Hardware Interrupts - How to determine source

[Trevil]

Greetings,

Running Win-XP-Pro w/all updates on HP-Xt958.

Brief version is I would like to know how I can determine who/what is driving the high number of hardware interrupts that I am seeing. Is there a monitor or log that can be used.

Longer version: Over a month ago I noticed when moving my mouse the cursor would freeze for .5 or 1 second, also WinAmp started acting if my music was 'scratched'. Did the usual (SpyBot, AdAware, etc.); used Task Manager with refresh set to fast, and noticed CPU was pulsing from 3% to 47% on each screen refresh. Found no obvious culprit, booted in Safe Mode, played around with stopping all programs, then slowly adding back and testing. Finally wound up with ALL programs from original config running, and problem *went away*.

Now a month later, problem came back. About two weeks ago I had download 'Process Explorer from Sysinternals' (find/pcworld.com/47596) which shows a lot more info than Task Manager. When I opened that today, it showed Hardware Interrupts were pulsing between 0 and 35% of my CPU. So I think that's where I would like to drill down for more info.

Of course, like when you take your car to the mechanic -- the problem has stopped for the moment!

Any help / suggestions would be appreciated.
Wayne

35+ years of 'progress' -- can't we all just go wire boards again?

 

[linney]

http://www.google.com/search?source...Hardware+Interrupts+-+How+to+determine+source

http://groups-beta.google.com/group...lr=&rls=GGLD,GGLD:2004-50,GGLD:en&sa=N&tab=wg

Error Message: An Interrupt Storm Has Caused the System to Hang

http://support.microsoft.com/default.aspx?scid=kb;en-us;290101&FR=1&PA=1&SD=HSCH

Have you tried your luck in the Hardware forum?

 

[Trevil]

Hi Linney,
Thank you for the suggestions, but the majority of the links I followed covered either courses taught somewhere, or internal hardware information. I had earlier done numerous searches of Google and Tek-tips (all forums), without an answer.
I was hoping there was some software product (or sample code) that would let me wrap my hands around the throat of just who originated the interrupt!
Is it acceptable to post this to the hardware forum, or is that considered "poor form"?
Thank you for your help!
Wayne

35+ years of 'progress' -- can't we all just go wire boards again?

 

[linney]

Of course it is acceptable to post in the Hardware forum, at least someone there may have come across your problem and can advise you better than I can.

It is still early days here so someone in this forum might offer some advise too.

Does MsInfo32 help you at all?

 

[Trevil]

I had checked MSINFO32, and under Hardware Resources there is a listing for IRS's, however it only shows which device is assigned to each channel. It would be GREAT if it would show the count by device.

Thank you for your suggestion. I will wait till later in the day to post in "Hardware".

35+ years of 'progress' -- can't we all just go wire boards again?

 

[Nelviticus]

This doesn't answer your question but it may solve your problem: is re-installing Windows an option for you?

I had a very similar problem when I first installed SP2. Despite extensive research and much uninstalling/reinstalling of apps and drivers I never did manage to track down what was causing it. Removing SP2 cured it but re-applying SP2 brought it back. In the end I started from scratch, re-installed Windows and applied SP2 and other OS updates before installing anything. That fixed the problem and it has never returned.

One other thing to try is to look at your services via Control Panel/Administrative Tools/Services - if any of them show up as "starting" in the Status column then that will give you a pointer for where to look next.

Nelviticus

 

[Trevil]

During the issue from a month ago, after trying everything and suspecting it may be some virus, I purchased a new 80gb hard drive and rebuilt entire system. I only installed essential programs that I had 100% faith in. Imagine my dismay when that didn't help!

I'll keep your suggestion about "starting tasks" in mind if this beast returns.

Thank you for your assistance.

35+ years of 'progress' -- can't we all just go wire boards again?

 

[bcastner]

As a guess, replace the network adapter or update the driver.

