Difference between u8, uint8_t, __u8 and __be8

Question

While browsing linux networking code, I came across these datatypes:

1. u8
2. uint8_t
3. __u8
4. __be8

(same things for 16, 32 and 64 bits)

Can someone please explain the difference between these datatypes and where to use which? I have seen the definitions of these datatypes but those were not clear to me.

Answer 1

uint8_t is Standard C and represents an unsigned 8-bit integral type. If you are on a system that does not have 8-bit addressable units then this will not be defined; otherwise it is probably a typedef for unsigned char.

Anything with __ in it is reserved for implementation use. This means that compiler writers and standard library writers can use those identifiers without worrying about a name clash with user code. You may see this when looking in the internals of standard library implementation.

u8 is non-standard but almost certainly means the same as uint8_t. A reason that u8 might be used is in code that was written before uint8_t was added to Standard C.

Answer 2

The last denotes big-endian (Not significant for a single byte), aka network-byte-order.

The others are all identical to each other, an unsigned type of 8 bits / 1 byte.

The above all holds for those types with 16 bits / 2 byte resp. 32 bits / 4 bytes too.

(Host byte-order on x86 for example is little-endian, and Linux only runs on 8-bit-byte machines.)

Out of the list only uint8_t is defined by the C standard (in <stdint.h), and the last two are in the implementations name-space, which is generally a bad idea.