SOLVED BSOD New Variation

[Fixer1234]

I've encountered an apparent new variation of the unresponsive black screen after sleep issue. Mine started several weeks ago (late July 2012). I've reviewed hundred posts and tried dozens of fixes without success.

Problem: Sleep mode always leads to crash on awakening. However, the system also similarly crashes after idle most of the time even though sleep mode is disabled, and it sometimes crashes in a similar manner during active use.

Symptoms:

• video actually shuts down (monitor goes into the analog/digital scan when the signal disappears and then goes to sleep)--no display at all.
• Computer power light remains on (sometimes solid, sometimes flashing; flashing seems to be associated with sleep mode even though disabled--crash after idle sometimes has a flashing power light; crash during use has a solid power light).
• The fans continue to run and some device LED indicators remain on. The hard disk continues to spin but there is no access activity. Attached USB peripherals remain on and do not indicate a loss of connection.
• The computer is unresponsive to keyboard or mouse.
• It is also unresponsive to the power button. The only way to restart is to pull the power cord, wait about 30 seconds and plug it back in. (Plugging the power cord back in after less than about 30 seconds leaves the computer powered off and unresponsive.)
• The computer then powers up without pressing the power button and goes to the safe mode option screen.
• Logs indicate that the computer shut down unexpectedly, which could be from pulling the power cord. The crash doesn't leave diagnostics that I can find.
• Time to crash varies. If I restart after about 30 seconds, the computer occasionally crashes again within minutes of boot up. If the computer sits powered off for 5 or 10 minutes, it usually runs for anywhere from several hours to a day before crashing again.

Remedies tried so far:

• All hardware tests show normal.
• Temperature tests show everything within normal ranges; in fact, crashes occur when temperatures are well below limits (using SpeedFan to monitor).
• Voltages are stable (at least up until the moment of the crash).
• I've swapped the positions of the RAM modules with no effect.
• All sleep/hibernate options have been disabled, including those for individual devices. The attached Speccy file indicates that these are enabled even though the control panel shows them as disabled (could this be related to the issue?)
• The computer ran for almost 2 days in safe mode without incident (although there isn't much activity you can do in safe mode).
• The crashing occurs even when MSCONFIG is used for clean boots. The difference between safe mode and clean boot may indicate where the problem lies.
• The registry has been cleaned (no effect).
• All drivers and AMD Catalyst Control Center were updated (no effect).
• SFC returns no errors.
• Various recommended Microsoft patches have been installed (no effect).
• There are no viruses or malware.
• Disabled Norton Security Suite as a test (no effect).
• The computer is OEM OTS, with no esoteric hardware and no gaming hardware or software. Plenty of RAM and hard disk space.
• I've run so many tests and tried so many things that I'm probably leaving some out.

It's possible that in cleaning up the system, I disabled a critical process or service, although the crashes sometimes occur in the middle of activity that does not involve the system looking for services (unless some background activity is the cause).

Anyone have any insight or suggestions? Thanks.

 

[Shintaro]

Welcome to the w7 forums.

I hope that we can help.

That is a pretty exhaustive list of things that you have tried.

I would suggest that you remove CCC (Catalyst Control Center) it is known for causing problems.

Do you have a CAPS lock light on the keyboard? When the system crashes, can you press the CAPS lock and make the light go off and on?

If you have another machine, can you ping the crashed machine?

Let us know how you go.

 

[Fixer1234]

Thanks. A couple of good thoughts.

Caps Lock didn't activate the LED. The keyboard has a number pad and the Num Lock LED is normally on. It remains on during sleep mode and the crash, going out only during reboot. Looks like power is still going to the keyboard during the crash but the logic doesn't work.

Pinging resulted in a message that the destination host was unreachable. My understanding is that there is no response to a ping if the computer is asleep or hibernating. If that is incorrect, the response could mean that the computer crashes going into sleep mode rather than on coming out.

 

[Shintaro]

Is there a file called "hiberfil.sys" in the root of C:\ ?
If there is that would mean that hibernation is still enabled. And can be turned off from a command prompt with powercfg -H off

 

[Shintaro]

What brand and model is your network card?
What version of driver is installed for it?

 

[Fixer1234]

Found hiberfil.sys as a hidden system file. It looks like that file will exist whether hibernation is enabled or not. Powercfg appears to be just a command line interface for the same settings as the power options in the control panel.

