Weird WinXP problem.

[Goragoth]

Normally I try google first to see if there's anything out there to explain what is going on but I really don't know even where to begin searching for this problem.

Ok, at first I thought it was just some freak occurance but it has happened a few times now so I know that something is definatively wrong. What happens is that sometimes the Windows UI becomes pretty much unresponsive. Some things do work. Ctrl+alt+del brings up task manager and it draws its pretty graphs looking pretty much normal but you can't click on anything (like the computer is too busy but this is on dual core machine and taskman is showing 1-2% CPU utilisation so that ain't it). The mouse still moves around but clicking seems to have no effect. Using hotkeys on the keyboard to launch apps (e.g. calculator) works but you can't type anything or navigate any menus or anything.

So far this has happened once particularly badly when trying to render something in Maya with mental ray, making the machine utterly unusable resulting in me doing a hard reboot after about an hour of trying to fix it. Other than that it seems to happen just about every time after exiting Diablo2 for just a few seconds before fixing itself. It also just happened while trying to copy some data from a CompactFlash card (also Windows was throwing some errors at me for some of the files); in this case it again fixed itself after a little while.

It seems like each time there is some system resource that is being blocked and Windows refuses to let me do anything (except for ctrl+alt+del and pressing hotkeys) until the resource is freed again (but its not the CPU, if taskman is to be trusted).

This is really driving me crazy and its not something that I have ever come across in about 7 or so years of heavily using Windows (NT4 -> 2k -> XP). If anyone has experienced anything similar, heard about it or even just has a good idea of what to google for, it would be really appreciated.

My system specs: a64 x2 4800+, 4gb ram, 74gb Raptor x2 (in raid0), 300gb Seagate Barracuda, ati x1900xtx, winxp sp2. Other bits and pieces: DEP on (for essential services), pagefile turned off, large system cache on. All the latest drivers installed. AMD cool&quiet driver installed (but turned off). That's about all the relevant information that I can think of. This is a pretty new system, only maybe three weeks old or so. I didn't notice anything at the beginning but then I've only really been pushing it lately and it only seems to happen under certain circumnstances as I described above.

 

[maaoouud]

Basically one or more of your programs are busy when windows UI requires an update from it. Can't really be fixed, windows will just keep at it until the program handles whatever it's supposed to do or is shut down (we all love the taskmanager). Sometimes killing the program doesn't fix it and then it's just an endless loop that windows is stuck in.

I do have some advice on what not to do. Do not change the resolution when a program is busy with something (like maya and mental ray) since that will make them redraw the GUI. Also I think manually refreshing [F5] can trigger it sometimes.

EDIT: Where'd that newline come from ?

 

[Goragoth]

Basically one or more of your programs are busy when windows UI requires an update from it. Can't really be fixed, windows will just keep at it until the program handles whatever it's supposed to do or is shut down (we all love the taskmanager). Sometimes killing the program doesn't fix it and then it's just an endless loop that windows is stuck in.

I do have some advice on what not to do. Do not change the resolution when a program is busy with something (like maya and mental ray) since that will make them redraw the GUI. Also I think manually refreshing [F5] can trigger it sometimes.

Well its nothing this simple. It isn't just the program that is unresponsive. All of Windows is. I've never seen anything like it and I've had to kill a lot of unresponsive programs. Like I said, you can't click on anything in taskmanager so there's no way to kill the offending program. I've had Windows be pretty unresponsive when a program is hogging all of the CPU (like when its stuck in an infinite loop) but this is different to that as well, and CPU usage is basically zero.

I've now had this happen everytime I start Warcraft3. This didn't happen before but now everytime I try to start it Windows becomes all unresponsive like that and just the splash screen shows and it sits there for about a minute and after lots of frantic ctrl+alt+deleting it finally goes into the game. This is new.

I'm currently thinking that possibly some driver is messed up, maybe causing some resource to be blocked when it shouldn't and Windows waiting until the resource is freed up again. Or there's some incompatibility between one or more applications or services running on my system. Or my system has been 0wn3d by some piece of nasty malware (although since then I've run ClamAV, AdAware and Spybot S&D over the system and kept a close eye on running processes using ProcessExplorer, none of which have turned up anything).

If I can't figure this out within the next few days, I'll just back everything up and do a wipe but I would really rather not have to resort to that. I've just installed nearly all of the 20-30 odd programs that I just can't live without.

 

[Rodéric]

I have the weird feeling this is related to SMP, try disabling one of your cores (if possible) and do one of the things you know will make the system unresponsive to see if it's coming from here.

 

[Ty]

Like I said, you can't click on anything in taskmanager so there's no way to kill the offending program.

Just a little FYI, you can use Tab to cycle through choices (menu focus areas). That will allow you to interact with Task Manager without a mouse.

 

[AlNom]

could the RAID array be borked?

 

[Frank]

Did you check the event log?

 

[Goragoth]

Did you check the event log?

Actually, no I didn't. How stupid of me. Now that I did... lot's of warnings there like this:

An error was detected on device \Device\Harddisk3\D during a paging operation.

I'm thinking that's not so good. Might be a borked RAID array. It could certainly explain a thing or two. I'll have to look into that.

 

[AlexV]

It may not be that simple. Did this behaviour by any chance become apparent after installing the 6.5 or 6.6 Catalyst drivers?

 

[_xxx_]

Try defragging your drives and turning on the swap file with a firm value for the size, just for testing. Although it sounds like you have some apps running which have some memory leaks or which fragment the RAM very badly.

