On Fri, Oct 21, 2011 at 10:37:17AM -0700, Jesse Gross wrote:
> On Fri, Oct 21, 2011 at 10:18 AM, Ben Pfaff wrote:
> > In testing, I found a bug: "later" fragments were being extracted with
> > a key_len that covered the transport ports, but the key sent to
> > and received back from userspace omit
On Fri, Oct 21, 2011 at 10:18 AM, Ben Pfaff wrote:
> In testing, I found a bug: "later" fragments were being extracted with
> a key_len that covered the transport ports, but the key sent to
> and received back from userspace omitted it, so that the flow that
> userspace set up didn't actually matc
On Thu, Oct 20, 2011 at 03:56:41PM -0700, Jesse Gross wrote:
> On Wed, Oct 19, 2011 at 10:56 PM, Ben Pfaff wrote:
> > On Wed, Oct 19, 2011 at 06:09:30PM -0700, Jesse Gross wrote:
> >> Otherwise, the incremental looks good. ??However, I realized that there
> >> is one more issue: when we pass up th
On Wed, Oct 19, 2011 at 10:56 PM, Ben Pfaff wrote:
> On Wed, Oct 19, 2011 at 06:09:30PM -0700, Jesse Gross wrote:
>> Otherwise, the incremental looks good. However, I realized that there
>> is one more issue: when we pass up the flow key for UDP GSO packets,
>> they will all reflect the first fra
On Wed, Oct 19, 2011 at 06:09:30PM -0700, Jesse Gross wrote:
> > switch (wc->tos_frag_mask & FLOW_FRAG_MASK) {
> > -case FLOW_FRAG_ANY | FLOW_FRAG_FIRST:
> > +case FLOW_FRAG_ANY | FLOW_FRAG_LATER:
> > ds_put_format(s, "frag=%s,",
> > f->tos_frag & FLOW_FRAG
On Wed, Oct 19, 2011 at 3:19 PM, Ben Pfaff wrote:
> On Mon, Oct 17, 2011 at 09:19:44PM -0700, Jesse Gross wrote:
>> On Tue, Oct 11, 2011 at 4:33 PM, Ben Pfaff wrote:
>> > diff --git a/datapath/flow.c b/datapath/flow.c
>> > index 7322295..9a9b0aa 100644
>> > --- a/datapath/flow.c
>> > +++ b/datapa
On Tue, Oct 11, 2011 at 4:33 PM, Ben Pfaff wrote:
> diff --git a/datapath/flow.c b/datapath/flow.c
> index 7322295..9a9b0aa 100644
> --- a/datapath/flow.c
> +++ b/datapath/flow.c
> + * Correct behavior when there's more than one fragment header is anybody's
> + * guess. This version reports wheth
On Wed, Oct 12, 2011 at 06:12:27PM -0700, Jesse Gross wrote:
> On Tue, Oct 11, 2011 at 4:05 PM, Ben Pfaff wrote:
> > Until now, OVS has handled IP fragments more awkwardly than necessary. ??It
> > has not been possible to match on L4 headers, even in fragments with offset
> > 0 where they are actu
On Tue, Oct 11, 2011 at 4:05 PM, Ben Pfaff wrote:
> Until now, OVS has handled IP fragments more awkwardly than necessary. It
> has not been possible to match on L4 headers, even in fragments with offset
> 0 where they are actually present. This means that there was no way to
> implement ACLs th