Network Card: Marvell Yukon 88E8071 PCI-E Gigabit Ethernet Controller
Driver: yk62x64.sys Ver 11.22.3.9. It's dated 9/28/09, which checks out as the most recent version.

 

[Shintaro]

No. If you run the command. Then reboot the file should be gone. If the file is Giga bytes in size then you did not not have hibernation disabled.

 

[Fixer1234]

OK, some "interesting" results. I've attached some reports from POWERCFG.

The reports showed hibernation disabled but I went ahead and disabled it through POWERCFG. The hiberfil.sys was, indeed, a bunch of GB. The computer crashed again after using POWERCFG (could be just coincidence). When I rebooted, hiberfil.sys was gone. I suspect its existence was just an artifact of the problem, not the cause. Half an hour later, the system crashed again so that didn't fix the problem.

No minidump files are being created so it looks like either Windows doesn't see a problem before the crash (ie, crash caused by pulling the power plug on the unresponsive system or Windows getting lost in its underwear), or the crash is outside of Windows (ie, hardware-related).

The system doesn't reboot for close to a minute after the crash (fans come on but the power light remains off and unresponsive). Not clear whether this could be related to the cause or just an artifact but the time would seem to indicate either a component overheating or the need for capacitors to discharge. I couldn't find any evidence of overheating.

I added some notes to the POWERCFG reports. The setting for shutting off the monitor (10 minutes), doesn't match what shows in the control panel (never),or in the Speccy report (20 minutes). How do we get three different reports of the same setting? There are also routes to sleep mode if there is corruption and the computer is acting like it's a laptop.

The problem presents like a sleep mode error but that doesn't explain why it becomes unresponsive to keyboard, mouse, and power button.

Thanks for hanging in there on this problem.

 

[Shintaro]

Nothing attached???

 

[Shintaro]

Mate,
After so many times of "pulling the plug" have you run chkdsk /r ?
It is not the solution but your drive is having a rough time.

 

[Fixer1234]

Sorry, I added the files to the original post to keep all of the attachments together (power settings.zip).

I run chkdsk regularly--so far it's come up clean. Pretty soon I'll need to run it on what's left of my brain.

 

[Shintaro]

Lol, yes it is a very frustrating problem.

What is interesting is that in "available sleep state.txt" it says "The system firmware does not support this standby state."

Have you looked in your BIOS? Turned all hibernation / sleep off?

 

[Fixer1234]

Looks like S3 and Hybrid Hibernation are the two states in play. I've attempted to turn off all sleep/hibernation but the various reports show exceptions. And if the problem involves going into sleep mode, it means that there is a second problem of the computer becoming comatose (unresponsive to everything) in that mode. I would think that going into sleep mode would leave some kind of trail so at least it could be verified that that is what's going on, but I don't know what to look for.

I'm not aware of anything that could have corrupted the BIOS. If there is now some incompatibility, I'd be more inclined to remove the program or update that conflicts. I'm not a big fan of updating the BIOS.

 

[Shintaro]

No, No, What I mean, is there a setting in your BIOS that relates to hibernation?

 

[Fixer1234]

I have no idea. How do I read it?

 

[Shintaro]

When your computer starts(Before Windows starts), it might be F2 to get to the BIOS. Do you have the manual for the computer from Gateway?
Or at least know the update site for your model?

 

[Fixer1234]

I'm getting an education.

There are no apparent settings for controlling sleep/hibernation but there are settings related to it:
* power on via keyboard and mouse are enabled
* power on via RTC Alarm, PCIE Devices, and PCI Devices are all disabled (I don't know what these are)
* Restore on AC Power Loss is set to Last State, which might explain why the computer immediately powers up without hitting the power button after pulling the plug.
* AMD Cool 'n' Quiet is enabled. I'm guessing these are the power saving features designed to save the planet. I've seen other posts that suggest disabling this. I'll need to do some research into exactly what this controls.

 

[Fixer1234]

Looks like Cool 'n' Quiet fine tunes CPU operation and temperature control and doesn't relate to sleep mode.

 

[Shintaro]

As far as I know Cool and Quiet powers down the CPU to keep the CPU temp down as well as stopping the fan noise.

Maybe turn that off.

