netdev - Re: [RFC PATCH 1/5] bpf: add PHYS_DEV prog type for early driver filter

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: <20160405101940.74a10ba2@redhat.com>
Date:	Tue, 5 Apr 2016 10:19:40 +0200
From:	Jesper Dangaard Brouer <brouer@...hat.com>
To:	Thomas Graf <tgraf@...g.ch>
Cc:	Brenden Blanco <bblanco@...mgrid.com>,
	Tom Herbert <tom@...bertland.com>,
	"David S. Miller" <davem@...emloft.net>,
	Linux Kernel Network Developers <netdev@...r.kernel.org>,
	Alexei Starovoitov <alexei.starovoitov@...il.com>,
	Or Gerlitz <ogerlitz@...lanox.com>,
	Daniel Borkmann <daniel@...earbox.net>,
	john fastabend <john.fastabend@...il.com>, brouer@...hat.com
Subject: Re: [RFC PATCH 1/5] bpf: add PHYS_DEV prog type for early driver
 filter


On Tue, 5 Apr 2016 00:07:14 +0200 Thomas Graf <tgraf@...g.ch> wrote:

> On 04/03/16 at 12:02am, Brenden Blanco wrote:
> > On Sat, Apr 02, 2016 at 12:39:45PM -0400, Tom Herbert wrote:  
> > > Is there a hard restriction that this could only work with physical devices?  
> > I suppose not, but there wouldn't be much use case as compared to tc
> > ingress, no? Since I was imagining that this new hook would be more
> > restricted in functionality due to operating on descriptors rather than
> > with a full skb, I tried to think of an appropriate name.
> > 
> > If you think that this hook would spread, then a better name is needed.  
> 
> The thing that comes to mind is that this prog type makes it easier to
> implement batched processing of packets in BPF which would require major
> surgery across the tc layer to make it available in cls_bpf. Batched
> processing will be beneficial for software devices as well.

+1 agreed, I would really like to see batched processing of packets in
BPF, for this packet type, designed in from the start.

-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  Author of http://www.iptv-analyzer.org
  LinkedIn: http://www.linkedin.com/in/brouer