Question

I am currently looking into socket programming in python and I am experiencing some strange behavior with socket.htons() … it appears to be flipping the bytes on every call.

I am implementing a simple ping script, as far as I am aware network byte order is big endian and my systems byte order is little endian .

If I use htons on my 16 bit checksum wireshark reports that it is incorrect , however if I just pack the checksum into a struct without using htons wireshark confirms it is correct.

This is what wireshark has captured when using htons

Checksum: 0xece4 [incorrect, should be 0xe4ec]

And here is a quick example…

>>> z = 0xFF00
>>> print z
65280
>>> z = socket.htons(z)
>>> print z
255
>>> z = socket.htons(z)
>>> print z
65280

Any thoughts on this would be greatly appreciated, hopefully its just me doing something very wrong.

Edit:

>>> print sys.byteorder
little

Answer 1

Everything looks to be working like it should. Even in your example, the 2 bytes are being swapped, then back again. If the system byte order is the same as the network, then its a no-op, but your question states that the network bytes order is Big, and the host is Little, so the swapping is what it should be doing.

In [1]: a = 0xECE4
In [2]: b = 0xE4EC

In [3]: a
Out[3]: 60644

In [4]: b
Out[4]: 58604

In [5]: socket.htons(a)
Out[5]: 58604

In [6]: socket.htons(b)
Out[6]: 60644

In [7]: socket.htons(a) == b
Out[7]: True

In [8]: socket.htons(b) == a
Out[8]: True

In [9]: hex(socket.htons(a))
Out[9]: '0xe4ec'

In [10]: hex(socket.htons(b))
Out[10]: '0xece4'

In [11]: bin(socket.htons(a))
Out[11]: '0b1110010011101100'

In [12]: bin(socket.htons(b))
Out[12]: '0b1110110011100100'

---

From the python socket reference:

socket.htons(x)
Convert 16-bit positive integers from host to network
byte order. On machines where the host byte order is the same as
network byte order, this is a no-op; otherwise, it performs a 2-byte
swap operation.