Memory error

August 10, 2006 6:51 PM

I've been having this error crop up for some time now and the address is usually 0x00000000, but today it popped up with a new number. What is causing this? It happens to many applications (was happening with firefox occaisionally, too). The whole text of it follows:

The instruction at"0x4000685e" referenced memory at "0x0106a298". The memory could not be "read".
Click OK to terminate the program
Click on CANCEL to debug the program

Is this a problem with my physical RAM or is there something sinister behind this and how do I go about tracking down the source of the problem?

posted by IronLizard to Computers & Internet (18 answers total)

If it was a problem with RAM, I would expect to see more bluescreens than simple application crashes. Try downloading something like Ubuntu Linux's live CD and see if your machine still has issues under Ubuntu.
posted by SirStan at 6:55 PM on August 10, 2006

Or MemTest86 it.
posted by SirStan at 6:58 PM on August 10, 2006

I second the memtest86 suggestion. Run it at least overnight ... Sometimes it takes a while for memory problems to manifest (and this sure looks like a memory problem to me, though normally I'd expect to get a proper blue screen) ...

Check your event viewer for errors ... (under XP it's start->control panel->Administrative tools->Event viewer->system).

If memtest86 comes up with a failure and you have multiple DIMMS (memory stick) in the machine, try removing one of them and running it again. You might be able to isolate a particular bad DIMM.

Good luck.
posted by crunchyk9 at 7:04 PM on August 10, 2006

Response by poster: I've run puppy linux from a live CD with no issues, though I never really pushed it. I've also tried the memory test, I forget which one (didn't require a reboot of CD though). Everything else has come up clean, though I do a have a fubarred install of windows defender (at startup it complains that it can't be started) I can't (haven't bothered much/too lazy) dispose of completely and I doubt that's the issue.

Event Type: Information
Event Source: Application Popup
Event Category: None
Event ID: 26
Date: 8/10/2006
Time: 8:46:06 PM
User: N/A
Computer: DEVIANCE1
Description:
Application popup: C++Builder 6: bcb.exe - Application Error : The instruction at "0x4000685e" referenced memory at "0x0106a298". The memory could not be "read".

Click on OK to terminate the program
Click on CANCEL to debug the program

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
posted by IronLizard at 7:13 PM on August 10, 2006

It means the program was attempting to access memory it does not have access to. This is probably not a physical memory error but rather a software bug.

0x00000000 is where null pointers point to. Dereferencing a null pointer is an extremely common programming error. Nothing you can do about it.

Why are you seeing these errors now, you ask? Perhaps you've installed Visual Studio, or some other IDE/debugger. The "cancel" option suggests that you have some development tools installed on your machine. Normal users don't get the option to debug errors. You wouldn't have seen this particular type of error window until you installed Visual Studio/C++/Basic or equivalent.

What can you do about this? Not much, really. Get less buggy software? I dunno.
posted by Khalad at 7:16 PM on August 10, 2006

I have *never* had a modern build of firefox crash. Are you running a clean install w/o any plugins? Maybe try creating a new user on your machine and testing under that?
posted by SirStan at 7:18 PM on August 10, 2006

Response by poster: I'm familiar with null pointers, but just barely. I did a BBS door in QuickC once (early nineties) and could never get rid of it because I had no idea what I was doing (it did, however, work in all other respects). It's annoying because the error hasn't been consistent with any software (with the exception of blender, there's a particular mistake I make almost every time I've used it that causes it and I don't remember exactly what it was). It goes from outlook to firefox(this one has since stopped) to word to whatever else, this is just the latest one. SVCHOST only stopped doing it after I tore it apart with procviewer, killed some odd looking things, ran hijackthis and killed a few more (this is when I trashed defender more or less purposely). I don't remember the specifics, but the malware was all confirmed through google searches and such. Now, however, everything comes up clean and the error still comes up at the most unexpected times.
posted by IronLizard at 7:29 PM on August 10, 2006

Excellent call Khalad.

Sure sounds like a buggy shared library to me. I'd try a Windows rebuild. If that doesn't work your best bet is a reformat (or new drive) and reinstall.
posted by crunchyk9 at 7:29 PM on August 10, 2006

Response by poster: That sounds good, can it be done with OEM XP disks?
posted by IronLizard at 7:37 PM on August 10, 2006

I'd just format the drive and re-install my operating system, but I don't keep much on my computer that I'd mind losing.
posted by Roger Dodger at 8:03 PM on August 10, 2006

I believe so, yes. The link above should walk you through the process. At some point you will be asked for a product key, so make sure you've got a valid one handy (they're often affixed to the side of the computer). Immediately after the install hit http://www.windowsupdate.com and install all the patches.