 

[Fixer1234]

I went through six days of Windows log data and found some things that illuminate the problem. There were three overlapping patterns of system behavior:

1. On one occassion, the computer went into and out of sleep mode normally, on its own, even though I've done everything possible to disable sleep mode. As noted in an earlier post, there are discrepancies in how the sleep mode settings are reported, and routes to sleep mode if something is corrupted. Event details:

Log entries show the system preparing for sleep mode, triggering sleep mode based on system idle, and then awakening from sleep mode triggered by the power switch. System behavior during sleep mode was normal (all internal devices off and the power light blinking, monitor asleep with no input signal).

2. If I manually trigger sleep mode through the shut down menu, this always happens:

The system enters sleep mode but completely freezes on awakening. During sleep mode, all internal devices are off, the power light blinks, and the monitor is asleep with no signal. Triggering wake-up via mouse, keyboard, or power button starts the process. The fans come on, the hard disk searches, then the optical drive searches (second boot device on the list; indicates the system isn't finding what it's looking for on the hard disk). The monitor comes out of sleep mode but has no input signal. Then the system totally freezes with the fans and LEDs on, the hard disk spinning, the power light flashing, and the monitor goes back to sleep. The system is unresponsive to input from the mouse, keyboard, and power button, and the power button will not shut the system down. I have to pull the plug to shut down.

When the system is subsequently powered up, the log shows activity up to the time I triggered sleep mode. The reboot event shows that the system was in sleep mode at the time it crashed. However, unlike the system-triggered sleep event, there are no log entries for sleep being initiated (normally logged before sleep) or wake-up initiated (normally logged after wake-up), and sometimes no error entry for the time at which the system shut down (normally logged after wake-up). Manually-triggered sleep is supposedly handled by the same process as system-triggered sleep (via kernel-power), but the event logging is obviously being handled differently.

3. The system crashes at random times, sometimes while idle, sometimes in the middle of input activity. The appearance is very similar to pattern #2. Differences:

The system does not shut down. The monitor loses signal and goes to sleep but everything else remains powered up. System fans and LEDs continue to be on and the hard disk continues to spin. The keyboard NumLock LED remains on but keyboard logic is off (eg, the Caps Lock button doesn't affect its LED). The system power light remains on (steady, not blinking). The system is unresponsive to input, including the power button; I have to pull the plug to shut down. This is all generally similar in appearance to manually-triggered sleep mode after it freezes on awakening.

In pattern #3, the log entries are generally similar to pattern #2. There are entries up until I pull the plug (if the system has been idle, there may be no entries recent to the freeze, which could also happen if there is a delay between freezing and pulling the plug; I don't know which event stops the logging). There are no entries about going into sleep or powering down. In this case, however, the reboot entry does not indicate that sleep was in progress. There is an error entry, posted after restarting, indicating that shut-down (when the plug was pulled) was unexpected.

In almost all cases of system freezing, there were no log entries of system or program activity just prior to freezing. It's possible that some common event causes the freeze but timing prevents the system from logging it.

There is an interesting exception. On two events (out of more than three dozen), there was an error report, code 6000 winlogon notification subscriber <...> unavailable, at approximately the time the plug was pulled. One subscriber was <Session Env>, the other was <Sens>. The first was a typical pattern #3 event. The second was a pattern #2 event and this was not the last log entry before pulling the plug. That would seem to indicate this this error is not the cause. It may just be some non-fatal corruption and coincidence.

The longest the system has gone without freezing is on the order of about 12 hours, even with a clean boot. However, it has never frozen in safe mode. I was able to leave it in safe mode for almost two days before I had to get some work done. I also tried simulating safe mode with a "super-clean" boot, enabling only those services that run in safe mode. That still froze just like a clean boot. Obviously, safe mode disables more than the user has access to with MSCONFIG. One obvious indicator of this is that sleep mode is an option in the shut down menu after a clean boot but is not in safe mode. It appears that the underlying problem is somewhere in the differences between safe mode and "safe mode lite" via super-clean boot.

There are two obvious errors on the system. One is the inability to completely disable sleep mode (it would be great of the system did not even recognize the option of sleep mode like in safe mode). The other is something missing or corrupted that causes the system to freeze. Whatever that is, it occurs randomly but is also reliably triggered by manually initiating sleep mode.

That's as far as I have been able to diagnose the problem with what I know.