[ovs-dev] [PATCH v10 3/5] dpif-netdev: Time based output batching.

[Jan Scheurich]

> From: Ilya Maximets [mailto:i.maximets at samsung.com]
> Sent: Monday, 15 January, 2018 08:23
> To: Jan Scheurich <jan.scheurich at ericsson.com>; ovs-dev at openvswitch.org
> Cc: Heetae Ahn <heetae82.ahn at samsung.com>; Bhanuprakash Bodireddy <bhanuprakash.bodireddy at intel.com>; Antonio Fischetti
> <antonio.fischetti at intel.com>; Eelco Chaudron <echaudro at redhat.com>; Ciara Loftus <ciara.loftus at intel.com>; Kevin Traynor
> <ktraynor at redhat.com>; Billy O'Mahony <billy.o.mahony at intel.com>; Ian Stokes <ian.stokes at intel.com>
> Subject: Re: [PATCH v10 3/5] dpif-netdev: Time based output batching.
> 
> On 12.01.2018 19:42, Jan Scheurich wrote:
> > Hi Ilya,
> >
> > I understand that for the purpose of the present patch this is good enough to count rx and tx packets together, but the PMD performance
> metrics patch will depend on counting them separately. It would have been nice if you had simply included the changes I proposed. Now I
> will have to revert some of your changes to make the tx_packets available to pmd_perf_end_iteration(). But so be it.
> 
> 
> Hello, Jan.
> I tried to find anything about tx_packets in your v5 of supervision patch-set
> and RFC rebased series on mail-list but failed. At the first glance these series
> just ignores the tx packets and the cycles spent for sending them. Am I missing
> something? Or you didn't send these changes to mail-list?

I guess that was difficult to see in the jungle of patches ;-).
Now that I look for it I can't find it myself any longer in those patches.

> 
> In general, I think, it's better if you'll made this changes yourself, because
> you know your code better.

Fine with me, I have done that now when rebasing my series on top of your v10. Will send out soon.

BR, Jan