Monitor Sleeps, then BSOD

Page 1 of 2 12 LastLast

  1. Posts : 13
    Windows 7 Home Premium 64-bit
       #1

    Monitor Sleeps, then BSOD


    Got a brand-new desktop, this has happened at least 4 times:

    During normal use, the monitor will fall asleep on its own. After waking up, BSOD happens involving nvlddmkm.sys and not being able to restart the display driver. I've disabled every sleep feature I could find (Sleep, Hibernate, Hybrid sleep) and updated all my drivers, but the problem persists.

    I later activated the driver verifier tool and while that was running, BSOD within seconds after log in, saying it has found a faulty driver.

    System specs as follows and minidumps attached:

    HP Pavilion p6210f
    Windows 7 Home Premium 64-bit
    AMD Athlon II x4 620
    6GB DDR2 RAM
    Nvidia GeForce 9100 Integrated Graphics
    HP 2009m 20" Monitor
      My Computer


  2. Posts : 1,377
    Win7x64
       #2

    Short version: try a downgrade of the driver package, just as a test. Make sure you're not overclocking anything. Forget about driver verifier - it can't help with your video BSOD.


    ==============================================
    Detail: A stop 0x116 is a very specific type of bugcheck (BSOD). The OS Timeout Detection and Recovery (TDR) component has given up on trying to get a response from the [driver+video_card] combination, and in desperation it has crashed the OS. Before it triggered the bugcheck, TDR first attempted to send multiple "hello, are you there?!?" messages to the video [driver+card] combo, all of which were met with silence, as well as an attempt to fully reinitialise the video subsystem from scratch. Everything failed. The [driver+card] combo was unresponsive. TDR gave up and made everything go blue.

    Driver verifier (DV) can't help in this instance, and besides you already know the name of the unresponsive driver:

    FAILURE_BUCKET_ID: X64_0x116_IMAGE_nvlddmkm.sys

    In a curious twist, DV can reveal problems in drivers which would otherwise potentially pass unnoticed, and that is exactly what is happening in your 0xC4 crashes. Depending on the amount of DV driver scrutiny specified, (all) drivers on the system are subjected to some truly savage ordeals, such as for example denying their memory requests or forcing out all of their data to the disk so it has to be paged in again... DV is primarily a tool for driver developers hunting down bugs in their code.

    On your box, DV found an apparent bug in an AVG filter driver:

    0: kd> k
    Child-SP RetAddr Call Site
    fffff880`08d205b8 fffff800`02f003dc nt!KeBugCheckEx
    fffff880`08d205c0 fffff800`02f00e1b nt!VerifierBugCheckIfAppropriate+0x3c
    fffff880`08d20600 fffff800`02f11df8 nt!ExAllocatePoolSanityChecks+0xcb
    fffff880`08d20640 fffff800`02f1209d nt!VeAllocatePoolWithTagPriority+0x88
    fffff880`08d206b0 fffff880`02c6f707 nt!VerifierExAllocatePoolWithTag+0x1d
    fffff880`08d206f0 fffff880`02cddb72 afd!WskTdiEHReceive+0xe7
    fffff880`08d20780 fffff980`2838aea0 avgtdia+0x7b72

    0: kd> lmvm avgtdia
    start end module name
    fffff880`02cd6000 fffff880`02d4b000 avgtdia T (no symbols)
    Loaded symbol image file: avgtdia.sys
    Image path: \SystemRoot\System32\Drivers\avgtdia.sys
    Image name: avgtdia.sys
    Timestamp: Thu Oct 15 03:37:34 2009 (4AD5FE4E)

    Not that it helps you any with your video timeout issues, but avgtdia.sys dated 15-Oct-2009 has been outed as a "pool corruptor". If you disable DV, you may never see a crash because of that problem, or it may silently corrupt your data, which is far, far worse.
      My Computer


  3. Posts : 13
    Windows 7 Home Premium 64-bit
    Thread Starter
       #3

    Thanks for the quick response. Do you know where I can get older drivers for this card? I've been to the nvidia site and turned up nothing. I'd search the internet more thoroughly but I have tons of school work to do.

