Re: [PATCH 1/6] conf, fwd: Make overall forwarding mode local to conf path

[Stefano Brivio <sbrivio@redhat.com>]

On Tue, 10 Mar 2026 15:16:00 +1100
David Gibson <david@gibson.dropbear.id.au> wrote:

> The 'mode' field of struct fwd_ports records the overall forwarding mode.
> Now that runtime forwarding decisions are made based on the forwarding
> table, this is almost unused outside conf().
> 
> The only exception is the auto-port scanning code, which uses it to
> determine if a port scan is necessary.  We can instead derive that from the
> forwarding table itself by checking if there are any entries with the
> FWD_SCAN flag.
> 
> Once that's done, make the mode purely local to conf().  While we're there
> rename the constants to FWD_MODE_* to avoid confusion with the forwarding
> rule flag bits, which are also prefixed with FWD_.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> ---
>  conf.c | 82 ++++++++++++++++++++++++++++++++++++----------------------
>  fwd.c  | 27 ++++++++++++++-----
>  fwd.h  | 10 -------
>  3 files changed, 72 insertions(+), 47 deletions(-)
> 
> diff --git a/conf.c b/conf.c
> index 11d84536..c436b88e 100644
> --- a/conf.c
> +++ b/conf.c
> @@ -199,15 +199,27 @@ static void conf_ports_range_except(const struct ctx *c, char optname,
>  	}
>  }
>  
> +/**
> + * enum fwd_mode - Overall forwarding mode for a direction and protocol
> + */

Nit: I'm actually trying to document enums in the matching kerneldoc
style when it comes to it, see for example enum udp_iov_idx.

I don't think it's so much of a problem to omit it here, even though
"SPEC" and "ALL" might not be that obvious. If you respin, maybe:

 * @FWD_MODE_UNSET	Initial value, not parsed/configured yet
 * @FWD_MODE_SPEC	Forward specified ports
 * @FWD_MODE_NONE	No forwarded ports
 * @FWD_MODE_AUTO	Automatic detection and forwarding based on bound ports
 * @FWD_MODE_ALL	Bind all free ports

?

-- 
Stefano