Re: [nft PATCH 13/28] Define string-based data types as Big Endian

[Phil Sutter <phil@nwl.cc>]

On Wed, Oct 29, 2025 at 07:22:42PM +0100, Pablo Neira Ayuso wrote:
> On Thu, Oct 23, 2025 at 06:14:02PM +0200, Phil Sutter wrote:
> > Doesn't quite matter internally, but libnftnl should not attempt to
> > convert strings from host byte order when printing.
> > 
> > Fib expression byte order changes with NFT_FIB_RESULT_OIFNAME to Big
> > Endian.
> > 
> > Signed-off-by: Phil Sutter <phil@nwl.cc>
> > ---
> >  src/ct.c                  |  2 +-
> >  src/datatype.c            | 10 +++++-----
> >  src/evaluate.c            | 18 +++++++++---------
> >  src/fib.c                 |  5 +++--
> >  src/intervals.c           |  5 -----
> >  src/json.c                |  2 +-
> >  src/meta.c                | 16 ++++++++--------
> >  src/mnl.c                 |  2 +-
> >  src/netlink.c             | 12 +++++-------
> >  src/netlink_delinearize.c | 14 +++++++-------
> >  src/osf.c                 |  3 +--
> >  src/parser_bison.y        | 10 +++++-----
> >  src/parser_json.c         |  4 ++--
> >  src/segtree.c             | 10 +++++-----
> >  14 files changed, 53 insertions(+), 60 deletions(-)
> > 
> > diff --git a/src/ct.c b/src/ct.c
> > index 4edbc0fc2997f..e9333c79dfd42 100644
> > --- a/src/ct.c
> > +++ b/src/ct.c
> > @@ -273,7 +273,7 @@ const struct ct_template ct_templates[__NFT_CT_MAX] = {
> >  					      BYTEORDER_HOST_ENDIAN,
> >  					      4 * BITS_PER_BYTE),
> >  	[NFT_CT_HELPER]		= CT_TEMPLATE("helper",	    &string_type,
> > -					      BYTEORDER_HOST_ENDIAN,
> > +					      BYTEORDER_BIG_ENDIAN,
> 
> No, this is not big endian, this is confusing.

I agree, it is confusing. Maybe we could use BYTEORDER_INVALID for it?
No idea if that would work or not, though. Since string values are
always in the same ordering despite host's byteorder,
BYTEORDER_HOST_ENDIAN is certainly not correct, either.

Cheers, Phil