NVDA(OpenGL): Channel exception!

Update to my original post in this thread: http://discussions.apple.com/message.jspa?messageID=11792654#11792654

In brief: it may be my RAM.

Read on...

My system managed to stay stable for about 10 hours on June 29, from 7am to roughly 5pm. I was running OS X Mail and VMWare Fusion (with Win 7 + Firefox 3.6.6 + Selenium IDE + VS 2010). No trouble.

Later that evening while simply watching videos on YouTube (via OS X Firefox 3.6.6) my screen started flickering like mad and though I could move the mouse, no input was being registered. Eventually I just held down the power button to power off.

I spent all day on June 30 and July 1 trying to find a solution to the problem. I've run AHT from my original install disk, for 5 hours in loop mode. I've run ASD 3S116, both EFI and OS level tests in loop mode for 8 hours. I've tested the SMART parameters on my disk. I've run a surface scan on my MBP's disk by booting it into target disk mode and running the scan from another machine. I've tested the VRAM.

Not one thing indicates a hardware problem on my MacBook Pro (MacBookPro3,1) 15" 2.2 GHz Core 2 Duo from mid-2007 (Santa Rosa chipset). No RAM errors, no bad blocks on the disk, no sensors or fans broken. Absolutely nothing.

So I figured that maybe it was a software issue, as it cropped up soon after the 10.6.4 update.

I was ready to erase the drive and reinstall OS X, but then I found this thread over at the Steam forums which posted a clever work around that replaces the 10.6.4 NVidia and GeForce kernel extensions (kexts) with their 10.6.3 versions:

http://forums.steampowered.com/forums/showpost.php?p=15705859&postcount=73

NOTE: the instructions in that post aren't very good, and I don't have time to post corrections. So if you attempt this, you MUST be technically savvy with Unix commands or you could screw up your system. You've been warned.

I did this and it appeared to work... for about 2 hours. Then flicker, flicker, flicker forever. I left it in that state overnight to see if the flickering would sort itself out. It didn't. Flickered all through the night.

This morning, in a last ditch effort, I replaced my current RAM with the original RAM that came with my machine. I've been running stable for about 2.5 hours now. I've stress tested by running Portal on Steam for a few levels, then quitting that and simultaneous watching two HD movie trailers on YouTube in Firefox. As I type this I'm running Mail, Firefox (streaming a movie on YouTube), and my Win7 virtual machine in VMWare.

Still a bit early to be certain, but... so far so good.

Point of this wall of text: if you've upgraded your RAM and you still have your original RAM, swap it and try it even if your memory tests don't reveal anything wrong with your current RAM. It might save you a trip to the Genius Bar.

FWIW, no test using software running on the host computer is 100% reliable at identifying defective or sub-par RAM. Even dedicated hardware memory testers can take hours to run through all the possible memory states of just one large module & intermittent problems that occur in only a few states could take weeks of testing to reveal.

The only completely reliable test users can perform is substitution with known-good modules (sometimes a problem in itself) to locate bad ones by process of elimination.

Making matters worse, some modules work fine with one OS version but not another because memory use patterns & timing can vary from version to version. Because of this, it is a good idea to buy aftermarket memory modules with a lifetime replacement warranty.

Yeah... I know this but still fooled myself into running memory tests for hours rather than trying the simplest thing first: switch to the old chips. I think that's the most galling thing of all. I could've saved hours of nonsense and stress had I just tried different RAM.

That being said, a TON of people are getting this problem right after the 10.6.4 update so there's SOMETHING in there that is devastating people's machines, and there's no way that they're all configured exactly like mine. So it may not be RAM in all cases - but then again, maybe there is some connection. Maybe it has to do with amount of RAM - not everyone is disclosing their exact specs when talking about this issue and thus it gets muddled (for instance, I was running with 4 GB and that was causing problems; perhaps there is no problem with my RAM but rather with using more than 2 GB on particular machines).

I just don't want people to automatically jump to the "NVidia 8600M is broken, get it fixed" conclusion and rush out to their Apple service department when it might be something much more subtle and less time consuming to repair.

Hi guys,

I too have intermittent freezes of up to 10-20 seconds on my 2006 24" iMac with NVIDIA GeForce 7300 GT. After each freeze the machine comes back up normally, sometimes with video corruption of my Tweetie for Mac window.