 

[PeterAce]

An error was detected on device \Device\Harddisk3\D during a paging operation.

Very much a early sign of Bad Blocks on the HDD(s).

Or a disk controller error (controller going bad?)

 

[Bobbler]

Generally low CPU usage + unresponsiveness will tend to mean something with either the HDD or other media drive not being handled correctly. Optical drives are especially offensive in this regard. Sounds like your issue has to do with the HDD/RAID though (be it caused be something else messing with it or the actual raid/drives failing).

 

[Goragoth]

I've been meaning to reply to this thread for a while but was caught without Internet for over a week (ugh!). Anyway, this odd freezing continues to annoy (and intrigue) me. The disk error seems so far to be unrelated (after some searching on the Internet I found this to be related to the Raptors and the fact that they don't support NCQ, and turning off command queing in the nvSata drivers has made these errors go away completely).

So, some more information:
The Visual Studio debugger will trigger this behaviour every time. If I launch a program with debugging it will sit there with a busy cursor and being unable to click on anything until hitting ctrl+alt+del, which will usually clear the problem as soon as taskmanager pops up (sometimes multiple presses are required to get to this). Same thing happens when the JIT debugger attaches itself to a crashed program. The only other reliable trigger seems to be launching Warcraft3:TFT (which worked fine for a while). Again, pressing ctrl+alt+del will usually clear the problem up after a bit, following which Warcraft3 runs fine. I did suspect for a while that it wasn't ctrl+alt+del that was bringing it back but just a timeout but leaving it for a little while always seems to keep it in that unresponsive state until I give it the three finger salute. I do remain skeptical on this point though as it seems really unusual.

Another point of interest is that doing a user switch (with Winkey+L since the mouse won't let me click anywhere), logging into another user, and then killing the offensive program in taskman there, will always clear up the problem as well.

Furthermore, in my other test account, all of these applications (i.e. VS debugger and Warcraft3) work fine without any problems whatsoever (even with my other account still logged in, i.e. using fast user switching). Of course my first suspicion was some sort of software conflict, maybe a service or something. Using the autoruns tool from SysInternals I compared all the autoruns for both accounts and found no significant differences (and yes, I had tried my other account with no other programs running). Since Warcraft3 ran fine until some point I suspect it was something I installed in between that time but I can't remember what it might have been although possible culprits are:

* Moving from the nVidia display drivers to the Catalysts after swaping my defective GF7 out for the ATi card.
* Installing Daemon tools 4.03 (tried uninstalling this to no effect and it runs in the other account as well anyways so I'm doubtful to it being the culprit)
* Using the files and settings transfer wizard to move my account from my 2k box. Maybe messed up the registry somehow? This one at least might explain why the other account appears to work fine.

At this point the simplest fix may just be to create another account and move everything to it (assuming of course that it really is tied to my current account). I'll keep investigating and should I come to find the source of the problem I'll share it. I might try using a different account for a while and maybe try using regmon and filemon to see if I can spot anything suspicious.

 

[Frank]

In my experience, if the harddisk goes, it goes all at once. Strange errors mean, that your IDE/SATA interface is broken. Which is the most common thing to break on motherboards.

 

[_xxx_]

* Using the files and settings transfer wizard to move my account from my 2k box. Maybe messed up the registry somehow? This one at least might explain why the other account appears to work fine.

Duh, people are actually using this crap "feature"? I vote for this as the cause for your problems, the same happened to me a while ago at work, we were migrated to a new domain and the account transfer f**ked it all up. After the ITidiots (don't ask, our IT are idiots indeed) did a manual reinstall/configuration, it was all fine.

 

[Guden Oden]

Anyway, this odd freezing continues to annoy (and intrigue) me.

Dude, just reinstall already, okay? It's windows rot you're experiencing, reinstall will make it go away. Well, unless there's an underlying hardware fault of course, but then you'll notice that for sure when your fresh OS starts misbehaving immediately.

So go fish that OS install CD out of the back of your computer junk closet and get cracking, hehe!

 

[L233]

Could you post a list of the processes you are running (task manager, second tab)?

 

[Frank]

Could you post a list of the processes you are running (task manager, second tab)?

Hijackthis is the best tool for that.

 

[Goragoth]

Heh, I was getting the disk error on Disk3 in the event viewer again, so I had a hunt around the device manager to figure out exactly which disk this is (being rather worried about failing harddisks and the like). Turns out this is my CF card reader slot and the errors were happening while reading photos off the card. Phew. I should probably not trust that particular CF card anymore but it isn't used for anything important anyhow.

It also seems like I have fixed my problem, probably to do with the network adapter installed by the ATi drivers (why, oh why, do they need a network adapter anyway?). I have disabled it and now I'm not having the freezes anymore (at least as far as I can tell after a short time of testing). I wouldn't be surprised at all if the debugger and Warcraft3 both do a quick query to all network adapters for some reason or another and freeze up while waiting for a response. I hope this has finally fixed it.

 

[The_Wolf_Who_Cried_Boy]

Dude, just reinstall already, okay? It's windows rot you're experiencing, reinstall will make it go away. Well, unless there's an underlying hardware fault of course, but then you'll notice that for sure when your fresh OS starts misbehaving immediately.

So go fish that OS install CD out of the back of your computer junk closet and get cracking, hehe!

Without wishing to hijack the thread, what exactly is Windows problem. I've defragged weekly but still see a clear difference between the trojan induced fresh install last weekend and the nine month old install it replaces.

I've never run a registy cleaner, do they make a difference?