Em Tue, 29 Mar 2016 09:45:07 +0200 Hans Verkuil hverkuil@xs4all.nl escreveu:
Hi Mauro,
On 03/14/2016 06:28 PM, Mauro Carvalho Chehab wrote:
Em Mon, 14 Mar 2016 14:41:16 +0100 Hans Verkuil hverkuil@xs4all.nl escreveu:
I have these topics:
- Continue MC connector entity discussion
Although I'm not working on it, I think we should also discuss:
- support for dynamic removal/addition at the core and MC locking.
Are these two topics (dynamic removal/addition AND MC locking) or one (MC locking in the context of dynamic removal/addition)? I think it is one topic, right?
It is one topic, but from my side, I guess we can exclude it, as Sakari's suggestion of replacing the MC spin lock with the MC mutex seemed to address the issues the core used to have.
Shuah and I did lots of test keeping either snd-usb-audio or au0828 loaded and doing thousands of unbind/bind endless loops on the other driver. While we got a few minor bugs with such stress test (one of them is related to cdev at the V4L2 core), none are related to the MC code.
So, the core should now (after applying the fixups) be safe for dynamic removal/addition.
Thanks, Mauro