BSOD system_service_exception 0x3b

danielr0y · Jan 24, 2009

Hi all.
I've been experiencing a heap of different problems with my machine since updating to my ds4p mobo.
I've worked my way through the multitude of bluescreens it has kindly provided but cant seem to figure this one out, nor can i find any support for others having the same issues.

This is a fresh install of Vista business x64 sp1 that i had beleived to be heading for bluescreenless operation... Its the first error for this OS confirguration, but not the first ive had.
I originally believed it to be the fault of my everest installation, but have now recieved it without everest installed.

this is the closest i can find but provides no solutions for me...
http://www.vistax64.com/general-discussion/183791-bsod-system_service_exception-3b.html I was interested by the 4x1gb ram theory. im looking into new ram but would like to conclude on a problem before making a purchase.

My machine;
GIGABYTE ep45 ds4p- bios F8
E6750
4x1gb Corsair XMS2 DDR2-800
NVIDIA GeForce 9800 GT

the exact error i am receiving;
BCCode: 3b
BCP1: 00000000C0000005
BCP2: FFFFF960000BECD6
BCP3: FFFFFA600479EB80
BCP4: 0000000000000000

and although I've not had another since to confirm this, i believe I saw win32k amongst the message (it was set to reboot automatically at the time- the bluescreen only flashed quickly)

I'm not sure how to retrieve or analyze crash dumps but i'd be happy do provide i do you with some assistance.

A seemingly irrelivant but possibly connected problem was with the installation of itunes 8. I would stop incompleted with an error. Nothing has changed hardware wise since i have installed it successfully in the passed.

Any ideas/suggestions are greatly appreciated!
Thanks guys.

Dan

danielr0y · Jan 24, 2009

I've just realised the sheer number of errors i've had since this install yesterday :S
one has a solution from Microsoft. Think is could be the cause?

Thanks again guys.

danielr0y · Jan 24, 2009

i've had a good run without the bluescreen but there seems to be no stability and apps are crashing all the time...

danielr0y · Jan 24, 2009

One of the multitude of bluescreens I mentioned I've been having was apparently never resolved...

I've just had a stop error;

page fault in non paged area

BCCode: 50
BCP1: FFFFF800015386A8
BCP2: 0000000000000001
BCP3: FFFFF80002111E7B
BCP4: 0000000000000000

again, great thanks to anyone willing to help out.
dan.

dk70 · Jan 24, 2009

Stopcodes and error IDs dont mean much if memory is problem. You found the other thread so do Memtest86+.

I think we could have the same ram btw. Mine is 6400C4 with timing 4-4-4-12 - see more pages from the thread you linked to, cpu-z is a nice tool but also check bios. Voltage should be 2.1v or 1.9 or something else - go by Corsairs suggestions.

My 4x1gb cannot run at 800 speed even if only 750 in real life. Just wont work with the chipset I have. Corsair is very clear about only guaranteeing stated speed with 2 modules, not 4. Not saying that is relevant here but who knows. Why I know stop codes/minidump can be pointing in any other direction than where problem is

Possible bsod keeps on coming after memtest86+ give thumbs up. Error only show itself while ram is being stressed - memtest86 not sensitive enough. Use a tool called Prime95 for better test. Seems you have bigger problems than i ever had so i guess memtest86 will fail if ram is to blaim.

Corsairs page for your motherboard Welcome to Corsair :: Memory Configurator not many 4x1gb promises.

May be you have some mysterious Windows error but smells a lot like ram/bios/setting problem. No harm done in checking. Started after updating motherboard also yells ram - unless you forgot to power cpu fan

acdcfan · Jan 24, 2009

Danielr0y
BSoD screens and errors are usually saved in a way of minidump files.

how to find them
My Computer > C Drive > Windows Folder > Minidump Folder > Minidump Files. It is these files that we need you to attach to your next posting,or you can whim me and I'll give you my email address so that I can analyse them
Next time the Blue Screen occurs write down the error code that begins 0x followed by several zeroes and then a number followed by a letter or sometimes it just a number or two letters.
Like this: 0x0000007F

Zip up your minidump files ( with extension .dmp) and post here so that I can analyse them for you

dk70 · Jan 24, 2009

Just remembered that you should probably do some sfc /scannow. If memory setup is problem it can create no. 2, no. 3 and you and Debugger run around in circles

Before I lowered ram speed i would get "corrupted file detected and replaced" or whatever it says when sfc /scannow has finished. If I did same command few minutes later it spit out same result - i figured the whole process was too much and corruption happend during! May be you could try that as a very unscientific mem-test. Definitely do it at some point. Corrupt system file might even be what is bugging you. Memtest86 should come first though, rest dont matter if it fails.