My mouse pointer stays available during the freezes but doesn't do anything. These problems started when I upgraded to Snow Leopard - and it hasn't been fixed in any subsequent point releases (I'm on 10.6.4 now).

I looked in the console and noticed the same errors everybody is reporting:

--

9/7/10 1:41:51 PM kernel NVDA(OpenGL): Channel timeout!
9/7/10 1:41:51 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
9/7/10 1:41:51 PM kernel 00000069

--

It's pretty appalling that so many people have complained about this and it still isn't fixed. Don't think Apple doesn't read these forums because simply, they do.

Anyway - just chiming in that I'm having the same problem. Glad to know it's not related to my (trusty) hardware. I'm now off to scour the internet if there's a fix that doesn't involve Apple. Too bad I can't download new drivers for my videochip...

Lars Pasveer wrote:
It's pretty appalling that so many people have complained about this and it still isn't fixed. Don't think Apple doesn't read these forums because simply, they do.

It isn't as simple as you might think. As explained here, there is a group of Apple employees who among other duties forward info from user contributions to Discussions to other Apple groups. However, keep in mind two things:

• The forwarded info can be no better than the technical details of what users contribute. Thus, it is important to include all the relevant technical details in your posts. Things like "I have the same problem" with little other info is almost useless, especially when users don't take the time even to mention exactly which Mac they are using, what OS version it is running, or other basic details. Plus, when topics grow large they tend to become a grab bag of similar yet distinctly different problems, making it hard to know which "same" problem a post refers to.

• The volume of posts to Discussions is enormous. In fact, as a little browsing in Discussions > Feedback about Discussions will make clear, it is so large that it sometimes exceeds the limits of the only enterprise class software available that can run it. This volume makes it impractical for Apple Hosts to read all the posts, much less to comb through them for relevant technical details.

So don't assume that just complaining about a problem in Discussions will do any good. A hundred posts short on relevant technical details & long on what "everybody" is seeing isn't worth one post with comprehensive info like a complete crash log, all the steps tried to troubleshoot the issue, etc. Even if you don't submit anything to Mac OS X - Feedback, which guarantees someone will see it, try to include at least the same technical details in your post that you would in a feedback report.

Yes, I understand the ecosystem of this discussion forum and Apple's hands-off approach to it. I also understand the volume of posts here is daunting, but any forum has a kind of 'self redaction'. This topic for instance was started in 2009, has currently had 39,992 views and contains 200 posts.

It bubbles to the surface, so to speak, just by the shear number of people reporting problems. They all have basically the same error messages and symptoms, so that gives the Apple folk something to investigate.

I have left info at Mac OS Feedback, including tech details and so on. But by Apple's own admission they will read that, but probably not reply. So it's basically a memory hole you drop your support request into.

Which brings us to this forum and whether or not Apple reads it. Recent antenna reports on the iPhone 4 were heavily redacted and some threads were even removed, so there is al least some conscious editing going on.

I just find it curious that it takes a year of reports on this topic, with nearly fourty thousand views and there's still no solution or indication Apple is working on it. It is a genuine, recurring problem people have.

I'd put my money on an OpenGL driver problem. Just now I had a freeze again, and my Console reports the same stuff all the time:

--

10/7/10 2:23:24 PM kernel NVDA(OpenGL): Channel timeout!
10/7/10 2:23:24 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
10/7/10 2:23:24 PM kernel 00000069

--

This is only something Apple can fix and not something I'm eager to try myself, with kernel, kernel extensions and other low-level tweaks.

Lars Pasveer wrote:
I have left info at Mac OS Feedback, including tech details and so on. But by Apple's own admission they will read that, but probably not reply. So it's basically a memory hole you drop your support request into.

If you want to look at it that way, Discussions is an even bigger hole: it isn't for support requests to begin with & it is filled with user speculation, off-topic diversions, vague & incomplete information, & often Subject headings so broad or obtuse that they are practically guaranteed to accumulate posts about many substantially different issues. It is at its most useful when a topic concentrates on one user's issue, at its least when dozens of users add a bunch of "me too" comments without specifics that mostly confuse things instead of clarifying anything.

Which brings us to this forum and whether or not Apple reads it. Recent antenna reports on the iPhone 4 were heavily redacted and some threads were even removed, so there is al least some conscious editing going on.

