>10.5
Location: Greece

Re: 10.13 Failing installation; Kernel driver not loaded

Post by socratis »

bcassid wrote:It's looking to me like the problem is related to 10.13.x running on a Mac Pro, so the question is, has anyone successfully installed VirtualBox on any Mac Pro running High Sierra?
I'll try to mobilize the users and see if anyone has a MacPro, although they're kind of rare...
Excellent job there Bob!!!
Do NOT send me Personal Messages (PMs) for troubleshooting, they are simply deleted.
Do NOT reply with the "QUOTE" button, please use the "POST REPLY", at the bottom of the form.
If you obfuscate any information requested, I will obfuscate my response. These are virtual UUIDs, not real ones.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

Thanks to Shawn on this thread:
https://forums.developer.apple.com/mess ... 031#261031
I was able to get a successfully installation.

It turns out that remoting in causes the Allow button to have no effect when clicking it. All of the Mac Pro models we use are servers that we remote into with Apple Remote Desktop.

Installing without remoting in worked! So the problem is not related to Mac Pros, but rather to remote access.

Thanks to everyone who helped to get this workaround.
Bob Cassidy
IT Director
Somerton School District
michaln
Oracle Corporation
Posts: 2973
Joined: 19. Dec 2007, 15:45
Primary OS: MS Windows 7
VBox Version: PUEL
Guest OSses: Any and all
Contact:

Re: 10.13 Failing installation; Kernel driver not loaded

Post by michaln »

Now that is very interesting. I'm guessing the behavior is intentional in that there must be a user physically present to allow the kext loading (in other words, malicious software cannot simulate the approval). Obviously there was a complete user experience fail in that the remote user has absolutely no indication that what they're doing has no effect.

Sigh, modern security. Computers are well on their way to being so secure that they can't be used for anything.
KTK1990
Posts: 3
Joined: 22. Feb 2017, 04:17

Re: 10.13 Failing installation; Kernel driver not loaded

Post by KTK1990 »

I came here expecting to post about my iMac having the same issue and my MacBook Pro not experiencing this issue, but it looks like the physically present theory/hypothesis is correct. I attempted to reinstall virtual box multiple times over the past two weeks including booting in safe mode and each time resulted in failure. Each time I was remoting in via screen sharing on my MacBook. Once I physically went over there and plugged in a mouse to click on allow, I was able to install virtual box and download and run the installer again to update it to the latest version.

Virtualbox now works perfectly fine on my iMac, along with my MacBook.
Last edited by socratis on 10. Dec 2017, 05:06, edited 1 time in total.
Reason: Removed unnecessary verbatim quote of the whole previous message.
michaln
Oracle Corporation
Posts: 2973
Joined: 19. Dec 2007, 15:45
Primary OS: MS Windows 7
VBox Version: PUEL
Guest OSses: Any and all
Contact:

Re: 10.13 Failing installation; Kernel driver not loaded

Post by michaln »

Thanks for the confirmation. I suspect that a number of users ran into this problem but didn't bother mentioning the bit about remote access, because that does not normally cause problems, so they didn't think it relevant. Definitely good to know. A bit depressing that Apple did not document this behavior and didn't make the user interface flag it.
socratis
Site Moderator
Posts: 27330
Joined: 22. Oct 2010, 11:03
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Win(*>98), Linux*, OSX>10.5
Location: Greece

Re: 10.13 Failing installation; Kernel driver not loaded

Post by socratis »

michaln wrote:I suspect that a number of users ran into this problem but didn't bother mentioning the bit about remote access, because that does not normally cause problems
I am really puzzled and deeply frustrated by this behavior. How can someone not mention that they're RDPing into the freaking thing??? We've had from keyboard, to display, to sound, to... (name it) issues, and only after a lot of back and forth the OP decided to mention "This actually doesn't happen if I'm physically in front of the computer". Really Obi-Wan??? How odd!!!

Are you thinking that RDPing in a computer and actually, literally, physically sitting in front of it is the one and the same??? That the RDP-server and RDP-client are bug-free programs??? Of course they're not!!! Why is there the need for an added layer of complexity? So we can all together as a community get more grey hair faster? Or lose the ones we have because we're collectively scratching our heads, trying to figure out that it works in case 'A', but not in case 'A'?

