Why new[] adds 4 bytes to the returned address

[JeanDean]

MyClass *obj = new MyClass[3];

Even though address returned by new [] is 0x00365f50,
but obj gets automatically updated by 0x00365f54.
Why and who adds these addtional 4 bytes to the address?

[Victor Bazarov]

JeanDean wrote:

MyClass *obj = new MyClass[3];

Even though address returned by new [] is 0x00365f50,

How do you know that?

but obj gets automatically updated by 0x00365f54.
Why and who adds these addtional 4 bytes to the address?

Speculation: the 4 bytes contain some kind of control block
needed for the implementation to successfully deallocate the
memory (could be the size of the array you requested)...

Have you tried to see what those 4 bytes contain?

V
--
Please remove capital 'A's when replying by e-mail
I do not respond to top-posted replies, please don't ask

[JeanDean]

On Feb 28, 2:57 pm, "Victor Bazarov" <[e-mail]:

JeanDean wrote:

MyClass *obj = new MyClass[3];

Even though address returned by new [] is 0x00365f50,

How do you know that?

but obj gets automatically updated by 0x00365f54.
Why and who adds these addtional 4 bytes to the address?

Speculation: the 4 bytes contain some kind of control block
needed for the implementation to successfully deallocate the
memory (could be the size of the array you requested)...

Have you tried to see what those 4 bytes contain?

V
--
Please remove capital 'A's when replying by e-mail
I do not respond to top-posted replies, please don't ask

Actually, it is storing number of elements of the array:

eg. code outputs 0000000B

///////////////////////
MyClass* obj = new MyClass[11];
char* ptr = (char*)obj;
ptr = (char*)obj - sizeof(int);
printf("%p",*ptr);
/////////////////////

But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

[Ian Collins]

JeanDean wrote:

>
But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

Where's the problem? It's an implantation issue, which you never have
to worry about.

--
Ian Collins.

[Ian Collins]

Ian Collins wrote:

JeanDean wrote:

>>But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

Where's the problem? It's an implantation issue, which you never have
to worry about.

Implementation issue!

--
Ian Collins.

[JeanDean]

On Feb 28, 3:36 pm, Ian Collins <[e-mail]:

Ian Collins wrote:

JeanDean wrote:

>But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

Where's the problem? It's an implantation issue, which you never have
to worry about.

Implementation issue!

--
Ian Collins.

The problem is :

I am using overloaded new for getting the memory from the big chunk of
memory.
Now , when I use overloaded new[], it is assigning addr + some bytes
to the obj pointer.

MyClass *obj = new(POOL_TYPE) MyClass[3];

where new[] is overloaded like

void* operator new[](unsigned int sz, int POOL) {
..
..
..
MyMalloc (sz, POOL);
..
}
Even though address returned by new[] was 0x0074, but 0x0078 gets
assigned to obj in case of windows and
0x007c gets assigned in case of g++.

While calling overloaded delete :

operator delete[] (ptr , POOL);

I need to pass the ptr which was returned by MyMalloc (), not the one
updted with some additional bytes.

This is because during free I need to deal with the corresponding
pointers by myself.

So, I need to know what is the standard being followed for this
reserved memory in case of new[].

[Ian Collins]

JeanDean wrote:

On Feb 28, 3:36 pm, Ian Collins <[e-mail]:

>>Ian Collins wrote:

>>>JeanDean wrote:

>>>>But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

>>>Where's the problem? It's an implantation issue, which you never have
to worry about.

Implementation issue!

*Please trim signatures*

>
The problem is :

I am using overloaded new for getting the memory from the big chunk of
memory.
Now , when I use overloaded new[], it is assigning addr + some bytes
to the obj pointer.

MyClass *obj = new(POOL_TYPE) MyClass[3];

where new[] is overloaded like

void* operator new[](unsigned int sz, int POOL) {
..
..
..
MyMalloc (sz, POOL);
..
}
Even though address returned by new[] was 0x0074, but 0x0078 gets
assigned to obj in case of windows and
0x007c gets assigned in case of g++.

Returned by which operator new[]? If you are providing both new and
delete, how the memory is managed is down to you. You will have save
the number of items in the array somewhere in order to deallocate it.

While calling overloaded delete :

operator delete[] (ptr , POOL);

I need to pass the ptr which was returned by MyMalloc (), not the one
updted with some additional bytes.

The update will be done by the built in new/delete to facilitate its
housekeeping.

This is because during free I need to deal with the corresponding
pointers by myself.

Pointers that you returned?

So, I need to know what is the standard being followed for this
reserved memory in case of new[].

There isn't one, it's an implementation detail. Provided you don't mix
default and overloaded new/delete, you are sweet.

--
Ian Collins.

