SUpro 7 on OSX 10.5.6 locks up entire UI for all apps

Nicoloco

Having exactly the same problem on three different machines in the office. System locks up and requires a hard reboot after about 10 minutes. I have filed a bug report with Sketchup and have started a thread on the Goodle Groups Sketchup Pro board.

http://groups.google.com/group/Sketchup-Pro/browse_thread/thread/8f43b383f71ac697/7cb353f62aae219a?hl=en#7cb353f62aae219a

I have discounted Time Machine as we are not using it. Also, my Mac Pro doesn't have an Airport, so I don't think its that. Also, it seems independent of video cards as people with NVidia and ATI Radeon cards are experiencing the same problems.

Interestingly, it is only happening on one particular job and not on another. Both models had been created in SU6 from DWGs and were pretty complex. They have both been purged and checked/fixed. I will try at lunchtime with a completely new SU7 model and see if I can crash it; which should discount the legacy artifact theory.

Other things I have noticed:

Turning off 'Hardware Acceleration' seems to make it stable (no crashes in 8 hours of (very slow) use)
Turning off 'Auto-Save' seemed to be working, but then experienced an application crash.

Seriously though, I need to get back to doing my job of designing things, so if anyone finds a fix that would be great!

Nick

Gaieus

Hi Nick and welcome (although would be nicer if you didn't have to come here with this bad news).

I'm personally on PC so won't be much help however Catamountain over the Help Groups is quite knowledgeable at these fields as I've noticed.

Hopefully someone with more Mac experience will chime in.

tim

Nico - yes, I ran with TM turned off just in case and it made no difference. One interesting thing I have noticed is that the alloc/free error logged is always for the same address no matter what apps I am runnig, whether I've rebooted, got several SU models open or whatever.
I too have done various purges with no improvement. I haven't used any extension ruby apps either.

As an experiment to see if it was perhaps some apple kernel library change from recent updates I've been using SU6 for the last couple of days with none of the related problems.

Nicoloco

Tim,

Agree with that. Half of our office are still using Sketchup 6 Pro and none of them (out of 10) have reported a crash whilst it is happening to all (6 in total) of the Sketchup 7 Pro users. Everyone is running OSX 10.5.6. Happening to a variety of machines...Mac Pro, Intel iMacs and MacBook Pro. Can't see a common link in terms of video cards as they are a mixture of NVidia and Radeon cards. Turning off 'Hardware Acceleration' fixes the problem in all cases, but makes modelling nearly impossible.

Other programmes that are usually running at the same time:

Safari/Firefox, Mail, AddressBook, CS3, iTunes, Vectorworks 2008, Office.

However, I have started up the computer, disabled all of the startup items, booted it without extensions, not connected to the server and ONLY run Sketchup Pro 7 and the same thing happens.

How did you get the information about where the memory allocation problem was? I will do that next time (at lunchtime) and see if it matches yours.

Nick

tim

Nick, I've been running the Console application to keep an eye on the errors being reported as the system runs. It's quite.... alarming. So many apps reporting failed this, can't do that and the kernel won't do the other.

The Console has a list on the left that shows all the logs it maintains. The two that seem important here are the 'All Messages' log which has been listing a lot of alloc related messages, and the 'WindowServer.log' (expand the 'LOG FILES' entry by clicking on the triangle) which was showing the backing store type messages.

On the plus side I just received a message from the google-guys suggesting a full de-install and re-install since they suspect some component was not installed correctly. They suggest :-

@unknownuser said:

Uninstall Google SketchUp Pro 7 by visiting the following Google
SketchUp Help Center article:

http://sketchup.google.com/support/bin/answer.py?answer=36209

After uninstalling, we suggest repairing disk permissions:

a. Restart your computer.
b. Open a Finder window.
c. Click "Applications."
d. Click "Utilities."
e. Double-click "Disk Utility."
f. Select your hard drive.
g. Select the "First Aid" tab.
h. Click "Repair Disk Permissions."
i. After the permissions are repaired, exit Disk Utility.
j. Please repeat steps "a" through "i" once more.

Log in to your computer with a user account that has administrative
privileges.

Please visit
http://sketchup.google.com/support/bin/answer.py?answer=115548 and
download the Google SketchUp Pro 7 install file.

When the download completes, double-click on the install file and
follow the on-screen instructions.

Please launch Google SketchUp at least once while still logged into the
administrative user account.

It certainly seems to have massively reduced the alloc related messages so I'll see how it goes.

Nicoloco

Tim,

Thanks. Have made a clean installation of Sketchup 7 Pro as per the Google instructions. Had a look in the console and there are still lots (about 4000 in three seconds...) of the following messages.

09/02/2009 12;11;48 [0x0-0x110110].com.google.sketchuppro7[761] *** set a breakpoint in malloc_error_break to debug

and

09/02/2009 12;11;48 [0x0-0x110110].com.google.sketchuppro7[761] SketchUp(761,0xa077d720) malloc; *** error for object 0xa06eb6d8; Non-aligned pointer being freed

Have turned off 'Fast Feedback' and left 'Hardware Acceleration' on and this has eliminated the crashing (none yet today) without too much of a performance hit. It doesn't seem to have any effect on the error messages.