Also run chkdsk (see tutorials if in doubt what that is). You are torturing Vista and filesystem so better check it is working.

danielr0y · Jan 24, 2009

now to start typing this message again... i got another system service exception just now while running orthos. i wont conclude that as the problem as i've completed an orthos test a good number of times before.

yep, we've got the same RAM.

attached are the dump files from the bluescreens- titled appropriately. system_service_exception_recent being the one i've just had whilst stressing with orthos.

system_service_exception_recent.dmp
BCCode: 3b
BCP1: 00000000C0000047
BCP2: FFFFF80001EC1B48
BCP3: FFFFFA60090E9210
BCP4: 0000000000000000

system_service_exception_1.dmp
BCCode: 3b
BCP1: 00000000C0000005
BCP2: FFFFF960000BECD6
BCP3: FFFFFA600479EB80
BCP4: 0000000000000000

and

page_fault_in_non_paged_area.dmp

BCCode: 50
BCP1: FFFFF800015386A8
BCP2: 0000000000000001
BCP3: FFFFF80002111E7B
BCP4: 0000000000000000

Thanks

danielr0y · Jan 24, 2009

yay another...

also whilst running orthos;
driver irql not less or equal.dmp

BCCode: d1
BCP1: 0000000000000000
BCP2: 000000000000000C
BCP3: 0000000000000001
BCP4: FFFFFA800707E1A4

thanks

danielr0y · Jan 24, 2009

and this whilst or after vista memory diagnosis tool (i walked out... came back to see it)

Driver irql not less or equal

BCCode: d1
BCP1: 0000000000000037
BCP2: 0000000000000002
BCP3: 0000000000000000
BCP4: FFFFFA6002571CEC

nvlddmkm.sys- address FFFFFA6002571CEC

dk70 · Jan 24, 2009

Well feed acdcfan

But you really should do memtest86. Just to start off and if it fails there is not much reason to spend hours troubleshooting Windows. If Windows is completely upside down then you cant use Orthos test for much. You are worse off than me for sure but i can recognize some events - certainly stop codes. I loaded one of your files, 3B win32k.sys process: firefox or replace with whatever, i say hello old friend, heh. 50 is familiar to overclockers i believe. We will see but i put money on hw for now.

So ram are set up as they should be? Voltage/timings done manually or EPP? SPD will probably be wrong. Try run with 2gb, easy test (in case you dont fancy looking at memtest86 dos screen for hours). You could settle for test no. 5 and no. 8 - the most likely to trigger problems with timings/voltage. May be 1 or 2 modules are simply bad and need replacement so you need the full tour at some point. Since it seems like situation is very bad I doubt it will pass 1 cycle - if ram is problem that is.

Could even be power supply though that is a long shot. Motherboard change should not mean that much. You do have any chipset fan set up properly? Just asking to exclude hardware and installation bugs. Temp readouts under orthos might give a hint.

What about chipset drivers when you changed mobo? Should not be a problem if done right but another idea...

danielr0y · Jan 24, 2009

OK I've now concluded my RAM has had it.

Memory Diagnostics Tool reported;
"hardware problems were detected" and
i also have since received a new bluescreen I've never seen before...
MEMORY MANAGEMENT.

I wouldn't mind hearing what you can get out of the dumps i uploaded though.
It would seem there are far too many different bluescreens for one problem and i don't want to overlook another problem that i may have too.

Thanks again.

dk70 · Jan 24, 2009

I would rather leave that to those who ask for it, heh

But I see nothing in there that would make me conclude anything. athrx.sys is a driver for a usb thingy right? Might be ready for an update but again memory problems kind of overrides this. As does other hardware problems, like heat, overwhelmed psu, not working motherboard. Windows is far from intelligent with debug files, not when it loses control and hardware just stop working. Very nice if you want to find crappy driver and such - on a working machine.

Well about the bsods im trying to say: you could have 1 problem which is memory, either on its own as in broken, or in combo with wrong bios settings/chipset issue with 4 modules. You dont fix that by looking at dump files. Only way is memory test, preferably memtest86. Then move on from there, you have excluded.

Your Windows might be so shaken that files are corrupted etc. - errors breeding like rats. So very hard and risk of running around in circles. To avoid that start from A and move forwards in logical steps.

Definitely remember the sfc /scannow and chkdsk once things have improved. If my suspicion is correct waste of time doing it now.

May be let the debug guy look at files if you think they reveal something new. I can only translate them to a certain point, heh. As of now i would not take them literally, think they are fake because Windows is slapped silly by memory bugs - only one way to find out though. Memtest86 fail is final - it must be able to run forever or you will get problems. Dont race around in Windows until test is approved, not healthy.

dk70 · Jan 24, 2009

May be i should read posts before replying...

so you cant run windows mem thingy from boot menu?

if so yes you have mem problem

Check bios for a start, volt/timings, Memtest again. No better then lower ram speed 1 notch, Memtest again. No better then test them 1 at a time and/or in pair - whatever works, trial and error.

