Re: [PATCH] netfilter: xt_HMARK: endian bugs

On Mon, 2012-05-14 at 19:51 +0200, Hans Schillstrom wrote:

On Monday 14 May 2012 18:24:34 Eric Dumazet wrote:

quoted

On Mon, 2012-05-14 at 18:09 +0200, Hans Schillstrom wrote:

quoted

This context can contain both le & be machines,
so at least in hmark it make sense

Before jhash() and its shuffle ? What do you mean ?

I want that a Big endian machine should produce the same
hash value independent of flow direction as a Little endian.

So one machine can be both le and be ? at the same time ?

OK, I missed ntohl() before calling jhash_3words()

Correct me if I'm wrong here (have no big endian machine available for test)
jhash_3words() and __jhash_final() seems to be "endian" safe.

So by doing the expensive ntohl on addresses and ports into jhash_3words()
it will produce the same value on both be and le.

And what is the purpose of the jhash output ? Is is sent to other
machines on the network, or only localy used ?

That's why I want to have the ntohs() / ntohl() when comparing.

If xt_HMARK depends on a particular bit ordering to jhash() input, then
something is really wrong. I mean it.

jhash() primary purpose it to shuffle input.

We use (__force u32) everywhere in network tree to avoid sparse
warnings. Please grep for them.