On Mon, 15 Jun 2020 at 16:28, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: > > On Mon, Jun 15, 2020 at 04:14:06PM +0300, Vladimir Oltean wrote: > > On Mon, 15 Jun 2020 at 16:10, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: > > > > > > > > > > > It is a slightly different bug which so this patch should have a follow > > > up. > > > > > > Best regards, > > > Krzysztof > > > > > > > Why is it a different bug? It's the same bug. > > One bug is using devm-interface for shared interrupts and second is not > caring about suspend/resume. > > Best regards, > Krzysztof > The problem is that you don't have a way to stop servicing a shared interrupt safely and on demand, before clk_disable_unprepare. So it's exactly the same problem on suspend and on remove. Avoiding to think about the suspend problem now means that you'll end up having an overall worse solution.