Kernal Panic even with Live CD [SOLVED]

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

Hi,

I have been a gentoo user for a long time now and this is the first time I encountered this problem. I seem to get a kernal panic everytime I boot up the system. This could happen very early in the boot process or few minutes after a successful bootup. The worst thing is this even happens when I boot the system with the live CD. Below is screenshot of the kernal panic.

http://server6.theimagehosting.com/image.php?img=S7007452-1.JPG

work done so far:
1. I have checked all partitions with e2fsck and reiserfsck, no curruptions found.
2. Have replaced the CMOS battery with a new one because it died.
3. I have two 256MB DIMMs installed, checked them both individualy. same problem no matter which slot i stick them in.
4. Recently done a complete world update.

My machine is a 6 year old compaq presario desktop. The kernel version is 2.6.23. I could not get a emerge --info because I cannot get a shell prompt.

I'm beginningto think if both my memory cards have died.

Any help, ideas are greatly appretiated.
_________________
"it's not the bullets; beauty killed the beast"

Abraxas · l33t Joined: 25 May 2003 Posts: 814

What if you try booting with acpi=off?

NeddySeagoon · Posted: Sat May 23, 2009 11:29 am Post subject:

Kryptonite,

Remove the cover and look at the capacitors (tall cylinders) around the CPU.
These are (mostly) parts of the voltage regulator for the CPU core, which operates at a lower voltage than is supplied by the PSU.

The tops should not be domed, no fluid should be leaking out of the bottoms onto the motherboard.
These parts get a very hard life and when the fail, the CPU core voltage cannot be held within limits so all sorts of odd errors happen.

These parts can be replaced if you are already skilled in the use of a soldering iron.

While you are inside the box, operate the system briefly and check the fans are all working. They can fail suddenly.
Also, using a stiff natural bristle brush, clean the dirt out of the CPU heatsink.

Capacitor failures cause random issues, often but not always on boot.
Thermal issues (failed fans, clogged heatsink) normally show up after booting from cold due to the system thermal inertia.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

Guys..thanks for the tips. I will try them out little later on. at the moment I'm testing the memory with memtest86+ so far it has found two errors..strange thing is I did not get any errors in the previous run..but atleast it is an indication. would you know of any tools like memtest that I can use to test BIOS, CMOS etc..??

I did not see any capaciters gone bad at the first glance but will have a thorough look after this.
_________________
"it's not the bullets; beauty killed the beast"

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

Here is a screenshot of the memory test

http://server6.theimagehosting.com/image.php?img=S7007453-1.JPG
_________________
"it's not the bullets; beauty killed the beast"

NeddySeagoon · Posted: Sat May 23, 2009 1:13 pm Post subject:

Kryptonite,

memtest86 tests lots of things beside your RAM.
If the errors are repeatabale at the same locations, its probably RAM.

When the errors are random bits at random locations its likely to be something else.

If you think its RAM, you may want to try the badram kernel patch. I have no idea if its still maintained.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

NeddySeagoon,

It seems I spoke too soon. Here's what have done so far.

1. acpi=off kernal flag has no effect
2. I have removed all extra pci cards and replaced the vga card with another nvidia
3. tested each memory card on both slots individualy and together using memtest86: no errors

Non of the above had made difference. I still cannot get a successful bootup from HD or LiveCD.

weeks ago I started getting segmentation faults when I run transcode or mencoder. Then came the kernal panics. could it be the motherboard, CPU or perhaps the powersupply?

I'm out of ideas what to do next....!!!
_________________
"it's not the bullets; beauty killed the beast"

NeddySeagoon · Posted: Sat May 23, 2009 4:34 pm Post subject:

Kryptonite,

It can be all of those things.

Did you check the capacitors and clean the heatsink ?

Can you test with a replacement PSU ?
Thats really the only way to check the PSU.

Its good to do a swap to see if your PSU makes another known good system bad. Thats much more convincing than making your system good by swapping PSUs.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

I did check the capaciters..no bloated ones all seems fine..I gave heatsinks a good vaccum treatment as well. no extra SPU that I can swap i'm afraid. only sparepart I had was the VGA card.
_________________
"it's not the bullets; beauty killed the beast"

NeddySeagoon · Posted: Sat May 23, 2009 7:25 pm Post subject:

Kryptonite,

Can you borrow some spare parts to test with ?

Disconnect everything from the PSU that is not needed to boot. Likewise, remove all the plug in cards except the video card.
The idea is to reduce the load on the PSU to the bare minimum.

The next test is to feel the temperature of the CPU heatsink.
If its stone cold, when mentest86 runs, thats a bad sign.
It indicates that the heat cannot cross the boundary from the the CPU to the heatsink, so the CPU is running hot.