I have visited Corsair forums more than once so pretty sure that if you use correct volt/timing settings and still fail Memtest86 they will replace modules. I doubt they will if you only have problems running 4x1gb and each work fine while on its own or in dual/paired.

Im not into Intel chipset performance - possible it is better to keep ram speed but increase timings, like 5-5-5 instead of 4-4-4. Dont know but it could make all the difference.

danielr0y · Jan 24, 2009

It is memtest86+ that i cant get to run...
When I run Memory Diagnostics Tool it runs the diag and reports, 'Hardware problems were detected'. Once it boots to vista it says you have a memory hardware fault etc.

I'm a bit scared that it could be the psu though... when changing the fan i hadn't discharged the capacitor and got electrocuted. I tested all the rails and they seemed to be working fine.

Come to think of it I had squiggly lines through my monitor that I resolved by reseating the graphics card- but perhaps it was just not getting enough power to it.
Same too the RAM. Perhaps its not receiving enough volts. I'll check that now.

danielr0y · Jan 24, 2009

I don't know a great deal about voltages but I've gathered the seemingly relevant values from bios.

Do these all appear to be right?

DDR18V 1.888

MCH/DRAM REFERENCE 0.9

DRAM VOLTAGE 1.8 (by the link you supplied me I'm thinking this is the one that should be at 2.1)
DRAM TERMINAL 0.9

CHANNEL A REFERENCE 0.9
CHANNEL B REFERENCE 0.9

Thanks heaps and if its no trouble i wouldn't mind learning what each of those is exactly.

dk70 · Jan 25, 2009

Well check what type of ram you have, there are several versions of 6400

Settings of importance in bios is in "MB Intelligent Tweaker (M.I.T.)"

Look for "Dram Voltage" and "Dram Timing Selectable" - will list cas, trcd, trp, tras - or what matters. With my ram that would be 4-4-4-12. Same numbers are also printed on modules, required volt included - open case and have a look. Numbers must fit modules. Good chance they dont if you use SPD/Auto.

You should also see something about memory frequency, at 800. If nothing works and modules are proven to be ok when tested individually/in-pairs you can try forcing 667. What I have to do. For now 1st test must be after the book, correct timings/volt at rated speed - and then we hope 4 modules is not an issue. Trial and error after that if no luck - must narrow down which module is going back. May be try 4x1gb at 667 or if modules are 4-4-4-12 try 5-5-5-18 at 800 - still fail i would expect 1 or more modules to be bad.

What is memtest86+, what do I use it for? - The Corsair Support Forums

Sometimes it helps just moving modules about, could be a bad "contact". Doubtful but dont be afraid to use force. I think some even "clean" ram using rubber/eraser.

danielr0y · Jan 25, 2009

My modules are 4-4-4-12 and i have got them manually set to that. I'll take a look on the stickers for the voltage and check against it.

I'm overclocking my CPU but it is limited by the RAM- which is under clocked at 7something.

I'll run the suggested testing and report back.
All the modules have been reseated by the way. They all have good contact.

dk70 · Jan 25, 2009

If you are overclocking Im not so worried any more. May be "Orthos" was a hint! I assume you test properly. Can only judge by what is being told though - you could have been clueless about bios and settings. Bios do have ton of them, like fixed cpu oc-presets, fixed ram-tuning presets, even pci-e oc´ing - stuff that rarely do much else than create problems. If you use any of that go default - later when ram testing is finished you can go crazy.

Will know more when memtest86 is done.

Seems like using 2.1v could be solution but let sticker on sticks decide, so many revisions etc. because they change chip all the time. All 4 sticks have same model number i hope?

danielr0y · Jan 25, 2009

I dont overclock when stress testing... just though i'd mention that it usually is.
RAM does say use 2.1 on the sticker. i've done that now and ill test the following and report back.

4x1gb

voltage 2.1
timings 44412
speed 800

4x1gb

voltage 2.1
timings 44412
speed 667

4x1gb

voltage 2.1
timings 55518
speed 800

then with each stick seperately at

voltage 2.1
timings 44412
speed 800

BSOD system_service_exception 0x3b

New Member

My Computer

New Member

My Computer

New Member

My Computer

New Member

My Computer

Member

My Computer

My Computer

Member

My Computer

New Member

My Computer

New Member

My Computer

New Member

My Computer

Member

My Computer

New Member

My Computer

Member

My Computer

Member

My Computer

New Member

My Computer

New Member

My Computer

Member

My Computer

New Member

My Computer

Member

My Computer

New Member

My Computer