Nick

tim

Oh well, nice idea google-guys. I just had a repeat of the exact same crash as before. Let's see what the next suggestion is like!

bjanzen

Tim -

Those console msgs have been there for a couple of versions, and yes, it'd be nice if someone cleaned them up. Do you get the Apple crash log showing up? If you do, don't send to Apple - send it to us by copying the text in it and sending it to us in an e-mail so we can troubleshoot. I'll poke through this and other things and see if we know anything more. Thanks.

Barry

tim

Hi Barry -
no, there is no crash report as such, just the log messages in the windowserver.log and then total UI death. It may be of interest that iTunes keeps on playing, and don't forget that I can log in from one of my other macs and poke around. That might allow some sort of remote debugging experiments? Let me know. My time is my own for a week or so, so I can spend some if it working on this with you if you have tests for me to try.

It might sound callous but I'm glad someone else with different equipment is having similar crashes. At least that offers some more chance of reproducibility. I used to hate intermittent errors when I was managing s/w development...

TriciaS

A few more questions on this one.

Nicoloco - are you still not crashing with Fast Feedback off? If so, that is a really important clue. If you turn on Fast Feedback, can you recreate the issue with all models or just certain ones? It sounds like you were thinking it might be model related...

Tim - does turning off Fast Feedback in Preferences > OpenGL seem to fix your lock up as well?

Fast Feedback speeds up the drawing actions in your model (vs things like the speed at which you can orbit around) and only kicks in when performance degrades below a certain threshold. So Fast Feedback will only kick in when working on larger models. If you both are only seeing this issue with more complex models when Fast Feedback is enabled, that might be a further clue.

Also, we are wondering if you see a pattern in how many document windows you have open at the time of the crash. Or whether it seems to happen only with certain actions - e.g. only when adding geometry, or when selecting things etc...

Sorry for all the questions. We really appreciate your willingness to help us try to get to the bottom of this one.

Tricia

tim

@unknownuser said:

I'll try it and see.

@unknownuser said:

Also, we are wondering if you see a pattern in how many document windows you have open at the time of the crash. Or whether it seems to happen only with certain actions - e.g. only when adding geometry, or when selecting things etc...

I very rarely have more than 2 document windows open - so rarely I can't remember doing it. I do tend to have most of the tool windows open (layers, outliner, styles, components etc) most of the time. I can't discern any pattern of actions involved.

My model is not particularly complex
30,000 edges
13,000 faces
622 component instances
2 groups
3 images
235 component definitions
10 layers
47 materials
1 style
everything else 0

Don't worry about asking questions; I'll only get worried when you d

I'll try it and see.

@unknownuser said:

Also, we are wondering if you see a pattern in how many document windows you have open at the time of the crash. Or whether it seems to happen only with certain actions - e.g. only when adding geometry, or when selecting things etc...

I very rarely have more than 2 document windows open - so rarely I can't remember doing it. I do tend to have most of the tool windows open (layers, outliner, styles, components etc) most of the time. I can't discern any pattern of actions involved.

My model is not particularly complex
30,000 edges
13,000 faces
622 component instances
2 groups
3 images
235 component definitions
10 layers
47 materials
1 style
everything else 0

Don't worry about asking questions; I'll only get worried when you don't....

tim

@tim said:

@tricias said:

Tim - does turning off Fast Feedback in Preferences > OpenGL seem to fix your lock up as well?

I'll try it and see.

Sigh. No improvement at all. Exactly the same errors logged, exactly the same lock-up.

bjanzen

Nicoloco & Tim -

In Finder, type Command-Shift-g and type ~/Library/Logs/CrashReporter/ This directory contains all the crashes on your machine. You can send us a typical SketchUp crash.

Tim, yours messages point to WindowServer unable to create swap space. Is your hard drive nearly full, or are you running a lot of screens in Spaces? You posted that WindowServer threw errors - is WindowServer still running? If you remote'd in, do a "ps aux | grep Window" command is WindowServer there? You can also run "top", type "o" (to sort on) then "cpu" to sort on cpu usage. WindowServer will always be near the top. If it didn't crash, what's the memory usage (my machine, just typical usage, is 130 MB real memory, 500 MB virtual memory). If this is crashing or messed up, we can contact Apple and work with them to find out what's going on.

Please do or continue to send data (crash logs, system profiles, etc..) to

http://sketchup.google.com/support/bin/request.py

I'm still of the belief that your situations are unique - we have lots of users in 10.5.6 with zero problems, so we'll try to determine what's going on for you.

b

tim

@bjanzen said:

In Finder, type Command-Shift-g and type ~/Library/Logs/CrashReporter/ This directory contains all the crashes on your machine. You can send us a typical SketchUp crash.

There is just a single crash report - and that relates to a SU 6 crash that happened while I was trying to see if 6 had the same problem. It seems that SU itself is not actually crashing in any form that leaves a report.

