Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-29 Thread Jesper Dangaard Brouer
On Wed, 27 Sep 2017 10:32:36 -0700 Alexei Starovoitov wrote: > On Wed, Sep 27, 2017 at 04:54:57PM +0200, Jesper Dangaard Brouer wrote: > > On Wed, 27 Sep 2017 06:35:40 -0700 > > John Fastabend wrote: > > > > > On 09/27/2017 02:26 AM,

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread Waskiewicz Jr, Peter
On 9/28/17 2:23 PM, John Fastabend wrote: > [...] > >> I'm pretty sure I misunderstood what you were going after with >> XDP_REDIRECT reserving the headroom. Our use case (patches coming in a >> few weeks) will populate the headroom coming out of the driver to XDP, >> and then once the XDP

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread Daniel Borkmann
On 09/28/2017 10:52 PM, Waskiewicz Jr, Peter wrote: On 9/28/17 12:59 PM, Andy Gospodarek wrote: On Thu, Sep 28, 2017 at 1:59 AM, Waskiewicz Jr, Peter wrote: On 9/26/17 10:21 AM, Andy Gospodarek wrote: On Mon, Sep 25, 2017 at 08:50:28PM +0200, Daniel Borkmann

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread John Fastabend
[...] > I'm pretty sure I misunderstood what you were going after with > XDP_REDIRECT reserving the headroom. Our use case (patches coming in a > few weeks) will populate the headroom coming out of the driver to XDP, > and then once the XDP program extracts whatever hints it wants via >

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread Waskiewicz Jr, Peter
On 9/28/17 12:59 PM, Andy Gospodarek wrote: > On Thu, Sep 28, 2017 at 1:59 AM, Waskiewicz Jr, Peter > wrote: >> On 9/26/17 10:21 AM, Andy Gospodarek wrote: >>> On Mon, Sep 25, 2017 at 08:50:28PM +0200, Daniel Borkmann wrote: On 09/25/2017 08:10 PM, Andy

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread Andy Gospodarek
On Thu, Sep 28, 2017 at 1:59 AM, Waskiewicz Jr, Peter wrote: > On 9/26/17 10:21 AM, Andy Gospodarek wrote: >> On Mon, Sep 25, 2017 at 08:50:28PM +0200, Daniel Borkmann wrote: >>> On 09/25/2017 08:10 PM, Andy Gospodarek wrote: >>> [...] First, thanks for this

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-28 Thread Waskiewicz Jr, Peter
On 9/26/17 10:21 AM, Andy Gospodarek wrote: > On Mon, Sep 25, 2017 at 08:50:28PM +0200, Daniel Borkmann wrote: >> On 09/25/2017 08:10 PM, Andy Gospodarek wrote: >> [...] >>> First, thanks for this detailed description. It was helpful to read >>> along with the patches. >>> >>> My only concern

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-27 Thread Alexei Starovoitov
On Wed, Sep 27, 2017 at 04:54:57PM +0200, Jesper Dangaard Brouer wrote: > On Wed, 27 Sep 2017 06:35:40 -0700 > John Fastabend wrote: > > > On 09/27/2017 02:26 AM, Jesper Dangaard Brouer wrote: > > > On Tue, 26 Sep 2017 21:58:53 +0200 > > > Daniel Borkmann

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-27 Thread Jesper Dangaard Brouer
On Wed, 27 Sep 2017 06:35:40 -0700 John Fastabend wrote: > On 09/27/2017 02:26 AM, Jesper Dangaard Brouer wrote: > > On Tue, 26 Sep 2017 21:58:53 +0200 > > Daniel Borkmann wrote: > > > >> On 09/26/2017 09:13 PM, Jesper Dangaard Brouer wrote: >

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-27 Thread John Fastabend
On 09/27/2017 02:26 AM, Jesper Dangaard Brouer wrote: > On Tue, 26 Sep 2017 21:58:53 +0200 > Daniel Borkmann wrote: > >> On 09/26/2017 09:13 PM, Jesper Dangaard Brouer wrote: >> [...] >>> I'm currently implementing a cpumap type, that transfers raw XDP frames >>> to another

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-27 Thread Jesper Dangaard Brouer
On Tue, 26 Sep 2017 21:58:53 +0200 Daniel Borkmann wrote: > On 09/26/2017 09:13 PM, Jesper Dangaard Brouer wrote: > [...] > > I'm currently implementing a cpumap type, that transfers raw XDP frames > > to another CPU, and the SKB is allocated on the remote CPU. (It > >

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-26 Thread Daniel Borkmann
On 09/26/2017 09:13 PM, Jesper Dangaard Brouer wrote: [...] I'm currently implementing a cpumap type, that transfers raw XDP frames to another CPU, and the SKB is allocated on the remote CPU. (It actually works extremely well). Meaning you let all the XDP_PASS packets get processed on a

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-26 Thread Jesper Dangaard Brouer
On Mon, 25 Sep 2017 02:25:51 +0200 Daniel Borkmann wrote: > This work enables generic transfer of metadata from XDP into skb. The > basic idea is that we can make use of the fact that the resulting skb > must be linear and already comes with a larger headroom for supporting

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-26 Thread Andy Gospodarek
On Mon, Sep 25, 2017 at 08:50:28PM +0200, Daniel Borkmann wrote: > On 09/25/2017 08:10 PM, Andy Gospodarek wrote: > [...] > > First, thanks for this detailed description. It was helpful to read > > along with the patches. > > > > My only concern about this area being generic is that you are now

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-25 Thread John Fastabend
On 09/25/2017 11:50 AM, Daniel Borkmann wrote: > On 09/25/2017 08:10 PM, Andy Gospodarek wrote: > [...] >> First, thanks for this detailed description.  It was helpful to read >> along with the patches. >> >> My only concern about this area being generic is that you are now in a >> state where any

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-25 Thread Daniel Borkmann
On 09/25/2017 08:10 PM, Andy Gospodarek wrote: [...] First, thanks for this detailed description. It was helpful to read along with the patches. My only concern about this area being generic is that you are now in a state where any bpf program must know about all the bpf programs in the

Re: [PATCH net-next 2/6] bpf: add meta pointer for direct access

2017-09-25 Thread Andy Gospodarek
On Mon, Sep 25, 2017 at 02:25:51AM +0200, Daniel Borkmann wrote: > This work enables generic transfer of metadata from XDP into skb. The > basic idea is that we can make use of the fact that the resulting skb > must be linear and already comes with a larger headroom for supporting >