Re: [PATCH 1/3] dt-bindings: media: mali-c55: Allow wakeup source

[Rob Herring]

On Thu, Oct 08, 2026 at 03:22:18PM +0200, Linus Walleij wrote:
> On Wed, Oct 7, 2026 at 11:29 AM Dan Scally <dan.scally@xxxxxxxxxxxxxxxx> wrote:
> 
> > On 29/09/2026 13:02, Linus Walleij wrote:
> > > The Mali-C55 interrupt can be wired as a wake source on platforms that
> > > support it. Use the standard wakeup-source property.
> >
> > That's interesting! How's that work, though? In the RZ/V2H particularly the ISP is in an offline
> > mode, so I'm struggling to see how one could trigger the interrupt whilst the device is suspended,
> > since you'd need to instruct the IVC to queue a frame.
> 
> The implementation is IP-neutral and doesn't really ask this question.
> 
> The typical silicon solution would be if the IP was synthesized into
> and always-on power domain on the chip. This means the CPU
> which runs Linux is powered off, while the ISP is still powered
> on, and could generate a wakeup IRQ. This is how e.g. GPIO
> units that can wake up the system work: they are always
> powered in some sense.

In that sense, anything that uses 'interrupts' should accept 
'wakeup-source'. And we could implicitly allow that if that's the rule 
rather than the exception.

> 
> I don't know if that makes any *sense* from a system engineering
> point of view, but it is something that can be done.

But I don't see how it makes sense for an ISP. 'wakeup-source' is for 
system level suspend. When we're powering off idle cores and need a 
wakeup from an interrupt, that's a different, implicit mechanism. For 
example, we don't put wakeup-source on (tick) timers.

Rob