Data Alignment.

[vineet]

Hi,
To make data aligned to memory boundry, compiler pedded some byte. I
wanted to know if these bytes are initialized? or compiler pedded some
un initialized byte.

Thanks,

 

[Salt_Peter]

Hi,
To make data aligned to memory boundry, compiler pedded some byte. I
wanted to know if these bytes are initialized? or compiler pedded some
un initialized byte.

Thanks,

The padding, if any, is not initialized.
In fact neither is the data unless you have done so.
The question is: why would that matter?
Your code should be completely transparent to padding, regardless of
platform.

 

[nobyjos]

Hi,
To make data aligned to memory boundry, compiler pedded some byte. I
wanted to know if these bytes are initialized? or compiler pedded some
un initialized byte.

Thanks,

I think it is depends upon which part of the memory data sits - stack,
heap, initialized DS, Uninitialized DS


- nOby jOse

 

[Fei Liu]

I think it is depends upon which part of the memory data sits - stack,
heap, initialized DS, Uninitialized DS


- nOby jOse

Also some compilers (e.g. intel) has compiler flags such as
'-no-bss-init' to allow finer control of memory initialization. Really,
the answer is platform-compiler dependent and has not much to do with C++.

 

[Maxim Yegorushkin]

The padding, if any, is not initialized.
In fact neither is the data unless you have done so.

Unless it is POD with static storage duration, in which case it is
guaranteed to be zero-initialised.

 

[James Kanze]

Unless it is POD with static storage duration, in which case
it is guaranteed to be zero-initialised.

The members, for types where zero initialization makes sense
(not references, for example). And this is guaranteed to occur
even within types with constructors, etc.; zero initialization
takes place before any dynamic initialization. But the original
question concerned padding, and that's not guaranteed (for the
obvious reason that what is meant by zero initialization depends
on the type, and padding has no type).