Nvidia video driver crash

[satrow]

As a beta tester and user of PM who doesn't have a recent nVidia card, can I ask you for more details of the problems you encountered please? I was not aware of any such crashes.

[vrapp2]

Hmm. I wonder how you expect me to reply if you ban me from posting and delete original post.

[vrapp2]

Assuming this was some mistake, here's the log of the tech support incident with nvidia, attached. Read from the bottom.

[Moonchild]

Hmm. I wonder how you expect me to reply if you ban me from posting and delete original post.

Sorry, but not selecting a sane answer to the "sanity check" question, when you register, triggers a flag and a very likely cleanup of your user account into the "banned" group.
"I am:
A spammer"
was your answer.

This time around you seem to have selected the proper response so you should be good to go. Sorry if a legit post got nuked because of it.

[satrow]

My software doesn't show about the last 6 characters on each line so reading it is quite difficult, do you have the dmp files that you could upload to an accessible filehosting site so that I can take a look at them please?

[megaman]

My software doesn't show about the last 6 characters on each line so reading it is quite difficult, do you have the dmp files that you could upload to an accessible filehosting site so that I can take a look at them please?

Same thing happened to me, so I took it out on the document.

[satrow]

My software doesn't show about the last 6 characters on each line so reading it is quite difficult, do you have the dmp files that you could upload to an accessible filehosting site so that I can take a look at them please?

Same thing happened to me, so I took it out on the document.

Ah, so the dmp link is live? Ok, it looked like the conversation logged might have been in a non-public area so i didn't try it - thanks

[vrapp2]

My software doesn't show about the last 6 characters on each line so reading it is quite difficult, do you have the dmp files that you could upload to an accessible filehosting site so that I can take a look at them please?

Here are the two dmp files mentioned in the document:

http://www.mediafire.com/?6f9bbdcelt00tkm

[vrapp2]

Hmm. I wonder how you expect me to reply if you ban me from posting and delete original post.

Sorry, but not selecting a sane answer to the "sanity check" question, when you register, triggers a flag and a very likely cleanup of your user account into the "banned" group.
"I am:
A spammer"
was your answer.

This time around you seem to have selected the proper response so you should be good to go. Sorry if a legit post got nuked because of it.

I'd think, if "sanity check" is form of captcha, selecting "wrong" answer during the registration should probably have blocked the registration; to allow to proceed to register, then accept email verification, then allow to post, and then ban was somewhat unexpected.

[vrapp2]

Here's the copy of the transcript with Nvidia.

==================
Communication History
Response Ray via Email 01/31/2012 07:51 AM
Hello,

Can you please try the latest 295.51 beta driver that was released this morning to see if that helps? We fixed a similar problem in the 295.51 beta driver and that should also help resolve your issue. You can download the driver here: http://www.geforce.com/Drivers/Results/41051.


Best Regards
NVIDIA Customer Care


Customer Vadim Rapp via CSS Web 12/19/2011 10:25 AM
I also noticed another issue, unrelated. I will post it here, but if you prefer it in another incident, let me know.

In CCC, I run display optimization wizard. There are two issues:
1. on the page "Color temperature" it does not show any pattern, only instructional text. This may be because I have only one display, but I think it still should allow to set the color temperature "to the value that most accurately represents white to you", as it says.

2. The gamma settings set by the wizard don't persist. I set them in the wizard as instructed, but if I re-run the wizard, I find this page in reset state, rather than what I set before.
Customer Vadim Rapp via CSS Web 12/19/2011 08:17 AM
...still the question shouldn't be closed, as it did.
Customer Vadim Rapp via CSS Web 12/19/2011 08:16 AM
I confirm that without running Palemoon, version 290.36 is stable, no blue screen in several days