The fix for this is to replace the thermal compoud between the two. Its not easy as you must not damage the surface of the CPU or the heatsink and the two may be stuck fast - depending on the original compound.

Clean all traces of the old compound from both surfaces and put back a smear of new stuff.
The most common mistake here is to use too much. Thermal compound is supposed to drive the air out of the irregularities of the mating faces as it conducts heat much better than air. However, its not nearly as good a metals or the silicon of the CPU, so you do not want a 'layer' in the joint.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

DaggyStyle · Watchman Joined: 22 Mar 2006 Posts: 5941

question, do you have any boot parameters set?
the last error msg from your first image says unknown boot option
_________________
Only two things are infinite, the universe and human stupidity and I'm not sure about the former - Albert Einstein

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

I'm dont use any special boot parameters. just nox when i boot from the live cd.

I left it running for several minutes and the observations are below.

1. I can see the both cooling fans working on CPU and Power supply.
2. CPU heat sink does get heated as one would expect.

http://server6.theimagehosting.com/image.php?img=S7007456-1.JPG

I'll c if I can nick a powersupply when I get to the office on tuesday. my PSU is a HP (277979-001) 220-Watt Power Supply. I found cheap power supplys on ebuyer but well over 300watts. I'm wondering if I should by a new powersupply but not sure how to choose.
_________________
"it's not the bullets; beauty killed the beast"

krinn · Watchman Joined: 02 May 2003 Posts: 7470

NeddySeagoon · Posted: Sat May 23, 2009 9:22 pm Post subject:

Kryptonite,

AAAGHH!! A Compaq PSU. Be very careful.

Compaq used to use the standard connectors but with the wires swapped around.
The effect was that if you put a Compaq PSU on a non Compaq motherboard, nothing worked (it was harmless) but the other way round destroyed the motherboard.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

Thanks for tip..to tell you the truth this pc was shiped with some of the worst componants in their assembly line.

Microstar 6553 motherboard: which is not listed either on Compaq or MSI websites

apparently this motherboad and compaq powersupplys known cause problems as I found many posts online.

below link is what I think is a suitable replacement for the motherboard.
http://www.ebuyer.com/product/102909

but i'm thinking if should invest in a new powersupply first? perhaps buy this tester (http://www.maplin.co.uk/Module.aspx?ModuleNo=48647) and test the powersupply before that.

how can I find out if my pc is an ATX or ATX2 or 3 etc..?? this machine was purchased back in 2003 so I like to assume it's ATX. I have 2 HDDs, 2 DVD burners, a floppy connected so choosing a powersupply is difficult. the current PS is 220Watts..if I fit in a 300Watts or above will it harm the motherboard??
_________________
"it's not the bullets; beauty killed the beast"

NeddySeagoon · Posted: Sun May 24, 2009 11:59 am Post subject:

Kryptonite,

Maybe you don't have the unique Compaq PSU pinout problem then, as you have a non Compaq motherbaord then?
Its worth checking the 20 pin ATX power connector wiring (looking into the pins):-

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

NeddySeagoon,

thanks very much for the advice. unfortunetly I already bought a tester this morning and had a go at it.

http://server6.theimagehosting.com/image.php?img=S7007459-1.JPG

As you can see from pictures wiring colors are quite different. when I pluged in the 20pin motherboard code (one on the right) +12v2 check is showing a fault and started flasing and beeping. now I do not know if this an actual fault or simply becuase the PSU is wired differently..??

you asked an important question. it may not be worth buying replacements for a 6 year old pc. I build this pc in to a media centre over the years with DVB, new HDDs, fast DVD buner, 4.1 soundcard and everything so buing a replacement componant is still ceaper than buying a new pc. i'm still on the fence i'm afraid. on the otherhand I might actualy digg my self a very big hole buying replacements.
_________________
"it's not the bullets; beauty killed the beast"

poly_poly-man · Posted: Sun May 24, 2009 4:41 pm Post subject:

umm - compaq was standard always. You're thinking about dell, which swapped stuff around (also had an extra connector for a period of time)

EDIT: either that, or this is news by me. Looks like the OP just has a bad psu...
_________________
iVBORw0KGgoAAAANSUhEUgAAA

avatar: new version of logo - see topic 838248. Potentially still a WiP.

NeddySeagoon · Posted: Sun May 24, 2009 6:10 pm Post subject:

Kryptonite,

All the numbers look ok except the 12V2 which reads like LL in the photo.
Don't get to hung up on the colours of the wires as long as all the wires for the same voltage are coloured the same and are in the same pins as my ASCII art. Different vendors did change the colours but kept to the standard pinout.
Not all PSU vendors provide a -5v supply as it was only ever used for DRAMs that needed four power supplies and they stopped being used with the introduction of the IBM AT with the 286 CPU.

