Thread: Going out of Array bounds and how it affects input and output

Hey guys.

I tried experimenting with deliberately putting in more elements than an array has and printing out the results. The reason im doing this is basically in anticipation that should the user do something wrong, I want to understand how to handle these scenarios.

I write this code:

Code:

#include <stdio.h>

int main(void)
{
    int a[5], b, c;
    for (b=0; b< 10&&(scanf("%d", &c)); b++)
    a[b]=c;
    for(b=0;b<15;b++)
    printf("%d ", a[b]);
return 0;
}

So if you input, say:

1, 2, 3, 4, 5, 6, 7, 8, 9, 10

It prints out:

1, 2, 3, 4, 5, 10, 7, 8, 9, 10, a large number, another large number, etc until the fifteenth last element which, interestingly, contains a 1

Q's:
1) I am printing out elements of an array that dont really exist (>5), and yet, it still prints out the same numbers I put in input. Is this because, somehow, all your inputs are "contiguous" next to the defined array?
2) If so, how come the last input (10) replaces 6 in the start of the numbers that go out of array bounds?

I guess its a silly non trivial attempt to understand the deeper workings of C

Is this because, somehow, all your inputs are "contiguous" next to the defined array?

Yes, and you can easily confirm this by also printing the memory address.

Code:

for(b=0;b<15;b++)
     printf("%d %p ", a[b], a+b);

By the way, I see this forum finally decided to add line numbering and colouring to code tags! about time!

I agree with you all that it must be prevented in the first place.

But I was told on this forum that for variable length user input, say a string, a "sufficiently large" array must be made or I think it was termed large enough "buffer" (same thing?). So thats when I wondered what if it goes over this...

Originally Posted by Vespasian

I agree with you all that it must be prevented in the first place.

But I was told on this forum that for variable length user input, say a string, a "sufficiently large" array must be made or I think it was termed large enough "buffer" (same thing?). So thats when I wondered what if it goes over this...

You need to make sure it doesn't. That is why fgets() is recomended over gets() for example, since you can set the maximum amount of characters that will be accepted. Or in case of scanf use a length in addition to the %s format specifier.

Originally Posted by Vespasian

1) I am printing out elements of an array that dont really exist (>5), and yet, it still prints out the same numbers I put in input. Is this because, somehow, all your inputs are "contiguous" next to the defined array?

Yes, as you increase the array index (even out of bounds) C works from a calculated offset from the array pointer, so they will end up being contiguous in memory... But that's not what you should be concerned about. You should be wondering what other variables or stack data you clobbered in the process.
It worked this time as an act of pure dumb luck... don't count on it ever working again.

2) If so, how come the last input (10) replaces 6 in the start of the numbers that go out of array bounds?

It doesn't... i's probably that you are clobbering your C variable with out of bounds array content. If you put in different numbers... 100, 200, etc. you would probably also discover you've clobbered your loop counter (b).

I guess its a silly non trivial attempt to understand the deeper workings of C

What you've just discovered is one of the most important things you have to know about C ...

C contains no runtime error checking whatsoever; it is entirely on your shoulders as a programmer to make sure stuff like this doesn't happen.

Just to add...

Not using gets function, there is no other way to deal with variable user input other than making a "sufficiently large" array. I understand that accessing the array out of bounds must not be done, but what about simply overfilling the array by itself, is that a bad thing? Or the programme will simply ignore the overfilled elements in an array and wont cause any side effects down the line...

Originally Posted by Subsonics

You need to make sure it doesn't. That is why fgets() is recomended over gets() for example, since you can set the maximum amount of characters that will be accepted. Or in case of scanf use a length in addition to the %s format specifier.

Okay I took this advice. So my following code is safe? BUT it cant allocate extra memory should input exceed BUFFERSIZE. So if this is correct my next step is to do what Salem says and malloc some extra chunks. Ill do this next, tell me what you think!

Code:

#include <stdio.h>
#define BUFFERSIZE 10

char buffer[BUFFERSIZE];

void getinput()
{
    fgets(buffer, BUFFERSIZE, stdin);
}

int main ()
{
    int n;
    getinput();
    printf("%s", buffer);
    return 0;
}

Looks like you understand it.

I think of it like a soldier - one step out of the array bounds, and I'm stepping into the mine field. Will I step on a mine? Who knows? Maybe a bunch of goats have grazed there, and already set off every land mine, already. Maybe the map is wrong, and there are no land mines here. Maybe there's one waiting for my very first step into the mine field.

You just can't tell.

If you want to be safe - stay out of the mine field.