[Fixed by supplemental update] kernel panic from memory leak on 10.15.6

Discussions related to using VirtualBox on Mac OS X hosts.
fth0
Volunteer
Posts: 5677
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: kernel panic from memory leak on 10.15.6

Post by fth0 »

In the output of sudo zprint -d, you can watch the column named cur size as a simple measure for the amount of kernel memory lost, and you usually don't want it to be in the GB range. ;)
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

fth0 wrote:execute sudo zprint -d. If the output contains kalloc.32 every few seconds, then your scenario is affected.

Up to now, it is known that Oracle VirtualBox and VMware Fusion are affected, and it is assumed that Parallels Desktop and the macOS Hypervisor.framework are not affected.
this is not a safe assumption. i just ran the zprint, watched it sit quiet awhile, then intiated a time machine backup: kalloc.32, kalloc.32, kalloc.32.
fth0
Volunteer
Posts: 5677
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: kernel panic from memory leak on 10.15.6

Post by fth0 »

You are correct, it is not a safe assumption.

Regarding your Time Machine backup, out of curiosity: If you compare the cur size of the kalloc.32 memory zone before and after a Time Machine backup, what's the difference? When measuring, note that you have to wait for Time Machine doing its cleanup at the end, it may free up some kalloc.32 memory late.
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

fth0 wrote:If you compare the cur size of the kalloc.32 memory zone before and after a Time Machine backup, what's the difference? When measuring, note that you have to wait for Time Machine doing its cleanup at the end, it may free up some kalloc.32 memory late.
so the last backup was less than half hour ago. in this session, 386mb was copied.

kalloc.32 cur size
318312K before time machine
320916K after time machine cleanup

no other kalloc.32 activity visible before or after
fth0
Volunteer
Posts: 5677
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: kernel panic from memory leak on 10.15.6

Post by fth0 »

xbuntugest wrote:kalloc.32 cur size
318312K before time machine
320916K after time machine cleanup
Simplifying the calculation for convenience reasons, this Time Machine backup could be responsible for a memory leak of 2.5 MB in the kalloc.32 kernel memory zone. If this were a representative run, and Time Machine would create a backup once per hour, then you'd lose 1 GB in 400 hours or 16.6 days.

If you want a better estimate regarding your everyday computer use, you could use sudo zprint -t -s twice, 24 hours apart, and compare the kernel memory zones with the largest cur size values.
Last edited by fth0 on 16. Nov 2021, 19:00, edited 1 time in total.
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

fth0 wrote:If you want a better estimate regarding your everyday computer use, you could use sudo zprint -t -s twice, 24 hours apart, and compare the kernel memory zones with the largest cur size values.
i don't want anything. i thought you were asking for evidence that time machine (even running under safe boot, i now know) triggers the bug, to support my claim that other software cannot be assumed leak-safe.

i wrote about time machine in general because this morning i saw my first time machine backup in a day coincide with more than half a gb of wired memory increase. how much of that was leakage triggered by tm? i don't know. had i ever seen that fast or large a lasting increase when i wasn't running a vm? no, i had not.
fth0
Volunteer
Posts: 5677
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: kernel panic from memory leak on 10.15.6

Post by fth0 »

xbuntugest wrote:i thought you were asking for evidence
A slight misunderstanding, maybe because English is not my mother tongue. Let me try and explain:

I wrote that I'm curious, and I were asking for information, not meant as evidence, but just to collect more information about the nature of the memory leak, especially its size in respect to Time Machine.

I also didn't mean to indicate that you want something. I only offered an advice to you (and to all readers), how to proceed, if you want to investigate further. Ok, from reading the posts of kigoi, I could have known that you don't need any advice from me. ;)
airdrummingfool
Posts: 1
Joined: 27. Jul 2020, 18:56

Re: kernel panic from memory leak on 10.15.6

Post by airdrummingfool »

Apple has released an update to fix the issue with virtualization on 10.15.6.

Release notes:
macOS Catalina 10.15.6 supplemental update includes bug fixes for your Mac.

• Fixes a stability issue that could occur when running virtualization apps
• Resolves an issue where an iMac (Retina 5K, 27-inch, 2020) may appear washed out after waking from sleep
it's available in Software Update. I'm downloading it now and hope to test it shortly.
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

airdrummingfool wrote:Apple has released an update to fix the issue with virtualization on 10.15.6.

