netdev - Re: [PATCH net-next] efx: convert to new udp

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200721124811.3fb63afe@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 21 Jul 2020 12:48:11 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Edward Cree <ecree@...arflare.com>
Cc:     <davem@...emloft.net>, <netdev@...r.kernel.org>,
        <linux-net-drivers@...arflare.com>, <mhabets@...arflare.com>,
        <mslattery@...arflare.com>
Subject: Re: [PATCH net-next] efx: convert to new udp_tunnel infrastructure

On Tue, 21 Jul 2020 13:05:39 +0100 Edward Cree wrote:
> On 20/07/2020 18:21, Jakub Kicinski wrote:
> > On Mon, 20 Jul 2020 12:45:54 +0100 Edward Cree wrote:  
> >> I think I'd prefer to keep the switch() that explicitlychecks
> >>  for UDP_TUNNEL_TYPE_GENEVE; even though the infrastructure
> >>  makes sure it won't ever not be, I'd still feel more comfortable
> >>  that way.  But it's up to you.  
> > 
> > To me the motivation of expressing capabilities is for the core 
> > to be able to do the necessary checking (and make more intelligent
> > decisions). All the drivers I've converted make the assumption they
> > won't see tunnel types they don't support.  
> 
> Like I say, up to you.  It's not how I'd write it but if that's how
>  you're doing all the drivers then consistency is probably good.

I'll put a WARN() there, as a sign of "this can never happen".

> >> Could we not keep a 'valid'/'used' flag in the table, used in
> >>  roughly the same way we were checking count != 0?  
> > 
> > How about we do the !port check in efx_ef10_udp_tnl_has_port()?
> > 
> > Per-entry valid / used flag seems a little wasteful.
> > 
> > Another option is to have a reserved tunnel type for invalid / unused.  
> 
> Reserved tunnel type seems best to me.  (sfc generally uses all-ones
>  values for reserved, so this would be 0xffff.)

I'll do

#define TUNNEL_ENCAP_UDP_PORT_ENTRY_INVALID 0xffff

Can I add that in mcdi_pcol.h or better next to struct efx_udp_tunnel?

mcdi_pcol.h looks generated.