On 04/06/2018 02:53 PM, Justin Pettit wrote:

On Apr 6, 2018, at 12:27 PM, Mark Michelson <mmich...@redhat.com> wrote:

I just had a look, and unfortunately, the current code doesn't translate directly to a latch. The 
reason is that you can't control the data that is sent to and read from the latch. It's essentially 
a boolean of "set" or "not set". In the current stopwatch implementation, data 
packets are written to the pipe and that data is then read by another thread. The actual data is 
important in this case.

I noticed the same thing.  I spoke with Ben off-line, and I'll push my changes 
(broken into the fix and the style changes) so that we reduce the chances that 
people run into at least the issue I had.  You'll take care of the bigger 
change to handle Windows?

Thanks!

--Justin



Yes, I'll do that. I'll investigate Ben's idea of using a guarded_list and a seq.

Mark!
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to