On Fri, 2019-04-05 at 09:28 +0700, Mark Brown wrote:
On Thu, Apr 04, 2019 at 07:23:54PM -0700, Ranjani Sridharan wrote:
Agree that this might make the clean up robust. But today we call this only when the component is unregistered and that's causing a null pointer dereference. So what we really need to do is to remove the topology in remove() first.
If that's what this is doing then the changelog should reflect that it's fixing a null pointer dereference rather than talking about these separate issues.
Sure, I think I will drop the first patch as like you said it is good for robustness. Adding the remove() callback for skl and sof will fix the null pointer dereference issue.
But I really need some feedback from you on the third patch. I think that's the most invasive and I'm not sure about how to go about removing the topology before the snd_card is freed without removing the link components while unbinding the card.
Thanks, Ranjani