Response Ray via Email 12/15/2011 10:15 AM
Just a quick update. The lab is now running systems with Pale Moon installed to see if that may help trigger the failure. Our engineering team have analyzed the minidump files you provided and the stacks ares exactly the same as a known bug that we already fixed. Engineering believes that the problem are likely related since the stacks are identical to the other bug. The other bug is already fixed so as soon as we can replicate the failure then we can verify a potential fix. The fix is currently not available in any public drivers yet and will not be available until the 295.xx release, which is targeted for early January, beta may come sooner. If we are still unable to replicate the failure then we'll have to wait for the 295.xx driver release so you can verify on your system. Engineer assigned tot he bug is pretty confident that this is the same bug and will be fixed with the 295.xx release. I'll notify you as soon as the 295.xx driver is available.


-Ray

Customer Vadim Rapp via CSS Web 12/14/2011 10:38 AM
Yes, palemoon is indeed installed and is heavily used. In essence, this is Firefox recompiled with allowed optimizations for the modern processors. Because of that it runs noticeably faster and won't run on older hardware. I will now try to use standard Firefox and see if the problem goes away.

Response Ray via Email 12/14/2011 10:30 AM
I just got an update from another engineer who is analyzing the minidump file you provided. The minidump file points to the palemoon.exe as the cause of the fault in the driver. This appears to be an open source based project [http://www.palemoon.org/], based on the firefox browser (if our search is correctly). Are you familiar with this Pale Moon browser? Can you check you programs to see if this is indeed installed and then try uninstall it to see if that helps. This is not a fix but more a test to confirm if this is indeed related to the Pale Moon browser. We will also install this to our lab system to see if that will help trigger the failure.


-Ray
Response Ray via Email 12/14/2011 08:41 AM
Thanks for the file. Yes, the lab tried manually moving mouse and keyboard per your instructions for over 30 minutes without any problem. They then allow the systems to run automated tests. I will forward to engineering so they can review. We don't know for sure if this SMS Mirror Driver has anything to do with the problem, and since this does work with previous drivers it's definitely points to the new driver.


-Ray
Customer Vadim Rapp via CSS Web 12/14/2011 08:25 AM
here's the monitor infor attached.

SMS driver is indeed there, but disabled in device manager. I will however try to uninstall it completely and try again. Though, old driver works ok even with SMS driver as well.

When you tried and it was running for days - I assume you do have something that simulates user's mouse movements?
Response Ray via Email 12/14/2011 08:20 AM
Hello,

Our lab tried multiple motherboards with GeForce 6100 and so far all systems have been running for days without any such failure. I suspect the problem may be specific to your setup and not a general driver issue. Unfortunately we don't have this Abit NF-95 in our lab to confirm. Our engineering team review the system information log and noticed the system is using the Microsoft SMS Mirror Driver, not sure if this has anything to do with it but they would like more detail information of your display configuration including the monitor EDID. You list a Samsung MY23WS monitor connected via the VGA connection, is this the only monitor attached? Can you follow the steps below to extract the monitor EDID so we can review.


To get the EDID, please do the following:

1) Download the program here: http://www.entechtaiwan.com/util/moninfo.shtm
2) Install and run the program
3) Click on File->Safe As
4) Safe it as a *.Bin file and attach it to your case.


-Ray


Customer Vadim Rapp via CSS Web 12/05/2011 10:13 AM
The motherboard is Abit NF-95.
Response Ray via Email 12/05/2011 09:06 AM
Thanks for the update. I've submited a bug to our quality lab and see if they can replicate the failure. Based on the deviceID this appears to be an integrated GeForce 6100 graphics card. Can you tell me the brand/model of the motherboard you have? We may need to match the same motherboard if this is indeed integrated graphics card. If we can replicate the failure in-house then we can escalate to engineering to investigate.


-Ray


Customer Vadim Rapp via CSS Web 12/02/2011 09:41 AM
OK, I confirm that with the beta driver it's the same. Actually now I recall that I already had these crashes not less than one year ago, with then-current driver, that's why I always returned to this very old version 6.14.11.9107.