It should come as no surprise that the forum moderators edit posts or remove threads that do not meet Discussions guidelines; it may be a surprise that a large number of requests for moderator attention come from forum users -- anyone who reaches level 2 or above has "report this post" links below every post to do that. The number one reason posts are removed is for violating the terms of use, a frequent occurrence when users are more interested in ranting than posting anything constructive. That is very annoying to many users that volunteer their time to help others, not to waste their time wade through a lot of useless venting, so the most blatant posts tend to get reported very quickly.

I just find it curious that it takes a year of reports on this topic, with nearly fourty thousand views and there's still no solution or indication Apple is working on it. It is a genuine, recurring problem people have.

Views don't mean much. I subscribe to dozens of topics & read every new post even when my interest is in helping solve an issue I don't have. Many others do the same. Others are just curious or are looking for issues that might apply to them. What really counts is the number of unique users posting that they have the identical issue & you can't get that from either the view or post count.

Aside from that, it is well known that Apple almost never comments on an issue until it has a solution, & when it does you are likely to find out about it in Discussions only if some user posts a reference to it.

This is only something Apple can fix and not something I'm eager to try myself, with kernel, kernel extensions and other low-level tweaks.

Apple usually fixes what it breaks, & occasionally what third party software does. But an error like a channel timeout exception is not necessarily due to a fault in the OS X kernel or an Apple-supplied kernel extension. Exceptions occur all the time; the robustness of a kernel is measured by its ability to recover from them. If a particular app is the only one causing them, chances are good that the fault is in the application's programming; if several of them cause them it is more likely an extension is involved, but not necessarily an Apple-suppled one.

This is why it is important to include comprehensive info -- what may seem to be the fault with something Apple supplied may or may not actually be the real culprit.

I have a little more information just in case it can help. I can't consistently cause the Channel exception messages to start appearing in the log, but I can make them go crazy.

One I have a few consecutive Channel exception messages, if I grab the Console window and scrub it around the screen quickly then I'll get a string of messages. For example, I shook it at 4:54, then started writing this, then shook it again at 4:57, here's what the log showed:
7/13/10 4:54:21 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
7/13/10 4:54:21 PM kernel 00000069
7/13/10 4:57:54 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
7/13/10 4:57:54 PM kernel 00000069
7/13/10 4:57:55 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
7/13/10 4:57:55 PM kernel 00000069
7/13/10 4:57:56 PM kernel NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
7/13/10 4:57:56 PM kernel 00000069


Hardware Overview:

Model Name: iMac
Model Identifier: iMac6,1
Processor Name: Intel Core 2 Duo
Processor Speed: 2.33 GHz
Number Of Processors: 1
Total Number Of Cores: 2
L2 Cache: 4 MB
Memory: 2 GB
Bus Speed: 667 MHz
Boot ROM Version: IM61.0093.B07
SMC Version (system): 1.10f3
Serial Number (system): QP7250JPVGP
Hardware UUID: 00000000-0000-1000-8000-001B632E7F59


NVIDIA GeForce 7300 GT:

Chipset Model: NVIDIA GeForce 7300 GT
Type: GPU
Bus: PCIe
PCIe Lane Width: x16
VRAM (Total): 128 MB
Vendor: NVIDIA (0x10de)
Device ID: 0x0395
Revision ID: 0x00a1
ROM Revision: 3022
Displays:
iMac:
Resolution: 1920 x 1200
Pixel Depth: 32-Bit Color (ARGB8888)
Main Display: Yes
Mirror: Off
Online: Yes
Built-In: Yes
Display Connector:
Status: No Display Connected

My open applications are: iCal, Mail.app, Safari, BBEdit, Activity Monitor, Console, System Profiler

I'm sure you're all enthralled with my last message, but wait, there's more!

I can isolate it more:
It's not scrubbing back and forth that makes it happen, it's up and down. The messages happen when the Console window touches the dock and it is reflected on the 3D-ish surface of the dock. I believe that I also get it when a save sheet drops down from a document.

So I believe that this has something to do with transparency. That would also explain why SL and WoW make it happen.

Wow, these 7300GT cards just stink huh.

I've had this happen to me on my 2006 Mac Pro with the 7300GT. In fact, I just had it happen.....exactly like you Lars.....Screen froze, mouse worked, couldn't interact with anything, came back after 10-20 seconds.