This sounds like a hot beaconing situation:

http://www.microsoft.com/whdc/device/network/NetAdapters-Drvs.mspx

My first guess: the reporting is wrong, and this is completely driver dependent. A common erronous report is that the NIC is creating billions of packets:

http://www.experts-exchange.com/Operating_Systems/WinXP/Q_20785766.html

Generally speaking, the reporting of statistics are left to the driver manufacturers. They do no always comply with the Windows XP DDK in all respects. Most reports such as yours are the result of unsiqned and non-WQL drivers that falsely report traffic, and are harmless.

 

[Trevil]

I have replaced network driver w/latest; checked the links suggested, but they all focus on network traffic.

I'm still looking at the monitor showing hardware interrupts are eating me alive, and wondering WHICH one(s) it could be.

If someone can't help me get to the root cause of this in a few days, I'll waste a few $$ to call Microsoft and see what they say.

Thank you for your assistance.
Wayne

35+ years of 'progress' -- can't we all just go wire boards again?

 

[bcastner]

Natively, it does not exist.

 

[bcastner]

If you are absolutely sure that the NIC is not "hot" due to a hardware failure, the next object to inspect is the keyboard. I admire your NIC driver upgrade, but that did not answer my original question as to whether the NIC was "hot" and creating by hardware unnecessary hardware IRQ requests.

For non-PIO and pure PIC activity:

NIC -- sampled roughly 3,000 / second
Keyboard -- sampled roughly 1,000 /second
USB devices -- samples roughly 300 times /second
Parallel device -- sampled roughly 800 times /second

You need to pull each device one-by-one to see the sampling drop.

And I think you under-appreciated my original comment. The reporting for overall IRQ activity is very much non-Microsoft and very much the author of device driver dependent. Translation: it means nothing.

Strip your machine of all devices and test by some same metric the results.

It would help in future replies if you distinguish between NMI interrupts and those handle by the BIOS in the IRQ jump table.

 

[Trevil]

Let's try to understand one another. If you are aware of some monitor or reporting tool that you would like to see the results from, please let me know what / where. When you say "It would help in future replies if you distinguish between NMI interrupts and those handle by the BIOS in the IRQ jump table" -- I don't have a clue where I can get either, but would be happy to try.

As mentioned originally, I am using a product named 'Process Explorer' from Sysinternals. This tool provides more information than Task Manager and this is where I see the category of 'Hardware Interrupts' using 35+% of the CPU.

If there is a 'recommended order' of diagnostic steps that should be followed, I would be happy to try that. I had previously tried unplugging devices to see if there was any impact, but nothing seemed to work.

I pulled the power plug to the CD drive and my PC has been behaving normally for about the past three hours now. I just plugged it back in and rebooted, but the problem seems to appear after a certain period of time.

I appreciate your patience. Thank You!



35+ years of 'progress' -- can't we all just go wire boards again?

 

[bcastner]

I was not impugning your intelligence, or technical skills. In a situation like this:

. there are no native IRQ statistids
. and "hot" devices are discovered by trial and error

If there is a 'recommended order' of diagnostic steps that should be followed, I would be happy to try that. I had previously tried unplugging devices to see if there was any impact, but nothing seemed to work.

I pulled the power plug to the CD drive and my PC has been behaving normally for about the past three hours now. I just plugged it back in and rebooted, but the problem seems to appear after a certain period of time.

"Hot" CD devices anymore are not unusual. There is nothing wrong with your diagnostic steps, this is a decidly an area were you unplug, test, plug, test, .etc issue.

I suspect a "hot" CD Rom, and this is not an unusual problem.

Best wishes,
Bill Castner

 

[Trevil]

Hi Bill,
I'm at work now and away from the PC, but I'm suspicious of my CD drive and/or WinAmp. Out of the clear blue, last night the CD drive popped the door open (there was no CD in it), so I yanked it's power cable and rebooted. Problem departed for over three hours, so put power cable back and rebooted. After a while, opened WinAmp again and problem started again (40% CPU used for Hardware Interrupts). So decided to use Norton and blocked Internet Access for WinAmp. It was running for 30 min. w/o problem when I went to bed. Checked it on the way out the door this morning and it was still OK. Will bang on it with a hammer tonight and see what happens.

I appreciate the time you are spending and your suggestions!
Wayne

35+ years of 'progress' -- can't we all just go wire boards again?