Re: [2/2] genirq: Warn when IRQ_NOAUTOEN is used with shared interrupts

[Paul Burton]

Hi Thomas,

On Wednesday, 6 September 2017 07:14:50 PDT Thomas Gleixner wrote:
> On Wed, 6 Sep 2017, Paul Burton wrote:
> > On Wednesday, 6 September 2017 01:16:48 PDT Thomas Gleixner wrote:
> > > What's the problem with IRQ_NOAUTOEN and do
> > > 
> > >        setup_percpu_irq();
> > >        enable_percpu_irq();
> > > 
> > > on the boot CPU and then later call it when the secondary CPUs come up
> > > in
> > > cpu bringup code or a hotplug state callback?
> > 
> > There's no problem with that at all, apart from that it triggers your
> > warning when the boot CPU calls setup_percpu_irq().
> 
> Ok, I misread the whole thing somehow. What you are saying is that you need
> per cpu interrupts which can be shared ....
> 
> Hrmpf, ok. I'll have a look from that side then

So actually it turns out that getting this to work involves more than just 
avoiding your warning. I have some code which is mostly working; I'll clean up 
& send it as an RFC hopefully tomorrow.

Thanks,
    Paul

Attachment: signature.asc
Description: This is a digitally signed message part.