Author Topic: [HELP!!] Win XP reboots sometimes after startup [Any Ideas?] (Read 32980 times)

tomos · « **Reply #25 on:** July 14, 2011, 03:15 PM »

Looks like the CPU might be the problem.
I booted with a Parted Magic CD - it got as far as this with the CPU message at the end

Screenshot - 14.07.2011 , 22_01_33.png

"kernel panic - not syncing: cpu context corrupt"

Ath · « **Reply #26 on:** July 14, 2011, 03:31 PM »

Hmm, the Bank 0: / Bank 5: messages would suggest me memory errors, not CPU, but both are alarming

Stoic Joker · « **Reply #27 on:** July 14, 2011, 03:54 PM »

Bank 5? With a 2GB memory kit (generally 2x1GB). I don't think the bank references are memory, on chip cache maybe... *Shrug* ...Not sure. But Memtest86+ already passed the memory.

Still could be the PSU; brownout on boot load spike.

Or just for fun disable all the onboard USB stuff and see what happens...it is the last thing it was loading before it go poof.

tomos · « **Reply #28 on:** July 14, 2011, 05:00 PM »

^ yeah, memtest did three full passes without errors...

Or just for fun disable all the onboard USB stuff and see what happens...it is the last thing it was loading before it go poof.
-Stoic Joker (July 14, 2011, 03:54 PM)

how do I disable USB? Physically - unplug cables?
There's usb on the front of the machine & I can see where the cable goes into the motherboard, the usb connection in the back is encased in metal - arrows going to the box containing the usb sockets, and to the right/above the circular thingy I see USB139 written

I also see "LAN USB 1" centre-right

40hz · « **Reply #29 on:** July 14, 2011, 06:06 PM »

A machine check exception can be caused by a board, CPU, RAM, temperature, power voltage, or driver problem.

To get to the bottom of the error code you'd need to go to the CPU manufacturer's website and see if there was more info on that message under the specific processor. Since decoding machine check errors is a real nightmare, your best bet is to post a message on the manufacturer's troubleshooting forum with details and ask for help if you suspect the CPU.

FYI: Intel does publish a diagnostic utility for their CPUs. It will do a stress test so if there's a flakey in t he chip, it can help spot it. Unfortunately, it's Windows software so it's not going to help you out too much under Linux.

Addendum: did a little research. Under Linux, this error is most commonly caused by bad cache on the CPU according to one source. See here.

Stoic Joker · « **Reply #30 on:** July 14, 2011, 06:17 PM »

^ yeah, memtest did three full passes without errors...

Or just for fun disable all the onboard USB stuff and see what happens...it is the last thing it was loading before it go poof.
-Stoic Joker (July 14, 2011, 03:54 PM)

how do I disable USB? Physically - unplug cables?
-tomos (July 14, 2011, 05:00 PM)

I was thinking more along the lines of a BIOS option to disable them, or at least cut them back to a lower speed and making sure nothing is plugged into them.

That or just try a different flavor of Linux, as some are touchier than others ... and Partition Magic is owned by Symantec these days...

tomos · « **Reply #31 on:** July 15, 2011, 01:46 AM »

Addendum: did a little research. Under Linux, this error is most commonly caused by bad cache on the CPU according to one source. See here.
-40hz (July 14, 2011, 06:06 PM)

from that link:

The Machine Check Exception looks like this:

CPU 3: Machine Check Exception 0000000000000005
Bank 0: b200004010000400
Bank 5: b200121020000400

On Windows systems a MCE (Machine Check Exception) could also mean bad RAM, Motherboard or Processor. Reseating the processor is also another good option if you experience this error message.

So, "reseating" the processor is another option.
RAM I'm pretty sure is okay, I'll try disabling USB as Stoic Joker suggests and then look at this.
I think it's no harm btw having a "sensitive" version of linux - as obviously windows is sensitive to the same problem.
OTOH the UBCD booted up okay - that has some CPU stress test programmes, if I can figure the damn thing out, but one thing at a time ;-)

tomos · « **Reply #32 on:** July 15, 2011, 01:52 AM »

how do I disable USB? Physically - unplug cables?
-tomos (July 14, 2011, 05:00 PM)

I was thinking more along the lines of a BIOS option to disable them, or at least cut them back to a lower speed and making sure nothing is plugged into them.
-Stoic Joker (July 14, 2011, 06:17 PM)

Doh!
you can tell I dont have much experience at this kind of thing!
Tried that, disabled USB controller - got exactly the same screen as above.