[JeanDean]

On Feb 28, 4:00 pm, Ian Collins <[e-mail]:

JeanDean wrote:

On Feb 28, 3:36 pm, Ian Collins <[e-mail]:

>Ian Collins wrote:

>>JeanDean wrote:

>>>But the problem is , in VC++ it reserves 4 Bytes but if I compile same
code in g++ it reserves 8 bytes for the same.
I am not able to understand what is the standard for this additional
memory reservation?

>>Where's the problem? It's an implantation issue, which you never have
to worry about.

>Implementation issue!

*Please trim signatures*

The problem is :

I am using overloaded new for getting the memory from the big chunk of
memory.
Now , when I use overloaded new[], it is assigning addr + some bytes
to the obj pointer.

MyClass *obj = new(POOL_TYPE) MyClass[3];

where new[] is overloaded like

void* operator new[](unsigned int sz, int POOL) {
..
..
..
MyMalloc (sz, POOL);
..
}
Even though address returned by new[] was 0x0074, but 0x0078 gets
assigned to obj in case of windows and
0x007c gets assigned in case of g++.

Returned by which operator new[]? If you are providing both new and
delete, how the memory is managed is down to you. You will have save
the number of items in the array somewhere in order to deallocate it.

While calling overloaded delete :

operator delete[] (ptr , POOL);

I need to pass the ptr which was returned by MyMalloc (), not the one
updted with some additional bytes.

The update will be done by the built in new/delete to facilitate its
housekeeping.

This is because during free I need to deal with the corresponding
pointers by myself.

Pointers that you returned?

So, I need to know what is the standard being followed for this
reserved memory in case of new[].

There isn't one, it's an implementation detail. Provided you don't mix
default and overloaded new/delete, you are sweet.

--
Ian Collins.

Thanks for the prompt replies.

[Earl Purple]

On 28 Feb, 07:10, "JeanDean" <[e-mail]:

On Feb 28, 4:00 pm, Ian Collins <[e-mail]:

JeanDean wrote:

On Feb 28, 3:36 pm, Ian Collins <[e-mail]:

>>Ian Collins wrote:

>>>JeanDean wrote:

>>>>But the problem is , in VC++ it reserves 4 Bytes but if I compile same
>>>>code in g++ it reserves 8 bytes for the same.
>>>>I am not able to understand what is the standard for this additional
>>>>memory reservation?

>>>Where's the problem? It's an implantation issue, which you never have
>>>to worry about.

>>Implementation issue!

*Please trim signatures*

The problem is :

I am using overloaded new for getting the memory from the big chunk of
memory.
Now , when I use overloaded new[], it is assigning addr + some bytes
to the obj pointer.

MyClass *obj = new(POOL_TYPE) MyClass[3];

where new[] is overloaded like

void* operator new[](unsigned int sz, int POOL) {
..
..
..
MyMalloc (sz, POOL);
..
}
Even though address returned by new[] was 0x0074, but 0x0078 gets
assigned to obj in case of windows and
0x007c gets assigned in case of g++.

Returned by which operator new[]? If you are providing both new and
delete, how the memory is managed is down to you. You will have save
the number of items in the array somewhere in order to deallocate it.

While calling overloaded delete :

operator delete[] (ptr , POOL);

I need to pass the ptr which was returned by MyMalloc (), not the one
updted with some additional bytes.

The update will be done by the built in new/delete to facilitate its
housekeeping.

This is because during free I need to deal with the corresponding
pointers by myself.

Pointers that you returned?

So, I need to know what is the standard being followed for this
reserved memory in case of new[].

There isn't one, it's an implementation detail. Provided you don't mix
default and overloaded new/delete, you are sweet.

--
Ian Collins.

Thanks for the prompt replies.- Hide quoted text -

- Show quoted text -

You might consider using std::vector anyway. If you really don't want
your memory initialised then you might use boost::scoped_array or
shared_array.

[red floyd]

JeanDean wrote:

MyClass *obj = new MyClass[3];

Even though address returned by new [] is 0x00365f50,
but obj gets automatically updated by 0x00365f54.
Why and who adds these addtional 4 bytes to the address?

Please don't multi-post. You've also posted this to gnu.g++.help and to
comp.lang.c++.moderated

[Clark Cox]

On 2007-02-28 08:06:27 -0800, red floyd <no*****@here.dudesaid:

JeanDean wrote:

>MyClass *obj = new MyClass[3];

Even though address returned by new [] is 0x00365f50,
but obj gets automatically updated by 0x00365f54.
Why and who adds these addtional 4 bytes to the address?

Please don't multi-post. You've also posted this to gnu.g++.help and to
comp.lang.c++.moderated

Just as an FYI to the OP "multi-posting" (aka "double-posting") is
different than "cross-posting". The former is never appropriate, and
the latter is just slightly more approprate.

http://www.blakjak.demon.co.uk/mul_crss.htm