I also noticed after it came back the video corruption in Tweetie.

Here's my errors:

Jul 23 14:23:54 kernel[0]: NVDA(OpenGL): Channel timeout!
Jul 23 14:23:54 kernel[0]: NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
Jul 23 14:23:54 kernel[0]: 00000069


I guess we can all safely assuming that Apple and NVIDIA don't really care about our OLD machines and cards.

-Kevin

I got this error today. I was on Safari and a flash add loaded very badly making the screen flicker black once. After quitting it I moved a window of another app and all **** broke loss. The computer crashed and hung and the entire bottom of the screen started flickering black and the only way to fix it as to hard restart.
Permission fix and AHT didn't come up with any issues so atleast I know its not a hardware problem. Scared the life out of me as its not what you expect out of a Mac. I just hope it was a one off and that the computer isn't harmed.

It's not just limited to the 7300 series, I've got a 09 UniMBP 17" 2.93 Ghz and over the last 2 months it sometimes will freeze on wake (when Quartz Composer was running) and periodically just freeze. Now you mention it, I think it's been happening since I upgraded to 10.6.4.

This morning the bottom tenth of the screen started to flicker, periodically... I checked console and found this error message, the flickering started to fill more and more of the screen until the whole screen went black. Then the top 5th would flicker whilst the rest remained black.

The only way I could do anything was to force it to shutdown and reboot, the console was full or two errors repeating.

8/5/10 9:06:27 AM kernel NVDA(OpenGL): Channel exception! exception type = 0xd = GR: SW Notify Error
8/5/10 9:06:27 AM kernel 0000006e
8/5/10 9:06:27 AM kernel 00200000 0000502d 00000472 00000080
8/5/10 9:06:27 AM kernel 0000047a 000002a4 00000000 00000003
8/5/10 9:06:27 AM kernel 00000000 00000000 00000000
8/5/10 9:06:27 AM kernel 00000080 00000000
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f494e54523a3078323030303030
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f455843455054494f4e3a30783830
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f5052495f545043305f534d435f4857575f4553523a30
8/5/10 9:06:34 AM kernel NVRM-RC: 1 7838303031303030300a
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f5052495f545043305f534d435f534d5f4354585f353a
8/5/10 9:06:34 AM kernel NVRM-RC: 1 534d303a307831300a
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f5052495f545043305f534d435f534d5f4354585f383a
8/5/10 9:06:34 AM kernel NVRM-RC: 1 534d303a3078320a
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f5052495f545043305f534d435f534d5f4354585f353a
8/5/10 9:06:34 AM kernel NVRM-RC: 1 534d313a307831300a
8/5/10 9:06:34 AM kernel NVRM-RC: 0 4e565f5047524150485f5052495f545043305f534d435f534d5f4354585f383a
8/5/10 9:06:34 AM kernel NVRM-RC: 1 534d313a3078320a




and

NVDA(OpenGL): Channel exception! exception type = 0x6 = Fifo: Parse Error
8/5/10 9:07:08 AM kernel 00000069
8/5/10 9:07:08 AM kernel NVRM-RC: 0 4e565f504649464f5f494e54525f303a307831303030
8/5/10 9:07:08 AM kernel NVRM-RC: 0 4e565f504649464f5f4341434845315f444d415f53544154453a307838303030
8/5/10 9:07:08 AM kernel NVRM-RC: 1 343332630a


The MBP has two Graphics cards, I only use the 9400M (lesser one).

Also today MacRumors mentioned that supposedly Apple is working on a software update for this problem, I do hope so because it's a PITA.

I forgot to add that the problems were occurring way before I added more memory, I was actually hoping that by adding a little more memory the problem may not happen (although at that point I had begun to suspect it's a graphics card issue).

Message was edited by: Sam Rowlands

Let's hope that this Snow Leopard Graphics Update that is currently doing the rounds in developer circles will fix the freezing issues:

Link: http://forums.macrumors.com/showthread.php?t=985610

Is someone from Apple working on this. Apple phone support was not really helpfull (if you can talk to them).
Mac OSX 10.6 with all teh latest updates still has this problem occasionally. the scren flickers and the only thing you can do is turn the machine off ....

this is not good ....

Check your updates. Snow Leopard Graphics Update 1.0 is here.