So I guess the next step is to use UBCD to stress test some stuff

tomos · « **Reply #33 on:** July 15, 2011, 07:03 AM »

UBCD:

Unfortunately, I'm unable to complete CPU tests - computer sounds okay but screen goes blank after a while.
I'm was unable to get the hardware tests to work at all. Press [Y] to proceed - nothing happens...
There were two apps that gave info about my CPU. The second said I had a virus in memory I think. I was photographing everything and guess what, the SD card no longer works (two weeks old, what's going on here ).

So I tried a anti-virus boot disk from c't magazine here - it's ubuntu based. Downloads latest av info from 4 different AV's.
Tested the partition that kept getting checked on reboot - clean
Tested C: crashed twice

I get the impression that nothing works for long on this machine which kind of makes it difficult to test any damn thing...

Ath · « **Reply #34 on:** July 15, 2011, 08:12 AM »

Did anything else happen to that machine? A 'lightning skrike' like nudone had about 2 weeks ago, a power brown-out, or something similar?

It's just getting from bad to worst.
I had that with a PC that wasn't moved for about 5 years, and after moving it for half a meter ( ! ) it just about fell apart, rebooted spontaneously, or Windows hung without any warning. The ASrock MB inside I had to declare dead, nothing good could be had from it, and I simply replaced the entire machine to avoid endless trouble and support (the owner lives ~1/2h drive away, and I could only go there in the evening because of work obligations).
Maybe this is the point where you should drop this baby/culprit too? $:-\$

tomos · « **Reply #35 on:** July 15, 2011, 09:10 AM »

^ agreed Ath, I've already been looking at options for a new machine.
There is a shop in town that promise to find what part is wonky for a reasonable price - I could try that, but think I'll have to get a new machine anyways.

As to what happened I'd really love to know...

It's actually running now but I'm just getting essentials off of it for immediate use elsewhere so I dont have to go to online backups. Nah, crashed again - I can get it going (sometimes it reboots a couple of times then starts normally) but then it crashes after 5 or more minutes. Could well be a temperature thing....

40hz · « **Reply #36 on:** July 15, 2011, 09:58 AM »

^Sounds like a golden opportunity to further your education playing with it if you're getting something else to work with.

Look at it this way - it's already broken, so you don't need to worry about wrecking it.

I learned a lot about PC diagnostics playing with already broken machines. It's one thing to read about error screens and codes. But when see them happening in front of you, it makes for a more direct personal learning experience.

Just my 2¢ anyway.

tomos · « **Reply #37 on:** July 15, 2011, 10:14 AM »

^Sounds like a golden opportunity to further your education playing with it if you're getting something else to work with.
-40hz (July 15, 2011, 09:58 AM)

Yeah,
for a short, wild and free moment, I even considered building a new pc or rebuilding this one, but then sanity checked in. I think to be a builder you need to have extreme patience (well - it looks like extreme patience to me!) I get stressed out too easy - this whole thing was a good exercise in keeping cool under fire so-to-speak but I could do with a break

40hz · « **Reply #38 on:** July 15, 2011, 11:09 AM »

^IMO doing your own build is actually quite easy if you rule out overclocking, bleeding edge hardware or video, and water or other exotic cooling systems. A couple of screwdrivers, the willingness to read documentation, and some patience - plus a well lit and clean workspace - are all that's really needed. Take a few sensible and modest anti-static precautions and you're off to the races. A few hours later and you're enjoying your new toy.

The only problem is it will usually cost more (simetimes a lot more) than a comparably configured "standard" machine from one of the big manufacturers.

So if you're not going to go with an "enthusiast" or "hard-core gamer" build, why bother? At least with current pricing.

I think you're smart just wanting to buy one if you're not into the whole DIY thing.

tomos · « **Reply #39 on:** July 15, 2011, 11:45 AM »

So if you're not going to go with an "enthusiast" or "hard-core gamer" build, why bother? At least with current pricing.
-40hz (July 15, 2011, 11:09 AM)

I'd like a machine where I can have up to 3 harddrives, so extendable, also memory-wise. I think I'll do what I did last time - buy from a site where you can pick out the components and they'll build (for 20€). Ran well for almost 4 years - I expected more, but I guess that can never be predicted...

But will have to look closer at Dell later as well.

Stoic Joker · « **Reply #40 on:** July 15, 2011, 11:59 AM »

There were two apps that gave info about my CPU. The second said I had a virus in memory I think. I was photographing everything and guess what, the SD card no longer works (two weeks old, what's going on here ). [/li]
[/list]

So I tried a anti-virus boot disk from c't magazine here - it's ubuntu based. Downloads latest av info from 4 different AV's.
Tested the partition that kept getting checked on reboot - clean
Tested C: crashed twice
-tomos (July 15, 2011, 07:03 AM)

Can you try rerunning that with the drives disconnected and see what happens? Basically strip the mahine down to the bare essentials CPU, Mboard, Memory, power, and video. No modem, sound, or other cards, and no FDD/HDDs to draw power. Just to see if it'll make it far enough into the CPU test to (get a better error message, if any...and) hazzard a guess at a conclusion.

