Primitive C type for fields in Ethernet frame

Question

A char in C is not always 8 bits length as it is platform dependent. Its length using bits is given by CHAR_BITS value.

Howewer, definition for standard network protocol headers is done using a Byte as its length is 8 bits. For instance, ethertype field in 802.3 frame header has 2 bytes, each one has 8 bits.

I wonder, when char type in machine is 16 bits, and any type should be addressable using this char length, how is the memory content filled with a frame when it arrives to the machine?. Which primitive C type should be used to represent a ethertype in frame, in machines where char type length value is larger than 8 bits?

@EugeneSh. if char type is larger than 8 bits, wouldn't help. — Antti Haapala -- Слава Україні, Jul 20 '17 at 14:07
It would be a ridiculous amount of bit shifting and masking there. Or you could hope that a bitfield would help you... but then again, how they're laid in memory is implementation-defined. That's why `CHAR_BIT != 8` isn't popular in generic computers... — Antti Haapala -- Слава Україні, Jul 20 '17 at 14:08
@AnttiHaapala Right. I assumed `uint8_t` is mandatory.. Apparently it's [implementation defined](http://port70.net/~nsz/c/c11/n1570.html#7.20.1.1). — Eugene Sh., Jul 20 '17 at 14:10
I would say that the situation where `CHAR_BIT=16` is much better than the one with `CHAR_BIT=7`... — Eugene Sh., Jul 20 '17 at 14:13
and yes, `CHAR_BIT == 7` would mean that you couldn't program in C... — Antti Haapala -- Слава Україні, Jul 20 '17 at 14:14
@AnttiHaapala Wrong link? Actually, yeah.. the minimum is `8`. So replace that with `9` :) — Eugene Sh., Jul 20 '17 at 14:15
@EugeneSh. https://stackoverflow.com/questions/44755034/md5-implementation-for-non-byte-addressable-arch — Antti Haapala -- Слава Україні, Jul 20 '17 at 14:16
MABC, if this is just a hypothetical question, then just forget that non-8-bit arches exist and just use `uint8_t`. — Antti Haapala -- Слава Україні, Jul 20 '17 at 14:18

score 0 · Answer 1 · answered Jul 22 '17 at 23:14

0

If we look at how the PIC stores uint8_t strings in its flash, which is word-oriented and the word-size depends on the PIC (it is neither 8 nor 16 bits in some PICs), the upper bits are simply zeroed.

answered Jul 22 '17 at 23:14

Fredrik Lundström

74
5

Primitive C type for fields in Ethernet frame

1 Answers1