SOCD report detected: (AP watchdog expired)

In my case the fix was to stop using Chrome as a browser and use Safari instead.

For me, the issue was Chrome browser. I stopped using it and now my Mac is fine (using Safari instead).

Also, remove any internet-based apps.

Hope that works.

Further update:

Just looked in my 'IT box', and found a USB A to USB C adapter. I have since disconnected my Time Machine backup drive that was connected to USB A, and used the adapter to connect it.

Let's see if this error, "SOCD report detected: (AP watchdog expired)" happens again.

PS: Maybe it is just me, but, already feel as if my mouse is moving MUCH better than it was...

I just bought a new Mac Mini M1 16G 256 storage two days ago, Big Sur 11.2.3 installed and the same thing happened twice: green screen and reboot. Error log is [SOCD report detected: (AP watchdog expired). I was also using a WD external hard drive (circa 2012-13) to manually transfer documents (only one or two apps, both updated), and I believe the drive was attached during both incidents. Have detached and will update in a week or two.

I've had my Mac Mini M1 about a month and just got this crash/reboot error. Was watching Twitch at the time.

I did recently plug in a USB Blu-Ray Drive so I've unplugged that and will keep an eye on it. I'll unplug my time machine drive if it happens again.

anyway i disconnected my Toshiba external disk that was connected to a portable USB connector and waiting to see if it was this the issue but it seems odd that I could not use an external disk.....on a a brand new Mac with such small space. I hope Apple will release some updated ASAP for this issue!!!!! I also have the issue the some software run quite slow such Rhinoceros and Photoshop and I didn't expect this since the marketing was concentred on "HOW FAST IS THE MAC MINI". ...yep, maybe not mine (mAc Mini 16GB, 1 T space).

"Its like a component defect in the logic board"

i'm not entirely sure there's enough evidence to support this theory. you could be right though.

Just because 'Watchdog' has relation to logic board, doesn't necessarily point to 'logic board replacement'. correlation doesn't equal causation.

my theory: there's some x86 legacy driver that's being translated through some sort of abstraction layer like Rosetta, and there's some sort of leak or overflow that builds up over time, and eventually the system has to restart.

nice find.

to further speculate, based in this comment https://gist.github.com/Omar-Ikram/8e6721d8e83a3da69b31d4c2612a68ba#file-endpointsecuritydemo-m-L92

// This will cause discrepancies when converting Mach absolute time values from Endpoint Security Messages.

// The best option would be to compile your client as a universal binary:

// https://developer.apple.com/documentation/xcode/building_a_universal_macos_binary
// Note: Running x86_64 code running under Rosetta 2 will have timebase information for Intel CPUs.

not saying this is exactly the cause. but we do have a release note saying that this API was causing panics due a timeout type issue. it's feasible, not confirmed though, that these could be caused by something like x86_64 code spitting out bad values through Rosetta 2. wild speculation and could be completely wrong.

The same "SOCD report detected: (AP watchdog expired)" panic has occurred numerous times on my machine, but usually only once every few months, tops. I've had such a panic a total of maybe 3 or 4 times since purchasing the M1 mini in March. I am aware it is related to external storage.

I have an external Thunderbolt enclosure, a Sabrent EC-T3NS, which is based on an Intel Thunderbolt controller chip. I am using it with a 1TB Samsung 980 Pro.

I have sent a feedback post about this, which contains the panic report and a sysdiagnose captured after rebooting back into Monterey. FB9286550 if anyone wants to know, for reference. And if anyone wants to mention this Feedback in their own reports of this issue.

The Panic dump appears to contain a base64 encoded SOCDContainer object, which may be related to the SoC's microcontroller which locked up. Maybe someone will actually find this useful. Hopefully this isn't some errata that can't be corrected with at most a firmware update.

[Edited by Moderator] 

The string "SOCD" is only contained in kernel extensions related to SMC(System Management Controller) and PMU (Power Management Unit):

AppleD2422PMU.kext

AppleD2449PMU.kext

AppleSMC.kext

ApplePMGR.kext

AppleSPMIPMU.kext

AppleT8103PMGR.kext

Maybe the SOCD combines on M1 all the functions that on an Intel Mac the SMC and PMU did?

Since we also saw this base64 data in the panic log it could also have some NVRAM on it.

So maybe the SOCD is really daemon program doing former SMC and PMU tasks and runs on the AP (application processor). And everything is encrypted by the keys in the security enclave.

And on the AP is also a watchdog function that detects if the SOCD daemon is frozen. And since it handles all those important SMC and PMU stuff all they can do is a hard reset of the Mac.

Still a lot of speculation, but since nothing is documented...

pmu restarting
"SMCFW failed to handle restart request\n"SOCD: BASE 0x%04X, COUNT 0x%04X
"AppleDialogPMU::%s failed to alloc %d bytes"_handleSOCDSOCD: failed to read magic value
SOCD: Magic value %02Xh %02Xh %02Xh %02Xh
SOCDSOCD: failed to read trigger counters
SOCD: Failed to reset magic
SOCD: Failed to read socd container
SOCD: Failed to allocate OSData
SOCD: PMU %llu ns, TOTAL %llu ns (%u, %u)

some interesting strings in AppleSPMIPMU.kext

another possibility / theory on hardware side: bad or corrupted NAND (nonvolatile) flash storage (used by AES Engine) within SEP..... daemon times out and triggers pmu to restart

so many possibilities!

also, i'm not sure what line 109 refers to, but its kinda funny.

PMURTCHelperreadRTCUpcountTicks%s::%**** <rdar://problem/31800609> (retries=%d)

something to do with the pmu reading real time clock tickets... and looks like reference to internal help system

more errata/interesting information:

bug reported on 11.5.2 that claims that the "amount of user and system CPU time a given process has consumed over its lifetime" for processes run through Rosetta 2 on M1 is an incorrect value.

https://openradar.appspot.com/FB9546856

my guess would be SEP would potentially monitor some of these values. If there is skew over time, that may result in SOCD getting confused, watchdog times out on sepOS, and triggers pmu to hard reset.

Merlinsmate, I'm curious - your 10+ yrs on your prior Mac brought this to mind - when you made the migration to the M1 mini, did you use Migration Assistant? I presume you would, I always have.

If so, what kernel extensions came along for the ride? Anything particularly ancient?

Call it a hunch I'm looking for evidence to support or knock down.

I believe it is, in my case anyway. I ordered a USB C to DP cable, will see how that goes. I though about upgrading the monitor as well. Mine is the LG 31MU97-B, Has been great over the years but now show signs of screen burn in when left up for some time. Should do HDMI @ 30Hz but Preferences for display show 24Hz, not sure if that has anything to do with it. I have an Ultra High Speed HDMI cable which is supposed to support 8k, so I figure it should be fine. Will see what it works at on that USB C to Mini DisplayPort does.