PSU good came good in 250mS, so the PSU is within reasonable limits under the test conditions, which will be static loads.
Do your own ASCII art of your ATX connector or provide a photo or two of the wiring and I'll look it over.

If you are going to spend money, a PSU is lower cost than a motherboard, so its the one to replace first.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

NeddySeagoon,

Here are wiring for the ATX connector. I'm not sure how to place voltages though.

NeddySeagoon · Posted: Mon May 25, 2009 6:17 pm Post subject:

Kryptonite,

Thats a standard ATX pin out with the wire colours changed and the -5v supply missing.
Thats ok as the last systems to use the -5v were 286 based ones but the ISA and then the PCI spec required it to begin with.

Your Black is GND, thats standard,
Your Red is +5v, thats standard.
Your brown is 3.3v - my orange.
Your Orange is +12v, so you should have red orange and two black wires in your disk drive power connectors.
Looking in your PSU tester image, I can see two reddish wires but not tell them apart.

Its quite safe to replace your PSU with a standard ATX PSU provided the case and fixings are suitable.

Thinking about what the +12v would be used for on your motherboard, I'm a little bit at a loss. Its provided to PCI slots and its used by the serial port inputs and outputs which need +12 and -12v It is unlikely to be used for the Vcore regulator in a six year old system but I can't read the voltages on the two large silver capacitors between the CPU and the video card (close to the CPU).
If they are rated at less than 12v, your Vcore is derived from the +5v and the +12v is used only for the PCI slots.

That does not mean your PSU is good, just that the +12v supplied to the motherboard is unlikely to be the problem.

============ Edit =============

Compaq HP-Q220PC3P ATX 220W Compaq Evo Series Power Supply Bulk-Pack Refurbished
Thats a warning if ever there was one. Its the same age as your own PSU and probably had just as hard a life. It is the same part as your own PSU, both are pn 277910-001. Look in the top right hand corner of your image.

Unless its a special case size, get a standard ATX PSU, around 300w
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

You are indeed right about the wiring on molex HDD power connectors. Orange, red and in-between the two blacks.

it reads 6.3V on the two tall silver capacitors next to the CPU. so could it still be derived from +5V like you mentioned??

my understanding is that the PSU tester beeped on +12V2 test because it was too low and given the wiring on the ATX connector is indeed standard that might be the fault...the simplest reason

I'd looked up many PSUs on ebuyer but non of them that I can afford do not match the dimensions. they are 97mm(H), 130mm(W) and 154mm(L) approx.

If I buy a little smaller standard PSU I'm worried the screw nuts will not fit on the casing and I would still have to buy extra cable converters. e.g. P3 and a really rare two pin fan connector P9 I think.

I found an OEM PSU on ebay for 20quid...used of cause..I'm tempted to buy this one..If it could get me through for another 6, 7 months I'd be quite happy..
_________________
"it's not the bullets; beauty killed the beast"

NeddySeagoon · Posted: Mon May 25, 2009 9:18 pm Post subject:

Kryptonite,

As the capacitors say 6.3V, they cannot possibly be used on the +12v - they would explode.
Its unlikely that the +12v supplied to the motherboard is actually used there except by the UART for tthe serial ports and supplied to the PCI slots.

It is unlikely to directly account for the problems. However, regulation of all the different voltages is not independent.
It may be interfering with the +3.3v or +5v regulation.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

pappy_mcfae · Posted: Tue May 26, 2009 7:26 am Post subject:

If I may butt in, from the photo of the inside of the machine, I can see your heat sink is nearly plugged solid (or was at the time of the photo). Also, with those wires hanging precariously close to the heat sink, I can see an overheat in your past. Blow everything out (PS included). If this issue remains, changing the board might not help.

The only way to know for sure is if you set your kernel for MCE, and also loaded the MCE software. Rumor has it that MCE only works for x86_64 systems. However, if you find errors in /var/log/mcelog, your CPU has gone on to greener pastures. It should remain an empty file.

Was it an Intel or an AMD chip?

Blessed be!
Pappy
_________________
This space left intentionally blank, except for these ASCII symbols.

Kryptonite · n00b Joined: 08 Jan 2006 Posts: 60 Location: London

Hi..It's an AMD Athlon xp 2100..FYI I don't think there was any overheat as I did not see any indication of such. the motherboard looks absolutely fine.

NeddySeagoon,

my question is this..if the ATX connector is indeed standard and PSU tester indicates a fault in the 12v power supply. couldn't that be the problem..?? even if it is not directly powering up the motherboard...may be it's causing the PCI cards to act strangely hence by screwing up the interrupt system.

from your opinion whats the best way forward...? should I invest in the OEM PSU I found on ebay...?
_________________
"it's not the bullets; beauty killed the beast"