@unknownuser said:

Tim, yours messages point to WindowServer unable to create swap space. Is your hard drive nearly full, or are you running a lot of screens in Spaces? You posted that WindowServer threw errors - is WindowServer still running? If you remote'd in, do a "ps aux | grep Window" command is WindowServer there? You can also run "top", type "o" (to sort on) then "cpu" to sort on cpu usage. WindowServer will always be near the top. If it didn't crash, what's the memory usage (my machine, just typical usage, is 130 MB real memory, 500 MB virtual memory). If this is crashing or messed up, we can contact Apple and work with them to find out what's going on.

I have 150Gb free space on my 300Gb disk. I don't run Spaces at all (nor dashboard for that matter). I do have a secondary 20" display on the 24" iMac. Typically I have the tool windows on that surface.
Next time the system locks up I'll take a look with ps and report

Thanks for looking into this!

bjanzen

OK, scouring the net, iMac's have had reports of overheating on gaming & OpenGL apps. I've replaced a personal one myself. At this point, I'd try 3 things:

Getting a menu bar temperature meter, maybe something like http://islayer.com/apps/istatmenus/
Unplug the second monitor for a day and see if you crash.
Stop playing World of Warcraft every coffee break . OK, maybe just 1 and 2; this one's important.

Barry

Gaieus

I didn't know WoW was release for the Mac!
I'll change platforms immediately!

(Barry, can you arrange a shift from the Windows license to the Mac for me?)

tim

@bjanzen said:

OK, scouring the net, iMac's have had reports of overheating on gaming & OpenGL apps. I've replaced a personal one myself. At this point, I'd try 3 things:

Getting a menu bar temperature meter, maybe something like http://islayer.com/apps/istatmenus/

Been using one ever since my first iMac had this problem and got replaced within a couple of days of buying it. No apparent temp issues with this unit.

@unknownuser said:

Unplug the second monitor for a day and see if you crash.

Owwww, Mum.... no, don't make me do that. It's not fair. Yuck. OK, I'll try it. The things I do to help people out, I dunno....

@unknownuser said:

Stop playing World of Warcraft every coffee break . OK, maybe just 1 and 2; this one's important.

Never played it. Why would I downgrade to simulated violence when I can just - well, never mind, you don't need to know about my day job?

Today's little test result - I turned off hardware acceleration (Oh my goodness.... it....... is...... so....... sloooooooowwww) and the crash is exactly the same. The same address is being reported as not freeable (0xa063a6d8) and the windowlog has the same complaint about destination buffer to small for y index (3852). And I clean forgot to log in from the other machine to look at the process list. Rats.

tim

OK, same error with the external display disconnected. Exactly the same errors etc.

This time I remembered to log in from the macbook and this is what it said:-

Diziet;~ tim$ ps aux | grep Wind
_windowserver    94   0.0 44.8  2605312 1878828   ??  Us    4;56PM   7;33.79 /System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Resources/WindowServer -daemon

and

Diziet;~ tim$ ps aux | grep Sk
tim        739   0.0  0.0    86524   1204   ??  Ss    5;45PM   0;01.28 /Applications/Google SketchUp 7/SketchUp.app/Contents/Resources/SketchUpDaemon 14639
tim        738   0.0  8.4  2485168 351256   ??  S     5;45PM   5;36.58 /Applications/Google SketchUp 7/SketchUp.app/Contents/MacOS/SketchUp -psn_0_241723

A note I don't think I've mentioned previously - the mouse cursor is still responsive to the mouse, so some part of the graphics system is still running ok and it can't really be something like the GPU overheating.

Christopher Love

Tim,
Just getting this up to the top of the list again. I too have started experiencing UI freezes, its not isolated. Only occurs whilst using SU. Very annoying. Sounds like I have a similar setup. Running an external monitor off the latest MacBook Pro. Just wondering wether or not some of the ruby scripts that I use could be causing the problem. Smustards dashed lines runs like a dog on V7, dont know that I could live without it.

Console log error message attached

(1000)
13/02/09 5:15:30 PM SketchUp[442] _NSShapePlainWindow: error setting window shape (1000)
13/02/09 5:15:30 PM SketchUp[442] _NXPlaceWindow: error setting window shape (1000)

Seriously considering going back to V6

Any suggestions greatfully recieved

Regards
Chris

tim

@christopher love said:

Tim,
Just getting this up to the top of the list again. I too have started experiencing UI freezes, its not isolated. Only occurs whilst using SU. Very annoying. Sounds like I have a similar setup. Running an external monitor off the latest MacBook Pro. Just wondering wether or not some of the ruby scripts that I use could be causing the problem.

Hmm, it's probably not any (non-standard) rubys at fault since I'm currently running a totally vanilla SUP7 install to avoid any possible conflicts.

Can it really be true that my WindowServer is trying to use 44% of all memory? (see the ps aux output in my previous message? That must imply a truly staggering memory leak or abuse somewhere.