Shades · « **Reply #41 on:** July 15, 2011, 12:42 PM »

As previously suggested: try a new power supply. In my experience it is the reason behind all kinds of strange and unrelated errors popping up.

Or in case you still want to spent time with it, do the following:
- Track the cables from the power connectors (disregard the ones going to the main board)
- You'll notice that (in most cases) that these connectors are attached to two distinct sets of cables. Each of these sets is called a 'rail'.
- Switch the power connectors from one rail to the other rail
- Reboot and run your tests again.

It happens that sometimes one of the power rails doesn't supply the required voltages and/or currents anymore (just outside the margins set up by the ATX standard). So it appears that the power supply works, but in reality it doesn't.

Electronics behave very strange when they do not get the juice they need, leading into all kinds of strange and unrelated errors.

Most people/manufacturers skimp on the power supply because they overlook the importance of it.

tomos · « **Reply #42 on:** July 15, 2011, 02:23 PM »

Sorry guys, curiosity's got me...
...
Can you try rerunning that with the drives disconnected and see what happens? Basically strip the mahine down to the bare essentials CPU, Mboard, Memory, power, and video. No modem, sound, or other cards, and no FDD/HDDs to draw power. Just to see if it'll make it far enough into the CPU test to (get a better error message, if any...and) hazzard a guess at a conclusion.
-Stoic Joker (July 15, 2011, 11:59 AM)

Sounds like a good idea - I was going to give it into the shop without harddrives so may as well try it myself first

(but focusing on looking for new pc first, so it may take a little while)

tomos · « **Reply #43 on:** July 15, 2011, 02:39 PM »

Most people/manufacturers skimp on the power supply because they overlook the importance of it.
-Shades (July 15, 2011, 12:42 PM)

I did get a reputable (at the time) brand power supply - Corsair 450w - should have supplied what I needed no problem, but of course that's no guarantee....

tomos · « **Reply #44 on:** July 15, 2011, 03:37 PM »

Can anyone recommend a diagnostic boot disk that has a relatively familiar user interface and isnt too complex to create (i.e. not Win PE) - I often find myself lost in the UBCD one (I have no DOS experience)

cmpm · « **Reply #45 on:** July 15, 2011, 03:49 PM »

Could be one of your controllers. Go to device manager.
Don't disable them, but uninstall them one by one and reboot.

Windows XP will reinstall them, when found.

JavaJones · « **Reply #46 on:** July 15, 2011, 03:51 PM »

It's not a Windows driver issue as problems also occur in Linux bootup environment. This indicates pretty strongly that it's a hardware problem of some kind.

- Oshyan

Stoic Joker · « **Reply #47 on:** July 15, 2011, 04:01 PM »

Electronics behave very strange when they do not get the juice they need, leading into all kinds of strange and unrelated errors.
-Shades (July 15, 2011, 12:42 PM)

We on same page.

steeladept · « **Reply #48 on:** July 15, 2011, 04:36 PM »

My first idea from what I saw you post from this morning is get your hard drive out of the machine ASAP so stray currents don't fry that - you can always slave it into another machine to get your data later. After that, the other ideas are pretty good. Strip it down to the bare essentials (MB, power supply, minimum ram, CPU - that should be about it) and boot to a live drive (I suggest Hiren's boot CD - it is pretty good with lots of diagnostic tools, though nothing so easy as "run" -> "here is your problem..." solutions). If all the flakiness continues (I think you already did close to this and said it did), change the Power Supply (if possible). Once that is changed, if it still continues, swap ram (again, if possible). If it STILL continues, replace the MB/CPU assembly. The price difference (timewise) in general for finding the individual bad component and replacing either the CPU or Motherboard is not worth it compared to the cost difference of replacing both.

I think you are about 30% there already, so if you do continue to work on it yourself, that would be how I would do it. You don't need to find the problem, just isolate it and replace. BTW: If it is the MB/CPU, make sure you pay close attention to your PSU and backup plan religiously. It might well be a flaky PSU caused the MB/CPU to go bad....If it isn't too hard financially, it might be worth replacing the PSU anyway and keeping that one as a spare if it seems good.

cmpm · « **Reply #49 on:** July 15, 2011, 05:28 PM »

I think scsi refers to controllers, is why I posted that.
Reinstalling the controllers could address the issues.
Same drivers but maybe a fix of conflicts.
Never know till you try.

Author Topic: [HELP!!] Win XP reboots sometimes *after* startup [Any Ideas?] (Read 32980 times)

Author Topic: [HELP!!] Win XP reboots sometimes after startup [Any Ideas?] (Read 32980 times)