Re: eBPF - little-endian load instructions?

[Johannes Berg <johannes@sipsolutions.net>]

On Wed, 2017-04-12 at 20:08 -0700, Alexei Starovoitov wrote:

> it's really llvm bug that i need fix. It's plain broken
> to generate what effectively is nop insn for march=bpfeb
> My only excuse that when that code was written llvm had only
> march=bpfel.
> bpfeb was added much later.

So I'm confused now. Is bpf intended to be endian-independent or not?
It sounded at first like it was, even if I have a hard time imagining
how that would even work.

> > 	#define be32_to_cpu bswap32
> > or
> > 	#define be32_to_cpu(x) (x)
> > depending on the build architecture, I guess.
> 
> yeah. that's what we should have in bpf_helpers.h

But that sounds more like it isn't.

> ntoh is enough for any networking code,
> so I guess we can live without real bswap insn.

Well, my reason for asking this is that wireless actually as a little-
endian wire protocol, unlike other network stuff :)
(Even at a bit level it's defined to transfer the LSB first, but that
doesn't really get visible at the level of the CPU that can only
address bytes.)

johannes