Latest ElCap upgrade hanging 10.11.6 - "Stalling for detach"

All good thoughts here, but I have the "Incompatible Kernel Extension Configuration Data" from Jan 5 and there's even a "Core Suggestions Configuration Data" from today, and the issue persists. As I've mentioned, the only fix for me, is to disable the "Use Graphics Processor" in the Performance Settings of Photoshop.

My logs show that my uptime was about 6 days 2 1/4 hours when the detach messages started reappearing for me today. That is less than the 2^29 milliseconds that Robert Lipe mentions -- but it is about 40 minutes into 2^19 seconds.

I would tend to dismiss this as coincidence -- except that at the time it happened, I was sound asleep. Which indicates that user interaction is not required for the problems to manifest.

I've had the same problem since upgrading an MBP (late 2013) to Sierra shortly after its release (I'm on 10.12.2 now). None of the suggestions from Apple Support brought any help. Even repair was suggested but is pointless really : restoring my timemachine backup to another laptop also brought the issue to that machine. I'm no pro but this suggests that it is a software error (OTOH : both machines are very similar hardware-wise so in theory it still could be hardware-related)

Interestingly, the problem only occurs when I'm connected to the network thru a dongle and ethernet cable, NOT when I'm on Wifi. (yes, I tried to replace the dongle several times: we have a bunch of them at the office). So a problem due to the fact that I have the latest MS Office 365 running on the machine (what we were thinking at first) can be ruled out also (IMO)

How I ended up on this thread : I've only recently noticed the WindowServer...userspace_watchdog_timeout.spin logfile because after yet a new reinstall I let the beachball spin during lunchtime. When I came back the file was there in the console. (google did the rest)

(sorry for breaking-in in this thread since it is not El Capitan-related)

No, this is helpful. Thanks. I've been toying with the idea of upgrading to Sierra to fix the problem, but these are production machines and can't afford any more down time, for any additional Sierra related issues that invariably pop up. For now, it seems a hard reboot every morning is still required.

Well, it's a bit of a mystery, but when the problem arises I can quit Photoshop and the issue to goes away immediately. But yes, it eventually returns, and if it returns because of some other unknown app that is accessing the graphics processor, the only recourse is to perform a hard reboot. For now, I'm just doing one every morning to clear out anything that could interrupt my work. This bug is entirely disruptive when it occurs in the middle of working. 😝 I had it pop up the other day when I was in the middle of a presentation. Not just disruptive, but embarrassing, as it took a long time to even get the Mac to respond to the restart command.

Reminds you of the System 7 (or Mac OS 8/9) days when Guy Kawasaki would joke

"Did you sacrifice to the demo gods?" when a presentation froze on stage...

OS X, now back to macOS, has come a long way, but this is a big setback.

I just had my latest stall.

The dates between the "WindowServer_(snip)" reports are:

Date/Time: 2017-01-06 23:29:10 +0900

Date/Time: 2017-01-13 00:01:19 +0900

-> 6 days, 32 minutes and 9 seconds.

This happened after changing locations (LAN to Wi-Fi, also meaning after a wake from sleep).

The stall manifested itself after I logged in to a separate account via Fast User Switch and launched Firefox (50.1.0).

The first user account was not affected, but multiple attempts in the 2nd account to launch Firefox caused stalling.

Quitting Firefox in the 2nd account seemed to stop the stalling, but did not try long enough to confirm this.

Console was showing repeated "0:01.SS firefox: CVCGDisplayLink::setcurrentDisplay didn't find a valid display - falling back to 60Hz" messages.

Duplicate bug or not, Apple please fix it... Good Night!

If you do not have this kind of message:

kernel[0]: stalling for detach from IntelAccelerator

or:

kernel[0]: stalling for detach from AMDTurksGraphicsAccelerator

or similar with your graphics card name in your system.log then your problem is not related to the problem here. There are several other reasons why the window server spindumps. I have not seen any log from Sierra with this error message from the kernel yet. Please check if this is the case on your machine.

Yes, "Console" shows the message multiple times (as also mentioned in my original reply to this thread):

Jan 13 00:01:36 [ComputerName] kernel[0]: stalling for detach from IntelAccelerator

This was shown for about a dozen times while I was trying to regain enough control to quit Firefox and logout...

I'll add my "me too" as well. Running build 15G1212 on a Late 2012 27" iMac. No real info to add other than the first 2 episodes in my case were while using iMovie. The next time Firefox triggered it. Reboot cleared it in all cases. All has been good until this afternoon...My uptime is now 6 days 20:42. FF seems to have triggered it again every 10 minutes from 14:15:22 to 18:50:46, which I attribute to having 10-minute autorefresh page in one of my tabs.

This is the last few log entries occurred while trying to type this reply. Restarted FF so I could finish.

1/12/17 7:19:08.000 PM kernel[0]: stalling for detach from nvAccelerator

1/12/17 7:19:23.000 PM kernel[0]: stalling for detach from nvAccelerator

1/12/17 7:19:48.000 PM kernel[0]: stalling for detach from nvAccelerator

1/12/17 7:20:03.000 PM kernel[0]: stalling for detach from nvAccelerator

1/12/17 7:22:05.000 PM kernel[0]: stalling for detach from nvAccelerator

1/12/17 7:22:20.000 PM kernel[0]: stalling for detach from nvAccelerator

Adding my 'me too'.. I've noticed the issue since installing the Security Update.. (This doesn't seem like the kind of change that belongs in a security update.....)... same version 1212, same symptoms (kernel: Stalling for detach from IntelAccelerator)... It's to the point that this laptop has become mostly useless.. I should have time machine'd it before installing the update.. yeesh.. (Note: this is on a MacBook Pro (Retina, 15-inch, Mid 2015))...

My Late 2014 iMac Retina 5K, 27-inch on 10.11.6 does this. Anecdotally, running VLC seems to provoke it:

kernel: stalling for detach from AMTongaGraphicsAccelerator

then there will be a watchdogd and spindump entry for WindowServer.

Hi there,

I can confirm that killing UserEventAgent is fixing the problem temprorarily:

sh-3.2# ps -xa | grep -i UserEven

49 ?? 2:09.92 /usr/libexec/UserEventAgent (System)

330 ?? 0:13.25 /usr/libexec/UserEventAgent (Aqua)

96275 ttys000 0:00.00 grep -i UserEven

sh-3.2# kill -9 49

sh-3.2# kill -9 330

sh-3.2# ps -xa | grep -i UserEven

96303 ?? 0:00.21 /usr/libexec/UserEventAgent (System)

96309 ?? 0:00.23 /usr/libexec/UserEventAgent (Aqua)

96312 ttys000 0:00.00 grep -i UserEven

detach is gone by now without a reboot.

Great analysis, good find.

> Reminds you of the System 7

I was exactly thinking the same. It's uncanny. Similar circumstances, after Steve Jobs had left. Took about 6 month for the Mac System 7 to become so bug ridden it was unusable, and we had to switch to Windows 95 (which had pretty mediocre stability BTW, but was usable). The quality of recent Apple OS releases is similarly appalling. Attention to quality is gone in the name of frenetic feature release - the same as what killed System 7 and post-Jobs Apple. It's simply not tested properly, and releases of the poorly checked updates are much too frequent. As unreal as it seems, the Microsoft platform is presently better quality wise. I am actually forced to switch back on laptops! Aargh! It's 1995 all over again! Forced back into mediocrity. Hope this gets fixed.