Memory allocation problem

[AlabiChin]

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}

The above program causes a crash when destroying the allocated memory
via delete. I suspect the crash problem originates at the time of
allocation (i.e., possible memory corruption at time of allocation).

I am using MS Developer Studio 6.0 (SP 6) on AMD Athlon. I get the same
results on a Pentium 4. Any suggestions or comments on what I am doing
in the code is much appreciated.

TIA
Alabi

[Jacek Dziedzic]

AlabiChin napisa³:

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
// new does not return NULL on error, it throws
// so this check is meaningless
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}

The above program causes a crash when destroying the allocated memory
via delete.
That's because you should destroy it using 'delete[]',
not 'delete'.
I suspect the crash problem originates at the time of
allocation (i.e., possible memory corruption at time of allocation).
I don't think so.

I am using MS Developer Studio 6.0 (SP 6) on AMD Athlon. I get the same
results on a Pentium 4. Any suggestions or comments on what I am doing
in the code is much appreciated.

Notice that you don't clear the allocated buffer, therefore
it contains garbage. The function strlen() looks for the
first occurrence of the '\0' terminator, which luckily
falls 'about 15 bytes' after the end of your buffer.

HTH,
- J.

[Rolf Magnus]

AlabiChin wrote:

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
This check is unnecessary. new either returns a valid pointer or (if no
memory is available) throws an exception. It never returns a null pointer.
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}
strlen expects a C style string, i.e. a null terminated character array. You
gave it uninitialized memory, which means the behavior is undefined.
The above program causes a crash when destroying the allocated memory
via delete. I suspect the crash problem originates at the time of
allocation (i.e., possible memory corruption at time of allocation).

No, it originates at the time you used strlen on it. Probably, strlen went
past the end of the array to find the null character.

[Scott McPhillips [MVP]]

AlabiChin wrote:

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}

The above program causes a crash when destroying the allocated memory
via delete. I suspect the crash problem originates at the time of
allocation (i.e., possible memory corruption at time of allocation).

I am using MS Developer Studio 6.0 (SP 6) on AMD Athlon. I get the same
results on a Pentium 4. Any suggestions or comments on what I am doing
in the code is much appreciated.

TIA
Alabi

You have an uninitialized buffer. Calling strlen with such a buffer is
likely to cause a crash right then and there. strlen does not return
the length of the buffer, it returns the length of a nul-terminated
string that has been stored in the buffer. But you do not have such a
string in the buffer.

--
Scott McPhillips [VC++ MVP]

[Nick Keighley]

AlabiChin wrote:

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified.
whilst "new" can be expected to allocate a few more bytes than
requested
(housekeeping) 15 bytes sounds like a lot.

Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
you can't do strlen on a newly allocated chunk of memory. How do you
know
"new" returns a nul terminated string?
return buffer;
}

The above program causes a crash when destroying the allocated memory
via delete.
what does your delete look like?

I suspect the crash problem originates at the time of
allocation (i.e., possible memory corruption at time of allocation).
I suspect there's something else wrong with your program. Standard
libraries
can have bugs, but always suspect your own (newly written) code first
I am using MS Developer Studio 6.0 (SP 6) on AMD Athlon. I get the same
results on a Pentium 4. Any suggestions or comments on what I am doing
in the code is much appreciated.

--
Nick Keighley

[Neelesh Bodas]

AlabiChin wrote:

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}
strlen is a "C" library function which returns the length of a "null
terminated string". In other words, in the current case, it will count
the number of characters from start of buffer till it finds a null
character. Since char is a basic data type, buffer remains
uninitialized, and thus contains a junk. It is incorrect to relie on
strlen.

The above program causes a crash when destroying the allocated memory
via delete.

This is because usage of new[] must be matched via delete[], not
delete.

[Jim Langston]

"AlabiChin" <al**********@yahoo.com> wrote in message
news:11**********************@g47g2000cwa.googlegr oups.com...

Hello,

I noticed that when I dynamically create an array of chars, the
resulting size of the allocated memory block is larger by about 15
bytes than what I specified. Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.

strlen looks for the first null character to determine the end of the
string. Since you have not initialized the array, the array contains random
information, and your length will be some arbitary number.

To see this add this line just before your length =

buffer[0] = '\0';

Now your length will be 0 because it will find the null in the first
position.

When you delete this array you use delete[]

[Viktor Prehnal]

Hmm, I studied documentation in VS 2003.NET which I consider not too
out-of-date and found out that new returns 0/NULL. If you want exception
handling you need to use set_new_handler function.
From what time new throws exception? What type of exception is it?

[Nick Keighley]

Viktor Prehnal wrote:

Hmm, I studied documentation in VS 2003.NET which I consider not too
out-of-date and found out that new returns 0/NULL. If you want exception
handling you need to use set_new_handler function.
new throws an exception when it fails
From what time new throws exception? What type of exception is it?

std::bad_alloc
--
Nick Keighley

[Rolf Magnus]

Viktor Prehnal wrote:

Hmm, I studied documentation in VS 2003.NET which I consider not too
out-of-date and found out that new returns 0/NULL. If you want exception
handling you need to use set_new_handler function.
From what time new throws exception?
In Standard C++ (which is existing since 1998), this has always been the
case.
What type of exception is it?

std::bad_alloc

[Robbie Hatley]

AlabiChin wrote:

I noticed that when I dynamically create an array of chars
Why would you want to do that?
the resulting size of the allocated memory block
is larger by about 15 bytes than what I specified.
Why would you care?
Here is example code:

char *createTextBuffer(char *buffer, int length)
{
buffer = new char[length];

if (buffer == NULL) {
return NULL;
}
length = strlen(buffer); // Current buffer shows larger size.
return buffer;
}

Yuck!
Here's my version of your code:

{ // Begin block; stuff declared in here dies at end.

// Make a buffer to hold some text:
std::string TextBuffer = "This is some text.";

// Find the length of the text in the buffer, which is
// NOT the same thing as the memory used by the buffer:
int Length = TextBuffer.size();

// (do stuff with TextBuffer and Size)

} // End block; TextBuffer and Size are deallocated here.

In my version, allocation and deallocation are always done
for you automatically. There is no "memory allocation
problem". If you absolutely need a C string version of
TextBuffer (say, to pass to some C std lib function), then
use the .c_str() member function of TextBuffer:

int CSize = strlen(TextBuffer.c_str());

I think you'll find that Size and CSize are then both
equal to 18 (the length of "This is some text.").
The above program causes a crash.

Of course. For one thing, it's full of errors, as others
here have pointed out. But more importantly, it's C, not
C++. A zero-terminated array of char is a very antiquated
and crude way of handling strings. C++ std::string is much
better. It's easier, safer, and much more versatile, with
loads of cool member functions such as "size" and "find"
and "substr". Time to upgrade your approach, I think.

--
Cheers!
Robbie Hatley
Tustin, CA, USA
email: lonewolfintj at pacbell dot net
web: home dot pacbell dot net slant earnur slant