    EDIT: I apparently didn't have the most updated driver for my 9100. I knew there were drivers more recent than the one I had installed, but the nvidia site did not list 9100 under supported products of their most recent driver, 191.07. But I downloaded and installed version 191.07 anyways, and so far so good. Thanks again for helping me pinpoint the problem.
    Last edited by supermatt; 02 Nov 2009 at 02:41.
      My Computer


  4. Posts : 13
    Windows 7 Home Premium 64-bit
    Thread Starter
       #4

    UPDATE 11/1/09: Ok, so the new drivers slowed down the crashing, but did not stop them. The computer still crashes about once every 36-48 hours. It's the same error and it occurs at the most random times; the most recent time I had a blank firefox window open downloading a demo for Need for Speed Shift, nothing graphically going on at all.

    Updated to the new beta drivers released recently and still the same rate of crashing. The computer shrugs off the crash as if nothing has happened, but it annoys the hell out of me. Any more suggestions besides messing around with the driver versions?
      My Computer


  5. Posts : 5,705
    Win7 x64 + x86
       #5

    Please continue to upload the memory dump files as they occur - sometimes they'll expose a pattern that leads to a solution.
      My Computer


  6. Posts : 13
    Windows 7 Home Premium 64-bit
    Thread Starter
       #6

    Here are 3 more minidumps that i got recently, all one after another. All I was doing attaching a certain mp3 file to an email using Gmail, and it gives me the crash. It crashed every time i tried to attach the mp3, about 5-10 seconds after starting the attach. I will test if it's just Gmail, and/or just that mp3 that's the problem.
      My Computer


  7. Posts : 13
    Windows 7 Home Premium 64-bit
    Thread Starter
       #7

    I did a more thorough analysis to pinpoint this problem. I attempted to attach an mp3 file to an email using various clients and browsers. I used Firefox 3.5.4, Internet Explorer 8, and Internet Explorer 8 64-bit for the browsers. I used Gmail and Yahoo Mail for the clients. I tried to attach 2 different mp3s, 1.mp3 and 2.mp3 (abbr. for convenience). 1.mp3 was the one that crashed in the previous post.

    For each time I got a BSOD, I gave it a number and the corresponding mini dump can be found in Minidump Analysis.zip. Here are my results:

    Using Firefox:
    --Using Gmail:
    ----1.mp3--FAILED (Minidump in previous post.)
    ----2.mp3--FAILED Crash 1
    --Using Yahoo:
    ----Both mp3's successfully attached and sent.
    Using IE8:
    --Gmail:
    ----1.mp3--FAILED Crash 2
    ----2.mp3--FAILED Crash 3
    --Yahoo:
    ----Both mp3's successfully attached and sent.
    Using IE8-64 bit:
    --Gmail**:
    ----1.mp3--FAILED Crash 4
    ----2.mp3--FAILED Crash 5
    --Yahoo:
    ----Both mp3's successfully attached and sent.

    ** When using Gmail in 64 bit IE8, the progress bar did not show--instead it said something like C:\fake path\1.mp3.

    Clearly something in Gmail is triggering this crash. Perhaps the minidumps will hold the key.

    EDIT: I varied the file size and type and got crashes only on large (~10MB) files of any type. Small files (~500KB) were fine.
    Last edited by supermatt; 03 Nov 2009 at 01:06.
      My Computer


  8. Posts : 1,377
    Win7x64
       #8

    supermatt said:
    ...Clearly something in Gmail is triggering this crash. Perhaps the minidumps will hold the key.
    I'm torn between feeling impressed and guilty. Impressed because you did all this methodical testing, and guilty in case something I said previously sent you down this path.

    Fact is, it's effectively impossible to gauge from a minidump why the nvidia driver+card combination is sometimes becoming unresponsive.

    The results of your exhaustive testing do indeed suggest that some type of graphics activity performed on the gmail site - but not when visiting yahoo - is triggering the problem, but unfortunately there is no practical solution in that direction.

