random resets, no recognition of system not being shutdown correct

  • Thread starter Thread starter moo7-64
  • Start date Start date
M

moo7-64

Random system resets - sometimes when idle.

When XP (pro, sp2) boots back on, it never recognises the system wasn't
shutdown correctly, & doesn't ask to run the disk check or if I want safe
mode.

Event log shows errors (which occur sometimes prior to resets) with not
being able to start a service NVR0FLASHDev - there's entries for this in the
registry in 'legacy drivers' & it appears to be a common issue with older
Nvidia drivers. However I don't want to just delete the entries as I'm not
entirely sure the service not starting is causing the reboots.

I have the error reporting turned on, and it's never shown any problems the
whole time the systems been up (I had a blue screen recently, which was
linked to running Server 2003 in Virtual PC, and was to do with a mouse
buffer overflow, which I hopefully sorted).

The random reset occurred at least once running a linux OS too - however I
think that it was recognised that the system hadn't been shutdown properly
when it restarted; I can't remember exactly.

So, it seems a hardware issue - but I'd like to know if anyone has had a
similar problem, and knows why it isn't being logged as an error, or why it's
not seeing that it wasn't shutdown right; as this might help identify what
the actual issue is.

As far as things look, my hardware is ok....nothing is running too hot, and
there's plenty of memory and nothing is being used to its limits (it's a
4-core with 4 gig, has a temperature warning on in BIOS).
I'm a bit suspicious that these non-ECC boards and RAM, however nice, are
not as stable as they should be.......but there's no other way to build your
own rig! as all but the most expensive of server boards and memory are all
non-ECC overclocker kits these days.
The monitor has in-built speakers which have caused weird freezes when
playing back certain audio clips - some frequencies get 'stuck' and just keep
playing over and over until the system is reset, and the display freezes too.
So it could be related to that....but I can't tell as there's no error logs,
and nothing obvious; it's really random.

I have also had hangups (freezes), and random resets, when running virtual
machines in Virtual Box too; but that could be anything / is difficult to
troubleshoot, given it's probably not officially supported plus I don't know
enough about what conflicts may occur between resource sharing - just adding
this in in case it is useful.
(also, on Vista basic some virtual machines in Virtual Box cause EPP errors,
and go into the abort state - so there does seem to be some protection-error
related issues).
 
No. Given what I wrote it should be obvious that if I was then I'd be aware
it can make a system unstable. But do explain what it was I wrote that made
you think otherwise.
 
True. Or dust can cause it.

I did find out that due to ACPI I have very stupid IRQ assignments - two
major devices are sharing the same IRQ (onboard LAN and graphics card (x16
PCI-E)). The BIOS doesn't allow IRQ assignments! I had no idea things had
gotten that bad, now you can't even set your own IRQs; there may be another
way to do so - without rewriting the BIOS I mean, but it'd seem that all the
OSs I use have the same IRQ assignments, Windows or otherwise.

It is probably that, as the resets began after I got broadband (there were a
couple prior to that), so previously I wasn't using the onboard NIC. I could
add a PCI NIC, but then my case gets more crowded and it blocks the airflow
between the PSU fan and the GPU fan...

any info on changing IRQs in those circumstances would be appreciated.

It wouldn't surprise me if the RAM did have something not quite right - but
more likely just because it has no error-checking; though really a memory
space conflict should not be able to crash an OS or reset it. The RAM makers
sell it as having timings different to its actual timings. I called them
about it and they didn't actually know that not all BIOS allow timings to be
altered manually!
But at least they do lifetime guarantees, so if it ever does fail at least I
should be able to get a new set off them.

It's still worrying though, why any kind of shutdown or reset is not logged.
It should say that. Again, there'll likely be some very low-tech way to dig
out that info, some memory dump saved somewhere, that I'm not aware of.
 
Memory errors can cause random restarts especially
when near the max overclock speed of the system.
 
Yar, I'm well aware of that. If I had oc'ed them then I'd suspect that was
the problem.

Just for completeness' sake, memory diagnostic isn't finding any errors, and
the crashes are now displaying bluescreens - it's coming up as an
irql_not_less_or_equal.
So it's most likely a driver or other software issue; however as mentioned
below earlier - the fact ACPI assigned the same irq to two heavily used main
devices is incredibly stupid and bound to cause problems. I still haven't
looked up thoroughly on how to change them manually when BIOS doesn't allow
it.

Again though, it shouldn't matter if there is a memory space conflict, as it
shouldn't crash all the threads running even if it is in kernel mode! The OS
should have rock-solid threads & processes that keep the system up, and any
attempts to access space wrongly should result only in the requesting program
being shutdown. Of course if the graphics drivers are buggy then that's
difficult but they could always revert to a 'safe' mode display.
As far as anything running that has to be up to keep core services going, if
that is causing memory access issues then! - should have dedicated space
assigned maybe.

Could be that it's mostly 7 that's got the issues, it's a pre-release after
all. Other than that this seems to be a thing that'll happen if you use a lot
of media-intensive websites; the amount of things I've had to download and
install for browsers just to view some video. Back in the Netscape 4 days if
you opened that along with the IE of the time, they'd both cause visual
corruptions in one another. Try installing different AVs and they tell you to
delete the others they find whether you are running them or not. So unless
you don't want to use your machine in a client-y way ever, it can be
difficult to tell which of the many potential culprits known to cause
problems is actually the one causing it.

So it's likely a case of - either have no GUI at all and run only core
services, or use it for media-intensive stuff with the understanding that it
may crash sometimes. & if you video-edit for a living, then buy some
10-grand+ proprietary system.
 
Back
Top