General Protection Fault

[tinman0]

Before I start pulling a machine apart, whats the opinion of this fault.

Machine has died - go to start it up after 2 days powered off (to make sure it wasn't a heat related issue).

Machine goes through the BIOS fine, asks to boot from CD for 5 second, and then goes for boot device 1, which is an internal bog standard IDE drive with XP. At this moment i get a General Protection Fault. It doesn't look like a windows fault, its just a text screen with a few interesting bits of information. (Interesting that is if you are a nerd with no life).

Where could the fault be coming from? Is the main HD at fault? the MB? the install of XP (its run fine for a couple of years).

[thepassenger]

Blue screen with white text?

If so then yep, it's a BSOD. When it comes up on screen take a look near the top... your might find something like "IRQ_NOT_LESS_EQUAL_TO" or "PROCESSOR_TRAP" or really any sort of weird text. Underneath might be a driver or module that generated the error; you *might* be able to track down which driver (if any) is at fault and by taking the machine in to safe mode be able to fix it.

Also a repair install is none destructive and might enable you to get the system working for a while longer so you can check that your latest backup is still current.

This of course assumes it's a driver or peice of hardware causing the GPF; I've seen Windows brought to it's knees and blue screen with some crufty malware...

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

[TheExcession]

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

[tinman0]

i'm thinking its not HD actually.

suddenly remembered to stuff a redhat install cd in to see what would happen. it lasted about 10 seconds longer than XP did, so I'm thinking its MB (or memory or proc).

considering its a celeron 1.1Ghz, its not exactly cutting edge so I cannot imagine that would be at fault. memory is nothing special either. so i'm down to either MB or PSU. if i buy a PSU it'll be MB. If i buy a new motherboard (and proc) it'll be PSU.

the error isn't blue screen. its the straight forward DOS type text. it could be bios.

machine did get unreliable just before it went off for good. (eg freezing after a minute or two on reboot).

[mph999]

TheExcession said:

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

he he, found a unix box at work the other day that's been up for 4 years ...

Suspect the patches may be out-of-date, but's it's low priority, so I guess if it works ...

Martin

[greenlandy]

Any chance you could post a piccy of said screen?

[agent006]

TheExcession said:

but only on Windows eh? We have linux systems that haven't been rebooted since 2004...

And i've got linux boxes here that barely last a few hours before wheezing to an NT-esque standstill.

[Plotloss]

What are these interesting bits of information for nerds with no life?

Because you see thats the crucial bit.

[thepassenger]

TheExcession said:

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

[thepassenger]

TheExcession said:

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

Relative merits of operating systems aside, Desktop's tend to need software refreshes far more than servers. After all a server isn't having any "Ohh, I wonder if this is useful?" type stuff installed on it. Windows *CAN* get very good uptimes however to do so you need to forgo most security patches/service pack installs; same as Linux in that respect (need to reboot when you upgrade the kernel).

But back on topic. Black text could still be windows; the main problem is what exactly is the error message? "General Protection Fault" usually has something following or preceding it (i.e. General Protection Fault: IOS.SYS could not load).
Skipping that as you've thrown redhat at it and it last marginally longer it could well be the mobo or CPU... would be interesting to know where in it's boot process RH crapped out.

Hell it could be that the heatsink is gunked up and the processor is overheating; speaking of which does your BIOS have temp sensor readings? Leave it sat with them being displayed and see if the system still crashes... or indeed if the temp go to silly levels.

[tinman0]

mph999 said:

TheExcession said:

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

he he, found a unix box at work the other day that's been up for 4 years ...

Suspect the patches may be out-of-date, but's it's low priority, so I guess if it works ...

here is one of our sun boxes:

10:41am up 1667 day(s), 14 hr(s), 2 users, load average: 0.17, 0.12, 0.12

[tinman0]

thepassenger said:

But back on topic. Black text could still be windows; the main problem is what exactly is the error message? "General Protection Fault" usually has something following or preceding it (i.e. General Protection Fault: IOS.SYS could not load).
Skipping that as you've thrown redhat at it and it last marginally longer it could well be the mobo or CPU... would be interesting to know where in it's boot process RH crapped out.

