How to Avoid Structure Padding in C
Generally, when an object of some structure type is declared, some contiguous memory will be allocated in block form, which will be allocated to structure members. To understand structure padding we need to get an idea of how memory is allocated to structure variable.
Syntax Example
struct example
{
char x;
char y;
int z;
};
Memory Allocation in Structure
If we try to calculate the size, then we say it is 6 bytes why because we have two char values and one int value which is 4 bytes, and the character is 1 byte each, so finally, the memory allocated is 6 bytes, but we are wrong it is not 6 bytes.
This is because there is a concept called "Structure padding."
Structure padding
The basic thing is processor does not read one byte at a time from memory. It reads one word at a time.
For example, if we have a 32-bit processor, it can access 4 bytes at a time means 4 bytes at a time. A 64-bit processor can access 8 bytes simultaneously, meaning the word size is 8.
The memory looks like this.

Problem
Well, in one CPU cycle, one word can be accessed, if we consider a 32-bit processor here,4 bytes can be accessed means character x, character y, and two bytes of int can be accessed here at a time. Here, we have no problem with char a and char b. Still, whenever we want the value stored in "z," we need two cycles to execute it completely, whereas, in the first cycle, only the first two parts are accessed. In the second cycle, the last two bytes are executed means here simply CPU cycles are wasted unnecessarily. Because of this problem, the concept of padding came into the picture.
Structure padding
To align the facts in reminiscence, one or extra empty bytes (addresses) are inserted (or left empty) among reminiscence addresses which might be allotted for different shape contributors even as reminiscence allocation. This idea is known as shape padding.
The architecture of a laptop processor is any such manner that it could examine one word (four-byte in 32-bit processor) from reminiscence at a time.
To employ this gain of processor, facts are usually aligned as a four-byte package deal which ends up in empty insert addresses among different member addresses.
Because of this shape padding idea in C, the shape's length is usually no longer identified as what we think.
We can save the number of cycles by using the concept called "padding".
Here we create an empty room or space as the variable "z" is overlapping between two CPU cycles. We add space.
It adds 2 bytes empty bytes.
Total=1 byte+1 byte+2 bytes+4 bytes=8 bytes.
Memory Allocation Example
If we change the order like char, int and char then it occupies12 bytes. Check the below image for the memory allocation idea.

Structure Padding In C
#include<stdio.h>
struct example
{
char x;
char y;
int z;
};
int main()
{
struct example e;
printf("The size is %d",sizeof(e));
}
Output:

Here the allocation will be like this. The space is added automatically. Here 4 bytes will be accessed.

First CPU cycle:
char x will be accessed
Second CPU cycle:
int y will be accessed
Third CPU cycle:
char z will be accessed
So, the total size will be 12 bytes for 32-bit architecture.
Pragma
When we use structure padding, the CPU utilization is good, and we don't waste any extra cycles, but memory is wasted here. So, to avoid this structure padding and save memory or to stick with original memory, we need to use "pragma".
#include <stdio.h>
// Here, simply, we are forcing the compiler to use 1-byte packing
#pragma pack(1)
struct example {
char x;
char y;
int z;
};
int main()
{
struct example e;
printf("The size is %d",sizeof(e));
}
Output:

Here the size is 6 bytes only instead of 8 bytes as we used pragma concept.
Advantages of Structure Padding
Padding will increase the processor's overall performance on the memory penalty, in shape, or union records contributors aligned according to the dimensions of the best bytes member to save you the loss or penalty of overall performance.
Disadvantages of structure padding
1. It is slightly inefficient
2. The space will be wasted here. Sometimes memory usage plays a key role and here it is used unnecessarily.
3. It won't be binary, and wire compatibility is more.