0xc0000005 error

**Hankyaku** · 03-10-2008

These are interesting.
I mean of course I did insert new variables, and assigned some functions as well. But I use static arrays, and this makes such overflows (as far as I know) impossible, as if I try to write to them I'd run into exceptions upon programming, not only upon execution (since nothing changes, but in one case I compile&run&debug, and in the other I just run the exe).
Is there anything in C builder which would make these kind of errors traceable during development? Is there a way I can monitor the memory usage of my program? I didn't really need to use advanced C builder features until now...

**matsp** · 03-10-2008

I wouldn't trust the compiler to always [or even often] detect out-of-bounds array accesses. The C language doesn't do that very well, and whilst the latest MS products do have some extensions to make an attempt, it's quite possible that you can get these things wrong and still execute for quite some time.

--
Mats

**grumpy** · 03-10-2008

There is certainly nothing special about how BCB handles static arrays that allows any assurance that out-of-bound accesses will be detected, let alone generate exceptions. If you're assuming that, you're wrong. And it explains why you're having trouble: if you assume your code won't cause problems because it uses static arrays, you virtually guarantee you won't find any problems it causes.

The problem with out-of-bounds array accesses (or other forms of pointer molestation) is that there is no single effect, so there is no guaranteed way to detect them. The preferred approach is to prevent them in the first case: which, clearly, is a bit late for you.

**Hankyaku** · 03-11-2008

Okay. But how should I prevent these causes? I use static arrays, which take less than 1 Megs of memory in total. The ony way I access these arrays is that I load and read them in certain cycles. The only occasion I run through the whole array is at the start, when I clear them.

So even if I accept that not all of the out of bounds accesses are not detected, what can I do against it? When I'm saying that "I use static arrays", I also say, that I monitor their use to avoid out of bound access (in my cycles when I load or read them).

Is it possible by the way that I use too complex structures, or too big arrays?

(BTW: I began rewriting the codes, and I nearly finished, but I didn't run into any the problem.)

**matsp** · 03-11-2008

Well, it's certain the problem you are seeing is caused by a bad memory access - it is impossible to say without access to the source [and a failing test-setup] to say what the actual problem is.

Edit: That was probably a bit brief. You need to write code defensively, checking all indexes to ALL arrays, all over the code. And there is obviously a possibility that your library that you use is broken in some way.

Can you run a debugger and catch the actual error and get a stack-trace of where it is when it falls over?

--
Mats

**laserlight** · 03-11-2008

When you say static arrays, you mean arrays in contrast with dynamically allocated arrays, right?

It seems to me that you could use the at() member function of the fixed size std::tr1::array class template or std::vector.

**Hankyaku** · 03-11-2008

matsp: that's what I'm doing... I'm working with a few data, with huge arrays for it (so most of the array elements are actually empty), and when sweeping through the whole array, I check for the end index.
How do I get a copy of the stack trace in BCB?

The library I use, and it gets bad is the vcl50.bpl. That's practically a basic library, isn't it? So it shouldn't do anything wrong. Or is it possible, that I have modified it some way, I inserted something into it?

laserlight: yes, the contrast was right. I allocate them staticly. I don't really think I use the at() function. Maybe something similar. What does "at()" do BTW?

**laserlight** · 03-11-2008

Maybe something similar. What does "at()" do BTW?

Throw a std::out_of_range exception if the index passed is invalid.

**Hankyaku** · 03-11-2008

Can you give me a sample code?

**matsp** · 03-11-2008

Code:

vector<int> array(10);

for(i = 0; i < 11; i++)
   array.at(i) = i;

This will throw the exception when i is 10.

--
Mats

**laserlight** · 03-11-2008

Can you give me a sample code?

A really basic compilable example would be:

Code:

#include <iostream>
#include <vector>
#include <stdexcept>

int main()
{
    std::vector<int> numbers;
    try
    {
        std::cout << numbers.at(0) << std::endl;
    }
    catch (const std::out_of_range& e)
    {
        std::cout << e.what() << std::endl;
    }
}

**Hankyaku** · 03-11-2008

Thx

BCB used to give me execeptions when I produce such overflows without at(). But I'll insert something like this into the loops, and see what happens.

**Hankyaku** · 03-11-2008

One more thing I noticed is that I use several string arrays. Can they cause problems? (Even though the array is static, but the memory being used by many many string may be quite big.)

**matsp** · 03-11-2008

Originally Posted by Hankyaku

One more thing I noticed is that I use several string arrays. Can they cause problems? (Even though the array is static, but the memory being used by many many string may be quite big.)

If the string itself is a char array or char pointer, then it's certainly a potential candidate. If it's using std::string class, then it should be OK.

--
Mats

**Hankyaku** · 03-11-2008

Since I use C++Builder 5.0 I use AnsiString types.
And the real thing I'm wondering on is that since now, I never had any problems with the code I wrote, I mean the exceptions or overflows. Whenever a problem arouse, the compiler always told me that there's a problem, or during debug they appeared. A clear nice crash, at a definite place. But now? These programs are less complex than those I wrote earlier. Still, it's a mess now. One of them actually is just refreshing an older code, and making some additions. I'm having a similar problem at that one. (That's why I was thinking that a new installed component could've caused the trouble.)