I also upload new msinfo32-2.txt taken with beta driver installed, and new minidump.

In the instant before the crash, I was moving the mouse while working with Firefox window; I noticed that in that instant the mouse was moving slower than usual, as if it was experiencing some obstacles/delays on its way. In the next moment blue screen came up.


Customer Vadim Rapp via CSS Web 12/01/2011 11:17 AM
1. yes, with 6.14.11.9107 the computer was stable for weeks without restart.
2. Samsung MY23WS; 2048 x 1152; VGA connector.
3. default
4. unfortunately, nothing in particular. It already crashed 3 times, every time it was just another mouse movement that directly preceded it.
5. attached (at this point, I have rolled back to the old driver 6.14.11.9107 )
6. will try and post result after 1-2 days.


Response Ray via Email 12/01/2011 11:09 AM
Hello,

Thank you for your bug submission. Is the crash random or is there a sequence of steps to reliably trigger the crash on-demand? The crash dump does point to one of the graphics drivers component causing the crash. We'll need to replciate the failure in our lab so our engineering team can debug the problem. Can you provide the information below so we can investigate?

1. Had this worked with previous drivers? If yes what version?
2. Type of monitor attached and desktop resolution. Please also indicate the type of connection used to connect the monitor (DVI, VGA. HDMI, converter)
3. Any specific NVIDIA Control Panel setting, if using default then please note.
4. Step by steps sequences to reliably trigger this problem. We will have the lab try to follow the exact steps.
5. Complete system information by running "msinfo32.exe" to export to text file.
6. Verify that problem still occurs with the latest 290.36 beta driver: http://www.geforce.com/Drivers/Results/40024


Best regards,
NVIDIA Customer Care
Customer Vadim Rapp via CSS Web 11/30/2011 10:48 AM
I had old version of driver and it was stable for weeks without reboot. Yesterday upgraded to the current driver. Now received blue screen while working with Windows desktop, no games or 3D, just clicked a mouse on another window. Crash dump is attached.

[satrow]

Firstly, I'm no expert at debugging BSOD's.

Code: Select all

*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

Use !analyze -v to get detailed debugging information.

BugCheck 1000008E, {c0000005, bd08c7e2, b162253c, 0}

Probably caused by : nv4_disp.dll ( nv4_disp+7a7e2 )

Followup: MachineOwner
---------

0: kd> !analyze -v
*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

KERNEL_MODE_EXCEPTION_NOT_HANDLED_M (1000008e)
This is a very common bugcheck.  Usually the exception address pinpoints
the driver/function that caused the problem.  Always note this address
as well as the link date of the driver/image that contains this address.
Some common problems are exception code 0x80000003.  This means a hard
coded breakpoint or assertion was hit, but this system was booted
/NODEBUG.  This is not supposed to happen as developers should never have
hardcoded breakpoints in retail code, but ...
If this happens, make sure a debugger gets connected, and the
system is booted /DEBUG.  This will let us see why this breakpoint is
happening.
Arguments:
Arg1: [b]c0000005[/b], The exception code that was not handled
Arg2: bd08c7e2, The address that the exception occurred at
Arg3: b162253c, Trap Frame
Arg4: 00000000

Debugging Details:
------------------


EXCEPTION_CODE: (NTSTATUS) [b]0xc0000005[/b] - The instruction at 0x%08lx referenced memory at 0x%08lx. The memory could not be %s.