Release notes:
macOS Catalina 10.15.6 supplemental update includes bug fixes for your Mac.

• Fixes a stability issue that could occur when running virtualization apps
• Resolves an issue where an iMac (Retina 5K, 27-inch, 2020) may appear washed out after waking from sleep
it's available in Software Update. I'm downloading it now and hope to test it shortly.
seems to be working right. i watched various things, including zprint.
  • big post-update time machine didn't trigger problem
  • virtualbox didn't trigger problem
  • wired memory returned to original level after those jobs quit
what a frikkin relief this is!!!

ps. still, before i reinstate my full setup, i'm gonna watch memory pressure a day or so, until convinced.
trifster
Posts: 19
Joined: 16. Oct 2019, 01:54

Re: kernel panic from memory leak on 10.15.6

Post by trifster »

xbuntugest wrote:
airdrummingfool wrote:Apple has released an update to fix the issue with virtualization on 10.15.6.

Release notes:
macOS Catalina 10.15.6 supplemental update includes bug fixes for your Mac.

• Fixes a stability issue that could occur when running virtualization apps
• Resolves an issue where an iMac (Retina 5K, 27-inch, 2020) may appear washed out after waking from sleep
it's available in Software Update. I'm downloading it now and hope to test it shortly.
seems to be working right. i watched various things, including zprint.
  • big post-update time machine didn't trigger problem
  • virtualbox didn't trigger problem
  • wired memory returned to original level after those jobs quit
what a frikkin relief this is!!!

ps. still, before i reinstate my full setup, i'm gonna watch memory pressure a day or so, until convinced.
Woohoo!! this took way too long to be released.
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

trifster wrote:Woohoo!! this took way too long to be released.
i hope they tell the story. the download is large, and the update takes some time. were they hunting the bug the whole time? was it a symptom of deep problems? the public deserves details, and free popcorn.
xbuntugest
Posts: 17
Joined: 26. Jul 2020, 17:18

Re: kernel panic from memory leak on 10.15.6

Post by xbuntugest »

xbuntugest wrote:before i reinstate my full setup, i'm gonna watch memory pressure a day or so, until convinced
first yeah i'm convinced, and i haven't seen any new bugs either, but i am actually having a hard time relaxing my guard, maybe that's from the pressure of the outside world being scary.

second, i learned things from this! and one is a big change in how i use virtualbox: i was giving guest linux way too much memory! and with that adjusted down, the host system is running better than it was a month ago! very nice.
trifster
Posts: 19
Joined: 16. Oct 2019, 01:54

Re: kernel panic from memory leak on 10.15.6

Post by trifster »

xbuntugest wrote:
trifster wrote:Woohoo!! this took way too long to be released.
i hope they tell the story. the download is large, and the update takes some time. were they hunting the bug the whole time? was it a symptom of deep problems? the public deserves details, and free popcorn.
I don't think the bug fix itself was time consuming to resolve in the code... but I believe there isn't an adequate delta-update mechanism so that Catalina required a full build to be released with full installer. Anytime such an update is required for every variant of Mac the update needs to go through the full internal test/QA process before it can be released.
fth0
Volunteer
Posts: 5677
Joined: 14. Feb 2019, 03:06
Primary OS: Mac OS X other
VBox Version: PUEL
Guest OSses: Linux, Windows 10, ...
Location: Germany

Re: kernel panic from memory leak on 10.15.6

Post by fth0 »

Small summary of the current status:

Apple has released a macOS Catalina 10.15.6 Supplemental Update (19G2021) on 2020-08-12, addressing the memory leak. A VMware Fusion lead developer (Darius, involved in filing the bug with Apple) has confirmed the bugfix. My own tests also do not show the memory leak any more.

@iains, @yyc: Can you also confirm the memory leak as solved? TIA
Last edited by fth0 on 16. Aug 2020, 10:13, edited 1 time in total.
iains
Posts: 9
Joined: 23. Jul 2020, 22:31

Re: kernel panic from memory leak on 10.15.6

Post by iains »

fth0 wrote:Small summary of the current status:

Apple has released a macOS Catalina 10.15.6 Supplemental Update (19G2021) on 2020-08-12, addressing the memory leak. A VMware Fusion lead developer (Darius, involved in filing the bug with Apple) has confirmed the bugfix. My own tests also do not show the memory leak any more.

@iains, @yyc: Can you also confirm the memory leak as solved?
In my queue to check - machines are always busy ;)
Post Reply