On Wed, Mar 14, 2018 at 2:59 PM, Ben Pfaff wrote:
> I posted an alternative proposal:
> https://patchwork.ozlabs.org/patch/886052/
>
> I haven't checked that this solves the problem, so I would appreciate
> testing.
Thanks for this fix. With your patch, eth() is printed with
eth_type() for the k
On Sat, Mar 10, 2018 at 09:30:34AM -0800, Yi-Hung Wei wrote:
> Currently, using ofproto/trace to trace a datapath flow with eth_type()
> but without eth() may hit unexpected behavior because OVS sets
> the packet_type to be (1, eth_type) when decoding the odp flow key.
> This patch updates the logi
On Tue, 13 Mar 2018 09:36:23 -0700, Darrell Ball wrote:
> [Darrell] There was no suggestion otherwise in general. We were discussing
> for one the difference b/w kernel and userspace DP for handling packet type
> aware support. Specifically, the difference is here is as fundamental as it
> gets -
>
On Tue, Mar 13, 2018 at 09:36:23AM -0700, Darrell Ball wrote:
> On Tue, Mar 13, 2018 at 1:24 AM, Jiri Benc wrote:
>
> > Darell, please fix your email client configuration to conform to
> > RFC 3676 (https://tools.ietf.org/html/rfc3676#section-4.5). Your
> > replies are unreadable. In particular,
On Tue, Mar 13, 2018 at 1:24 AM, Jiri Benc wrote:
> Darell, please fix your email client configuration to conform to
> RFC 3676 (https://tools.ietf.org/html/rfc3676#section-4.5). Your
> replies are unreadable. In particular, the text you're replying to must
> be quoted by ">" character, clearly s
Darell, please fix your email client configuration to conform to
RFC 3676 (https://tools.ietf.org/html/rfc3676#section-4.5). Your
replies are unreadable. In particular, the text you're replying to must
be quoted by ">" character, clearly separating your reply from the
original email.
Also, wrap yo
Darrell Ball
; Yi-Hung Wei
> Cc: ovs dev ; Jiri Benc (jb...@redhat.com)
> Subject: Re: [ovs-dev] [PATCH] odp-util: Fix ofproto/trace with odp flow
>
> ovs-fields has this description regarding packet type:
>
> “Matching on packet_type is a pre-requisite for
ng Wei
> Cc: ovs dev ; Jiri Benc (jb...@redhat.com)
>
> Subject: Re: [ovs-dev] [PATCH] odp-util: Fix ofproto/trace with odp flow
>
> ovs-fields has this description regarding packet type:
>
> “Matching on packet_type is a pre-requisite for matching on any data field,
> b
ou will break the
L3 tunneling and PTAP for the kernel datapath.
BR, Jan
> -Original Message-
> From: ovs-dev-boun...@openvswitch.org
[mailto:ovs-dev-boun...@openvswitch.org] On Behalf Of Darrell Ball
> Sent: Sunday, 11 March, 2018 17:51
> To: Yi
or the kernel datapath.
BR, Jan
> -Original Message-
> From: ovs-dev-boun...@openvswitch.org
[mailto:ovs-dev-boun...@openvswitch.org] On Behalf Of Darrell Ball
> Sent: Sunday, 11 March, 2018 17:51
> To: Yi-Hung Wei
> Cc: ovs dev
> Subject: Re: [
g and PTAP for the kernel datapath.
BR, Jan
> -Original Message-
> From: ovs-dev-boun...@openvswitch.org
> [mailto:ovs-dev-boun...@openvswitch.org] On Behalf Of Darrell Ball
> Sent: Sunday, 11 March, 2018 17:51
> To: Yi-Hung Wei
> Cc: ovs dev
> Subject: Re: [ovs-dev
Thanks Yi-hung
One minor comment inline.
On Sat, Mar 10, 2018 at 9:30 AM, Yi-Hung Wei wrote:
> Currently, using ofproto/trace to trace a datapath flow with eth_type()
> but without eth() may hit unexpected behavior because OVS sets
> the packet_type to be (1, eth_type) when decoding the odp flo
Currently, using ofproto/trace to trace a datapath flow with eth_type()
but without eth() may hit unexpected behavior because OVS sets
the packet_type to be (1, eth_type) when decoding the odp flow key.
This patch updates the logic of odp flow key decoding so that the
packet_type defaults to (0,0)
13 matches
Mail list logo