FAULTING_IP: 
nv4_disp+7a7e2
bd08c7e2 8b4108          mov     eax,dword ptr [ecx+8]

TRAP_FRAME:  b162253c -- (.trap 0xffffffffb162253c)
ErrCode = 00000000
eax=e554b800 ebx=e554ec30 ecx=00000000 edx=e2651010 esi=e554b800 edi=e554ec24
eip=bd08c7e2 esp=b16225b0 ebp=00000003 iopl=0         nv up ei pl nz na pe nc
cs=0008  ss=0010  ds=0023  es=0023  fs=0030  gs=0000             efl=00010206
nv4_disp+0x7a7e2:
bd08c7e2 8b4108          mov     eax,dword ptr [ecx+8] ds:0023:00000008=????????
Resetting default scope

CUSTOMER_CRASH_COUNT:  1

DEFAULT_BUCKET_ID:  DRIVER_FAULT

BUGCHECK_STR:  0x8E

PROCESS_NAME:  palemoon.exe

LAST_CONTROL_TRANSFER:  from e5551b40 to bd08c7e2

STACK_TEXT:  
WARNING: Stack unwind information not available. Following frames may be wrong.
b16225ac e5551b40 e554de00 e606d640 00000000 nv4_disp+0x7a7e2
b16225b0 e554de00 e606d640 00000000 004ab700 0xe5551b40
b16225b4 e606d640 00000000 004ab700 bd0df360 0xe554de00
b16225b8 00000000 004ab700 bd0df360 e554de00 0xe606d640


STACK_COMMAND:  kb

FOLLOWUP_IP: 
nv4_disp+7a7e2
bd08c7e2 8b4108          mov     eax,dword ptr [ecx+8]

SYMBOL_STACK_INDEX:  0

SYMBOL_NAME:  nv4_disp+7a7e2

FOLLOWUP_NAME:  MachineOwner

MODULE_NAME: nv4_disp

IMAGE_NAME:  nv4_disp.dll

DEBUG_FLR_IMAGE_TIMESTAMP:  4ecd9cd0

FAILURE_BUCKET_ID:  0x8E_nv4_disp+7a7e2

BUCKET_ID:  0x8E_nv4_disp+7a7e2

Followup: MachineOwner

The above BSOD shows that Pale Moon was running (on top?) during the crash and the crash was caused by (happened with) the nVidia display driver.
The c0000005 in the first argument above (and below), I think, indicates that the problem is memory or drivers.
The below BSOD is almost identical, indicating a high level of consistency and therefore accuracy (unlike random BSOD's).

Code: Select all

*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

Use !analyze -v to get detailed debugging information.

BugCheck 1000008E, {c0000005, bd08b672, 8a89a53c, 0}

Probably caused by : nv4_disp.dll ( nv4_disp+79672 )

Followup: MachineOwner
---------

0: kd> !analyze -v
*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

KERNEL_MODE_EXCEPTION_NOT_HANDLED_M (1000008e)
This is a very common bugcheck.  Usually the exception address pinpoints
the driver/function that caused the problem.  Always note this address
as well as the link date of the driver/image that contains this address.
Some common problems are exception code 0x80000003.  This means a hard
coded breakpoint or assertion was hit, but this system was booted
/NODEBUG.  This is not supposed to happen as developers should never have
hardcoded breakpoints in retail code, but ...
If this happens, make sure a debugger gets connected, and the
system is booted /DEBUG.  This will let us see why this breakpoint is
happening.
Arguments:
Arg1: c0000005, The exception code that was not handled
Arg2: bd08b672, The address that the exception occurred at
Arg3: 8a89a53c, Trap Frame
Arg4: 00000000

Debugging Details:
------------------


EXCEPTION_CODE: (NTSTATUS) 0xc0000005 - The instruction at 0x%08lx referenced memory at 0x%08lx. The memory could not be %s.

FAULTING_IP: 
nv4_disp+79672
bd08b672 8b4108          mov     eax,dword ptr [ecx+8]

TRAP_FRAME:  8a89a53c -- (.trap 0xffffffff8a89a53c)
ErrCode = 00000000
eax=e5bef800 ebx=e5bf28a8 ecx=00000000 edx=e261b010 esi=e5bef800 edi=e5bf28a4
eip=bd08b672 esp=8a89a5b0 ebp=00000001 iopl=0         nv up ei pl nz na po nc
cs=0008  ss=0010  ds=0023  es=0023  fs=0030  gs=0000             efl=00010202
nv4_disp+0x79672:
bd08b672 8b4108          mov     eax,dword ptr [ecx+8] ds:0023:00000008=????????
Resetting default scope

