On 8/27/20 12:10 AM, Mathieu Poirier wrote: > I had another very long look at this... I haven't had the time to look in your > next serie so the end result is not yet clear in my head. But... > > In __rpmsg_create_channel() the new code is testing for VIRTIO_RPMSG_F_NS in > order to allocate the ops, which means that the rpdev that is fed to > virtio_rpmsg_announce_create/destroy() are associated with a virtproc_info. > Moreover there is a test in virtio_rpmsg_announce_create/destroy() that checks > for VIRTIO_RPMSG_F_NS already. So if the rpdev was created from another > interface than the name announcement then no message would be sent. > > So it seems to me that the last two patches could be dropped and things would > still work properly. Yes the last 2 patches concerns the TX part. Drop them could also be an option. > > The good news is that I reviewed V2 today and things look good. I will wait to > review your next set before adding RB tags. The down side is that having spent > a fair amount of time on your set, I need to look at other people's work if I > want to be fair to everyone. As such I have to push back the review of your > other set to next week. That's fair! And this will give me time to have a look to the detach series. Thanks, Arnaud > > Thanks, > Mathieu >
