Re: WireGuard recipes for OpenEmbedded

2017-05-14 Thread Stefan Agner
Hi Jason, On 2017-04-04 06:28, Jason A. Donenfeld wrote: > Hey Stefan, > > Very cool! Thanks for posting it. Let me know when it gets accepted, > and we can add official instructions for the package on > wireguard.io/install/ . The recipes are now in the OpenEmbedded pyro and master branch. The

Re: Installation instructions for Debian

2017-05-14 Thread Daniel Kahn Gillmor
On Sun 2017-05-14 10:44:56 +0200, Baptiste Jonglez wrote: > Note that the previous installation instructions (basically, just "sudo > apt install wireguard-tools wireguard-dkms") were valid for Debian sid. > > The new installation instructions are only valid for Debian jessie and > Debian stretch (

Re: TX counters not updated on wireguard interface

2017-05-14 Thread Jason A. Donenfeld
Hey again, I've committed a fix [1] for this which will be out in the next snapshot. I've also added a test [2] for this to the automated test suite on https://build.wireguard.io/ so that if the bug ever resurfaces, it will be caught. Before reading [1], be sure to wear protective eyewear, as your

Re: Performance of Wireguard on Infiniband 40G

2017-05-14 Thread Jason A. Donenfeld
Hey Greg, On Sun, May 14, 2017 at 12:48 PM, Greg KH wrote: > 4.9 is 6 months old, I'd be curious if 4.11 is any faster given the rate > of change in the network stack :) I imagine it might be. I think the biggest bottle neck, in any case, is still the poor algorithm in padata. Hopefully we'll ge

Re: Performance of Wireguard on Infiniband 40G

2017-05-14 Thread Greg KH
On Sun, May 14, 2017 at 11:55:52AM +0200, Baptiste Jonglez wrote: > On Sun, May 14, 2017 at 12:52:11AM +0200, Jason A. Donenfeld wrote: > > One small and unfortunate thought just occurred to me: the backporting > > to really old kernels I'm pretty sure is way less efficient than newer > > kernels o

Re: Performance of Wireguard on Infiniband 40G

2017-05-14 Thread Baptiste Jonglez
On Sun, May 14, 2017 at 12:52:11AM +0200, Jason A. Donenfeld wrote: > One small and unfortunate thought just occurred to me: the backporting > to really old kernels I'm pretty sure is way less efficient than newer > kernels on the RX, due to some missing core fast-path APIs in the old > kernels. In

Re: Installation instructions for Debian

2017-05-14 Thread Baptiste Jonglez
On Sun, May 14, 2017 at 12:58:37AM +0200, Jason A. Donenfeld wrote: > Hey Baptiste, > > Spectacular, thanks a bunch. I think I had asked about this same thing > a while back, but somehow I got confused in the ensuing mailing list > thread and eventually I forgot about it. So I'm glad you've brough