Hell it could be that the heatsink is gunked up and the processor is overheating; speaking of which does your BIOS have temp sensor readings? Leave it sat with them being displayed and see if the system still crashes... or indeed if the temp go to silly levels.

I'm leaning to cpu/mb at the moment. I'm going to pull the box apart and make sure heatsink is still in place as thats my first concern.

RH crapped out just after the boot loader "Uncompressing linux etc" and that was the last of that.

[onomatopoeia]

TheExcession said:

thepassenger said:

But ultimatley it might be time to look at the old warpig and decide that after 2 years of use it's time to format and reinstall from scratch.

but only on Windows eh? We have linux systems that haven't been rebooted since 2004... ahhh I see that's only 2 years too

My mail server got to over 700 days before SWEB rebooted it for me. Aside from power outages and being moved when I moved my study from bedroom 2 to bedroom 3 it has been up for 5 years, so is still running debian potato on 2.2.19 kernel .

[thepassenger]

tinman0 said:

I'm leaning to cpu/mb at the moment. I'm going to pull the box apart and make sure heatsink is still in place as thats my first concern.

RH crapped out just after the boot loader "Uncompressing linux etc" and that was the last of that.

If the heatsink is gunked up then a soft bristled paint brush is very handy. And Maplins can supply replacement fans if it's dead.

[Bodo]

agent006 said:

TheExcession said:

but only on Windows eh? We have linux systems that haven't been rebooted since 2004...

And i've got linux boxes here that barely last a few hours before wheezing to an NT-esque standstill.

Must be YOUR fault then

[thepassenger]

Bodo said:

agent006 said:

TheExcession said:

but only on Windows eh? We have linux systems that haven't been rebooted since 2004...

And i've got linux boxes here that barely last a few hours before wheezing to an NT-esque standstill.

Must be YOUR fault then

Nahh, kernel versions + drivers + 3rd party apps + high load. I've seen big unix machines (P690, AIX 5L) configured by experts in their fields collapse under its own weight due to odd issues that only croped up when the box was maxxed out. Building a machine that'll spend it's days nearly idle with the occasional peak of activity is easy; building a machine that can run flat chat for days on end is a lot tougher than you'd imagine.

[JonRB]

tinman0 said:

its just a text screen with a few interesting bits of information. (Interesting that is if you are a nerd with no life).

Yeah, those are called "diagnostics" and they're stuff that us programmers put into software in the vain hope that fwit users will report the information back to our support guys so that we have more than a gnat's chance in a hurricane of working out what the problem is.

Only 9 times out of 10 we just get a garbled message along the lines of "it don't work and prints out a load of shit that I couldn't be arsed to read".

[tinman0]

JonRB said:

tinman0 said:

its just a text screen with a few interesting bits of information. (Interesting that is if you are a nerd with no life).

Yeah, those are called "diagnostics" and they're stuff that us programmers put into software in the vain hope that fwit users will report the information back to our support guys so that we have more than a gnat's chance in a hurricane of working out what the problem is.

Only 9 times out of 10 we just get a garbled message along the lines of "it don't work and prints out a load of shit that I couldn't be arsed to read".

if there had been any information worth posting i would of done so. in fact, if there was any information worth posting i would of typed it into google in the first place.

i'm a nerd anyway. if nerds can't rip the piss out of each other, who can? chill.

[agent006]

Bodo said:

And i've got linux boxes here that barely last a few hours before wheezing to an NT-esque standstill.

Must be YOUR fault then [/quote]

Red Hat Enterprise 3, straight out of the box. Lasted about half a day before i lost patience with its flakiness and intentional complexity and put 2003 server on instead.