As always before doing major OS work ensure you've got the things you can't live without backed up. If you decide to back up to CD-R's make sure they're readable. Bad CD's of your most valuable digital assets makes baby jebus cry. Been there, done that, still makes me sad just thinking about it.

You may want to backup your files to your iPod if you've got one, or invest in an external hard drive. They've got some slick ones built specifically for laptops that don't require power other than the USB bus, a real win if you ever have the need to move around lots of data. As an aside, if you do backup to an iPod get the data off the iPod asap, they tend to get lost/stolen/damaged.

And by all means, a reinstall is a far more certain thing than a repair. If you can, I'd recommend a full disk-wipe/reinstall.
posted by crunchyk9 at 8:06 PM on August 10, 2006

Response by poster: Bad CD's of your most valuable digital assets makes baby jebus cry. Been there, done that, still makes me sad just thinking about it.

I have a stack of twenty useless cd's sitting around from my previous re-install. I will never use nero for backups again. I shouldn;t have needed those, so I didn't double check them because I assumed my data, being on a seperate partition, was safe. Boy, was I ever surprised when the oem reinstall hijacked the whole HD. Unforutunately, I don't have a better way to back up at the moment so it looks like this will stay for the time being.

I swear there has to be a better way than to re-install this damned OS every time something goes wrong. Even system restore breaks things done in between.
Any suggestions for near-bulletproof external data storage (2-10gig) that's not ridiculously expensive?
posted by IronLizard at 8:41 PM on August 10, 2006

I will never use nero for backups again.

Funny, it was nero screwed me as well.

for about 100 bux you can get a nice external drive from newegg. A friend of mine was just raving about the 60 gigger today. These drives are geared for portability and ease-of-use and not speed, so if you're thinking you might want to keep it hooked up to you desktop all the time a proper fill-sized drive external (for about 200 bux) may be the way to go.

Anything smaller won't get you a decent price break.
posted by crunchyk9 at 9:08 PM on August 10, 2006

Just before you decide to wipe the whole thing... You could try searching for a new version of the bcb.exe application, or you could uninstall borland.

bcb.exe seems to be an extra application layer that is catching memory referencing errors present in the code of the applications you are running. As an aside, these errors most likely occur for everyone else that is running these applications, but you have an extra application layer that is actually reporting on these errors.

You could even (as a last resort), rename this file to _bcb.exe, and then run the applications you frequently use to see of they still work. If you change the name of bcb.exe the application, it cannot be executed. This may not cause any problems with running the applications you actually want to run (firefox, etc.). Please take this as a last resort; I don't have this bcb.exe file so I can't test anything on my end.
posted by sleslie at 11:28 PM on August 10, 2006

Khalad writes "0x00000000 is where null pointers point to."

Pedantic quibble: in C and C++, while the programming language must ensure that equality comparison (==) of a null pointer with zero evaluates to true (and with != to false), the actual bit pattern of the null pointer is implementation (i.e., compiler/OS) defined, and is not required to be the same bit pattern as the bit pattern that represents zero.

Nevertheless, many implementations (i.e., compilers) do in fact use zero to represent the null pointer.

So it's possible, but unlikely, that 0x00000000 is not in fact an illegally dereferenced null pointer.
posted by orthogonality at 7:00 AM on August 11, 2006

Steve Summit, is that you?
posted by Khalad at 7:33 AM on August 11, 2006

I was getting errors exactly like this while using various applications which had previously run smoothly for over a year. memtest86 didn't show much, but I got a new stick anyway(I had been wanting to upgrade) and my problems went away. YMMV, but it might be worth seeing if you can borrow a known good and compatible stick and doing a temporary swap.
posted by Mr. Gunn at 8:58 AM on August 11, 2006

Response by poster: bcb.exe seems to be an extra application layer that is catching memory

Err, it's actually Borland C++ Builder 6. Though I really don't touch it much anymore since VS express came out. Then, of course, I'll just have to download borlands C# express when it becomes available and with any luck it won't require the .net framework, java, or some other ridiculous extra layer.

I was getting errors exactly like this while using various applications which had previously run smoothly for over a year. memtest86 didn't show much, but I got a new stick anyway(I had been wanting to upgrade) and my problems went away. YMMV, but it might be worth seeing if you can borrow a known good and compatible stick and doing a temporary swap.

Then this sounds about right. In november I added a 512m stick from amazon (kingston, I think) and while it didn't give me problems at first, it or the original stick might be doing this now. It's happening much less now, so I'll have to swap for a while to see which one it is.
posted by IronLizard at 10:34 AM on August 11, 2006