The least you could do, as a responsible reporter is to try the whole thing without another layer of software on top of an already complicated piece of software. You know, eliminate potential culprits, get rid of some variables to help solve the problem, get it to fail in the simplest possible, reproducible scenario (like 'bcassid' and 'KTK1990' did).

I needed to get that off my chest. Sorry for if it was too much of a rant...

<NoteToSelf>
From now on, ALWAYS ask as the first question, if any reported problem occurs on a stand-alone computer, or an RDPing one.
</NoteToSelf>
Do NOT send me Personal Messages (PMs) for troubleshooting, they are simply deleted.
Do NOT reply with the "QUOTE" button, please use the "POST REPLY", at the bottom of the form.
If you obfuscate any information requested, I will obfuscate my response. These are virtual UUIDs, not real ones.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

I share your frustration, but it's easy to overlook some things. That's why it's so useful for us to share, right?

I'm in charge of about 30 servers, mostly Mac Pros that are in different locations in our school district, some of which I haven't been in the same room as for years and all of which are maintained remotely using Apple Remote Desktop (ARD), an Apple product, which is not the same as RDP. I've been using ARD since version 1 and never once had a problem related to it, therefore, never considered it as being part of the problem. I was focusing on the machine model variable.

When Shawn mentioned the other thread about the remoting problem, I actually was able to solve it remotely by using the Avocent KVM that all of our servers are connected to. It has a web interface that lets you remote in and it is actually the same as sitting there physically.

Bob Cassidy
IT Director
Somerton School District
Last edited by socratis on 3. Jan 2018, 21:43, edited 2 times in total.
Reason: Removed unnecessary verbatim quote of the whole previous message.
socratis
Site Moderator
Posts: 27330
Joined: 22. Oct 2010, 11:03
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Win(*>98), Linux*, OSX>10.5
Location: Greece

Re: 10.13 Failing installation; Kernel driver not loaded

Post by socratis »

bcassid wrote:Apple Remote Desktop (ARD), an Apple product, which is not the same as RDP.
Potato, potahto. Tomato tomahto... ;)
bcassid wrote:I've been using ARD since version 1 and never once had a problem related to it, therefore, never considered it as being part of the problem.
There's always a first... Have you verified this behavior with other software that installs kexts and therefore needs that approval from the Security Settings? Does it also fail if tried while RDPing in the computer in question?
bcassid wrote:I was focusing on the machine model variable.
That's your classical, text-book, typical "red herring", right there ;)
Do NOT send me Personal Messages (PMs) for troubleshooting, they are simply deleted.
Do NOT reply with the "QUOTE" button, please use the "POST REPLY", at the bottom of the form.
If you obfuscate any information requested, I will obfuscate my response. These are virtual UUIDs, not real ones.
bcassid
Posts: 17
Joined: 29. Sep 2017, 02:35

Re: 10.13 Failing installation; Kernel driver not loaded

Post by bcassid »

bcassid wrote:Apple Remote Desktop (ARD), an Apple product, which is not the same as RDP.
socratis wrote:Potato, potahto. Tomato tomahto... ;)
The ARD "client" is built into macOS. It's one checkbox in System Preferences.
bcassid wrote:I've been using ARD since version 1 and never once had a problem related to it, therefore, never considered it as being part of the problem.
socratis wrote:There's always a first... Have you verified this behavior with other software that installs kexts and therefore needs that approval from the Security Settings? Does it also fail if tried while RDPing in the computer in question?
No I haven't verified this behavior with other software, but remember, this problem didn't start until High Sierra which was released less than three months ago and I had no hint whatsoever that it was related to ARD, until I saw Shawn's email.
bcassid wrote:I was focusing on the machine model variable.
socratis wrote:That's your classical, text-book, typical "red herring", right there ;)
That's for sure! Until I got that email from Shawn, I was chasing a wild goose!

At any rate, I'm happy to have a solution and thanks to all for helping.

Bob Cassidy
IT Director
Somerton School District
Post Reply