Re: [Ksummit-discuss] [TECH TALK] WireGuard: Next-Generation Secure Kernel Network Tunnel

Re: [Ksummit-discuss] [TECH TALK] WireGuard: Next-Generation Secure Kernel Network Tunnel

[Jason A. Donenfeld]

Confusingly, this talk was actually accepted for the main refereed
track at Plumbers, which probably makes more sense anyway for it. So I
should probably retract this submission.

However, the more important and relevant talk -- the one on Zinc -- is
really quite appropriate for the kernel track and is sure to spur a
lot of needed discussion. So please focus attention on that instead.

Thanks,
Jason

Re: [Ksummit-discuss] [TECH TALK] WireGuard: Next-Generation Secure Kernel Network Tunnel

[James Bottomley]

On Mon, 2018-09-24 at 02:12 +0200, Jason A. Donenfeld wrote:
> Confusingly, this talk was actually accepted for the main refereed
> track at Plumbers, which probably makes more sense anyway for it. So
> I should probably retract this submission.
> 
> However, the more important and relevant talk -- the one on Zinc --
> is really quite appropriate for the kernel track and is sure to spur
> a lot of needed discussion. So please focus attention on that
> instead.

Just so you know, this is the only email on the ksummit discuss list
I've seen from you ... there isn't one proposing either zinc or
wireguard as a tech topic.

The list archive also only sees this email from you:

https://lists.linuxfoundation.org/pipermail/ksummit-discuss/

so there may be an issue with your outgoing server.

James

Re: [Ksummit-discuss] [TECH TALK] WireGuard: Next-Generation Secure Kernel Network Tunnel

[Jason A. Donenfeld]

On Mon, Sep 24, 2018 at 2:21 AM James Bottomley
<James.Bottomley@hansenpartnership.com> wrote:
>
> On Mon, 2018-09-24 at 02:12 +0200, Jason A. Donenfeld wrote:
> > Confusingly, this talk was actually accepted for the main refereed
> > track at Plumbers, which probably makes more sense anyway for it. So
> > I should probably retract this submission.
> >
> > However, the more important and relevant talk -- the one on Zinc --
> > is really quite appropriate for the kernel track and is sure to spur
> > a lot of needed discussion. So please focus attention on that
> > instead.
>
> Just so you know, this is the only email on the ksummit discuss list
> I've seen from you ... there isn't one proposing either zinc or
> wireguard as a tech topic.
>
> The list archive also only sees this email from you:
>
> https://lists.linuxfoundation.org/pipermail/ksummit-discuss/
>
> so there may be an issue with your outgoing server.

Yikes! However, I'm certain it's not my outgoing server. My mail
server logs show that lists.linuxfoundation.org replied with the
following IDs for each of the two messages:

- 250 2.0.0 Ok: queued as AC122174
- 250 2.0.0 Ok: queued as 8EF1AF8

The one I sent earlier today following up gave a similar response:

- 250 2.0.0 Ok: queued as 63A72180

I'm adding Konstantin to the CC to see what's up.

[Ksummit-discuss] Fwd: [TECH TALK] Zinc: Minimal Light-weight Kernel Cryptography API

[Jason A. Donenfeld]

Following up from the other thread, it appears that the below message
was rejected silently by linuxfoundation.org's mailing list server. It
should have been accepted as either:

- 250 2.0.0 Ok: queued as AC122174
- 250 2.0.0 Ok: queued as 8EF1AF8

Hopefully the below submission does make it this time.

(Thanks to James for pointing out that these never made it the first
time around.)



---------- Forwarded message ---------
From: Jason A. Donenfeld <Jason@zx2c4.com>
Date: Fri, Sep 21, 2018 at 6:43 AM
Subject: [TECH TALK] Zinc: Minimal Light-weight Kernel Cryptography API
To: <ksummit-discuss@lists.linuxfoundation.org>


Zinc is a new minimal cryptography API for the kernel that is in the
process of being upstreamed. Rather than providing an abstracted
framework, Zinc provides simple functions. This talk will address the
design considerations of the new API, its approach to implementation
choice and fuzzing, and touch on formally verified cryptography
implementations. We'll explore the difference between, on one hand,
reference code for a particular algorithm, and on the other hand, the
present crypto API with its large and complicated abstractions, and
how Zinc fits into the middle of these extremes. We'll also examine
issues relating to using SIMD from kernel space and the costs
associated with it, and what we can do about it.