CUSTOMER_CRASH_COUNT:  1

DEFAULT_BUCKET_ID:  DRIVER_FAULT

BUGCHECK_STR:  0x8E

PROCESS_NAME:  palemoon.exe

LAST_CONTROL_TRANSFER:  from e5bf57c0 to bd08b672

STACK_TEXT:  
WARNING: Stack unwind information not available. Following frames may be wrong.
8a89a5ac e5bf57c0 e5bf1a80 e61ecd40 00000000 nv4_disp+0x79672
8a89a5b0 e5bf1a80 e61ecd40 00000000 00062400 0xe5bf57c0
8a89a5b4 e61ecd40 00000000 00062400 bd0ddaa0 0xe5bf1a80
8a89a5b8 00000000 00062400 bd0ddaa0 e5bf1a80 0xe61ecd40


STACK_COMMAND:  kb

FOLLOWUP_IP: 
nv4_disp+79672
bd08b672 8b4108          mov     eax,dword ptr [ecx+8]

SYMBOL_STACK_INDEX:  0

SYMBOL_NAME:  nv4_disp+79672

FOLLOWUP_NAME:  MachineOwner

MODULE_NAME: nv4_disp

IMAGE_NAME:  nv4_disp.dll

DEBUG_FLR_IMAGE_TIMESTAMP:  4e8fb4b1

FAILURE_BUCKET_ID:  0x8E_nv4_disp+79672

BUCKET_ID:  0x8E_nv4_disp+79672

Followup: MachineOwner

My usual advice for this would be:

To try for a quick 'fix', install a different set of nVidia drivers (the latest version?) and test, repeat with different versions if the BSOD persists. It could be that it is the memory on the graphics card that is faulty, test with a known good card.

For a fuller test/diagnosis, test the memory with Memtest86+ for at least 7 passes and update all drivers from the chip makers - use MSFT/Windows Update as a last resort as they are often months older. * You do have a lot of older drivers and some that are rather unusual!

[Moonchild]

I'd think, if "sanity check" is form of captcha, selecting "wrong" answer during the registration should probably have blocked the registration; to allow to proceed to register, then accept email verification, then allow to post, and then ban was somewhat unexpected.

It is not a form of registration-time captcha, it is a sanity check - a quick verification for administrators and moderators with banning powers alike to verify flagged accounts and take swift action. Yes, that means real people check and moderate this board *gasp* it still exists!
Nobody else who registers has a problem with this and nobody else who was a real person failed to select a proper response so far. Bottom line: read carefully when you register, and pay attention.

[Moonchild]

The above BSOD shows that Pale Moon was running (on top?) during the crash and the crash was caused by (happened with) the nVidia display driver.
The c0000005 in the first argument above (and below), I think, indicates that the problem is memory or drivers.

Close
Both dumps clearly show the owning process being palemoon.exe (so it happened because of something triggered by Pale Moon)
Both dumps also show the actual fault to be in nv4_disp, the NVidia display driver.
The faulting instruction is a load instruction for a memory address, and the error c0000005 means "access violation".

Conclusion: The NVidia driver, as a result of a command issued by Pale Moon, tries to access a memory address which it is not allowed to, leaving the system in an unrecoverable state (since the display driver is essential for Windows to function) ==> BSOD
Pale Moon may have triggered it, but NVidia's driver is the one at fault.

[vrapp2]

I think little clarification is in order: the post was not in order to point a finger a Palemoon or to suggest that it somehow needs to be fixed; there's no argument that the bug was indeed in Nvidia driver, which Nvidia has admitted and already fixed, so I don't quite understand the point of analyzing the dump. I posted this information only in order to let other users know that if anybody else was experiencing this problem, they now cold get rid of it by upgrading Nvidia driver.