    I examined all of the minidumps you attached, and they are all functionally identical.

    It sounds completely counter-intuitive, and I wouldn't blame you for being skeptical when you read this, but the browsers themselves and the websites you visit simply CANNOT be the root cause of this problem. That "privilege" is reserved for a kernel-mode driver, the BIOS, or a hardware problem in your video card. Gmail and Yahoo make somewhat different demands on the video subsystem (obviously - they look different). All those demands are legit, and on a healthy system all would be fulfilled without a hitch. For some reason that is deep in the bowels of the nvidia driver+card combo, on your machine some of those app demands cause the video subsystem to go into a tailspin and never recover.

    My advice is to update your BIOS, recheck again to make sure nothing is overclocked, reinstall the OS from scratch, test what happens with the in-built driver provided by Windows, and then the latest driver from nvidia. If you see the same symptoms again, return the hardware and demand a replacement.

    Given the machine is new, you've got a perfect right to expect it to function without bizarre low-level hardware-related crashes.
      My Computer


  9. Posts : 13
    Windows 7 Home Premium 64-bit
    Thread Starter
       #9

    Don't feel guilty at all, H2SO4, it's not what you said. It's mostly my insatiable desire to understand the world around us and likely a misunderstanding on my part of what minidumps are actually for. If I had a better understanding of them and how computers really work, I might not have done all that. But my inner engineer called for a methodical analysis, and I performed one and found a specific set of conditions that always causes the crash.

    I will follow your advice and update the BIOS and all that. You mentioned reinstalling the OS from scratch. I got the computer pre-built so I have no Windows 7 CD. Would a reinstall from the HP recovery partition suffice?
      My Computer


  10. Posts : 1,377
    Win7x64
       #10

    supermatt said:
    Don't feel guilty at all, H2SO4, it's not what you said. It's mostly my insatiable desire to understand the world around us and likely a misunderstanding on my part of what minidumps are actually for. If I had a better understanding of them and how computers really work, I might not have done all that. But my inner engineer called for a methodical analysis, and I performed one and found a specific set of conditions that always causes the crash.
    I salute your desire to understand the problem :)

    I'm hardly mother nature's gift to debugging, but I know enough to be confident that the answer is not in the minidumps, unfortunately.

    Think of the OS as one vast collection of workers, each doing one or more tasks while mostly oblivious to the others. Periodically, some of the workers send commands to each other or wait for shared resources to become available. In this instance, worker "C" long ago sent a command to the video subsystem, and worker "V" is responsible for nothing but periodically checking that the video is still alive and kicking. It's "V" which discovers the problem, simply by following its own monotonous "verify - rest a while - verify - rest..." pattern and suddenly noticing that the video is NOT responsive.

    A minidump is a tiny summary which in this case is like a recording of worker "V" going "WTF? This is very bad and unrecoverable. I am instructed to trigger a BSOD under these circumstances, so watch me..."

    There's no mention of what command "C" sent all those seconds ago (eons in OS terms), and even less info regarding the internal state of the video subsystem. It is possible to glean all that through additional debugging, but you really have to be an nvidia and/or MS driver developer with full physical access to the machine (not just minidumps), and obviously that's not practical for us mere mortals.

    supermatt said:
    I will follow your advice and update the BIOS and all that. You mentioned reinstalling the OS from scratch. I got the computer pre-built so I have no Windows 7 CD. Would a reinstall from the HP recovery partition suffice?
    Yes, it would suffice. If the reinstall doesn't fix the problem, then the main objective is to prove to HP that their hardware is somehow defective and in need of replacement under warranty. Since they provided you with the recovery partition image, the content presumably has to meet their standards for what it means to have a "clean and undefiled" install.
      My Computer


 
Page 1 of 2 12 LastLast

  Related Discussions
Our Sites
Site Links
About Us
Windows 7 Forums is an independent web site and has not been authorized, sponsored, or otherwise approved by Microsoft Corporation. "Windows 7" and related materials are trademarks of Microsoft Corp.

© Designer Media Ltd
All times are GMT -5. The time now is 13:17.
Find Us