I have had a lot of kernel panics when booting OSX guests if I configure 3 or more cpus. It never panics with 1 or 2 cpus, rarely happens with 3 cpus, and almost always happens with 4 or more cpus. Stragely it does eventually boot after many failed attempts.
I don't see the problem on an ESXi server so I'm guessing it may point to a hardware problem on my laptop (Late 2016 15" MBP w/ touchbar), but the machine passes all Apple hardware diagnostics. Is it possible this a known problem with Fusion? Is it possible to set cpu affinity to prevent Fusion from using certain cpu cores? I believe I have a directory full of boot failure logs but I forget where they are written. Should I attach a bunch of them?
Thanks in advance!
Edit: The same behavior occurs on Fusion 10.1.0 as well as 10.1.1. It appears that several other people experience the same issue, so it may indeed be a bug in Fusion.
Of course, you are right. But running into kernel panic with one single VM configured with 4 CPUs, 8 GB RAM on a i7 7920HQ with 16 GB RAM shows there is something bad going on.
I agree, it should just perform more slowly in the case of over provisioning/resourcing, not panic.
Yeah, very strange bug. I wonder if Apple's hard coded something in there that expects specific physical hardware.
I wish they'd spend at least a bit of the repatriated funds on some enhanced QA!
We've filed a radar with Apple, but I can't comment on their progress. It's a high priority, we don't like crashes at the kernel level... because ESX
Having numerous kernel panics while restarting...
Running latest sierra as guest on latest sierra running latest vmware fusion
With 4 processor cores in iMac Pro 8 core/ 32GB machine. Seems more stable with 2 cores allocated
Little ping. While you might not be allowed to share the whole status, would it be possible for you to advise if I can safely buy new (more powerful) hardware, with the expectation that a fix / workaround (other than limiting to 2 cores, of course) will be available sooner rather than later ? Let's say on a two to three months timeframe.
Fusion 10.1.1 (7520154) - SAME PROBLEM. Running 10.12.6 Sierra GUEST on a 10.12.6 Sierra HOST. Got it with 4 processors. Turned it down to 2 processors and it booted, but this is really problematic.
Definitely a VMware problem. PANIC BELOW:
*** Panic Report ***
panic(cpu 3 caller 0xffffff801b5362f8): "a freed zone element has been modified in zone kalloc.48: expected 0xc0ffeed4a7f26ce2 but found 0, bits changed 0xc0ffeed4a7f26ce2, at offset 0 of 48 in element 0xffffff8023cbb990, cookies 0x3f0011548439da72 0x53521720a438595"@/Library/Caches/com.apple.xbs/Sources/xnu/xnu-3789.70.16/osfmk/kern/zalloc.c:650
Backtrace (CPU 3), Frame : Return Address
0xffffff887bd33500 : 0xffffff801b4e953c
0xffffff887bd33580 : 0xffffff801b5362f8
0xffffff887bd335f0 : 0xffffff801b535549
0xffffff887bd33640 : 0xffffff801b533f84
0xffffff887bd33790 : 0xffffff801ba35de7
0xffffff887bd337c0 : 0xffffff801ba527ba
0xffffff887bd337f0 : 0xffffff801ba88fba
0xffffff887bd33820 : 0xffffff7f9c157661
0xffffff887bd33860 : 0xffffff7f9c156e99
0xffffff887bd339f0 : 0xffffff7f9c168d8c
0xffffff887bd33ac0 : 0xffffff801babd76a
0xffffff887bd33b30 : 0xffffff7f9c167b46
0xffffff887bd33b60 : 0xffffff7f9cbf4e9c
0xffffff887bd33c50 : 0xffffff7f9cbf07e8
0xffffff887bd33c80 : 0xffffff7f9cbe4542
0xffffff887bd33ca0 : 0xffffff801babd76a
0xffffff887bd33d10 : 0xffffff7f9cbe44f8
0xffffff887bd33d20 : 0xffffff7f9caf173a
0xffffff887bd33e90 : 0xffffff7f9caf10e8
0xffffff887bd33f30 : 0xffffff7f9cb2121f
0xffffff887bd33f50 : 0xffffff7f9cb211ce
0xffffff887bd33f70 : 0xffffff7f9cb21244
0xffffff887bd33f90 : 0xffffff7f9cb21346
0xffffff887bd33fb0 : 0xffffff801b4978f7
Kernel Extensions in backtrace:
BSD process name corresponding to current thread: kernel_task
Mac OS version:
Darwin Kernel Version 16.7.0: Thu Jun 15 17:36:27 PDT 2017; root:xnu-3789.70.16~2/RELEASE_X86_64
Kernel UUID: D3314D98-5D40-3CD8-98A4-F1DD46C20E03
Kernel slide: 0x000000001b200000
Kernel text base: 0xffffff801b400000
__HIB text base: 0xffffff801b300000
System model name: VMware7,1 (Mac-27ADBB7B4CEE8E61)
System uptime in nanoseconds: 4012925077
last loaded kext at 4004288890: com.apple.driver.pmtelemetry 1 (addr 0xffffff7f9c104000, size 90112)
Yah, it's one of those 'do we try to fix it (and then possibly have to un-fix it later) or wait for them to fix it?'
For now, if this is the same underlying cause as the RADAR we've filed, the issue can be mitigated by enabling 'code profiling' in the VM.
The failure should no longer occur.
This is not the same issue, my bad!
The issue corrected with the fix I describe is for 10.14 beta crashing when doing a sysctl -a. (same boat, Apple changed a thing, it breaks us)
Sure, but at some point, i'm not the one in touch with Apple, so it's my only way to: get status update, get potential workaround informations. I'm being blocked on that for going forward on some work, for weeks now.
@Mikero, since we are approaching 6 months without a fix from Apple are we nearing that threshold where VMWare chooses to go down the path of 'do we try to fix it (and then possibly have to un-fix it later)'? It does appear that this is a hardware related issue since some machines don't show the problem. Hopefully it is something that could be mitigated from your side? Oddly enough, once one of these VM's manages to start it seems rock solid so I imagine the kludge would only need to be in affect during initial start up.
Without going into much detail, we think this is fixed in 10.14.
We're keeping an eye on the situation, but we haven't been seeing this with our 10.14 testing. (tho it's early and we have plenty more testing to do)
We would be very interested in hearing if folks still have the same behavior with the new beta.
Mikero, I have some interesting news to report. I just upgraded to Fusion 10.1.3 and installed a fresh install of MacOS 10.13 (High Sierra) + the current updates. I have been able to consistently boot this VM with no kernel panics with 4 CPU's configured. I haven't tried other MacOS versions, but this is very encouraging. This is the same host computer that I was using when I first started this thread, so the hardware hasn't changed.
I've just upgraded to VMware Fuison 10.1.3 but kept the Sierra and not High Sierra host, set the number of VCPUs to 4, and it is still panicking. Either you are more lucky than I am, or there's some undocumented hotfix in High Sierra with all the updates and/or Fusion 10.1.3 is required as well. But just the VMware upgrade is not enough :/.
For what it's worth, my host is running macOS Sierra (10.12.6), not High Sierra. It is the VM that I set up that is High Sierra. My guess is that Apple corrected the issue in High Sierra so that it doesn't panic any more when it boots. We likely have to wait for a future version of Fusion to fix the issue for other guest OS's?