[PATCH v2 1/5] irqchip: add dumb demultiplexer implementation

[Rob Herring]

On Thu, Jan 15, 2015 at 3:11 AM, Thomas Gleixner <tglx at linutronix.de> wrote:
> On Wed, 14 Jan 2015, Rob Herring wrote:
>> On Wed, Jan 14, 2015 at 4:36 AM, Thomas Gleixner <tglx at linutronix.de> wrote:
>> > All attempts to work around that have resulted in horrible bandaids so
>> > far. That's why I guided Boris to implement this dummy demultiplexing
>> > mechanism. It solves the problem at hand nicely without adding nasty
>> > hackarounds into the suspend/resume code and inflicting platform
>> > knowledge on multi-platform device drivers.
>>
>> This change will break on old kernels with a new dtb. Would you be
>> happy if a BIOS update required a new kernel? Fixing this for any
>> platform requires a dtb update which may not be possible on some
>> platforms. I don't have a problem with this breakage for 1 platform
>> and the at91 guys may not care, but we'd ultimately be changing how
>> all shared irqs are specified for all DT. Maybe we decide that this is
>> how we want to describe things, but that needs much wider discussion
>> and agreement.
>
> We do not change shared interrupts in any way. We provide an
> alternative mechanism for braindead hardware. And if the at91 folks
> are fine with the DT change, then it's their decision. Nothing forces
> this on everyone.

We are changing how shared interrupts are described in DT. We don't
need 2 ways to describe them. We could say this is only for AT91 and
continue to describe shared interrupts as has been done forever. Then
the next platform that hits this problem will have to go thru the same
ABI breakage. Or we change DT practices to describe all shared
interrupts with a demux node. Given the way DTs are incrementally
created, it is not something we can check with review or tools, so we
will still have the same ABI breakage problem.

>> > If you have a proper solution for the problem at hand which
>> >
>> >    - avoids the demux dummy
>> >
>> >    - works straight forward with suspend/resume/wakeup
>> >
>> >    - does not add horrible new APIs
>> >
>> >    - does not add conditionals to the interrupt hotpath
>> >
>> >    - does not inflict platform knowledge about interrupt chip details
>> >      on drivers
>> >
>> > then I'm happy to take it.
>> >
>> > But as long as you can't come up with anything sane, the demux dummy
>> > is the best solution for this problem we've seen so far.
>>
>> What if during suspend you move all actions w/o IRQF_NO_SUSPEND to a
>> suspended action list? This would leave only the actions with
>> IRQF_NO_SUSPEND set in the active action list. The cost would be a
>> pointer in irq_desc and moving the actions during suspend and resume.
>
> That's exactly what we want NOT. Because it prevents us to do proper
> sanity checks for IRQF_NO_SUSPEND. We've been there and I rejected it
> for that very reason.
>
>> There are probably ways to do this demux irqchip without a DT change.
>
> What's the problem with a DT change for a single platform, if the
> maintainers are willing to take it and deal with the fallout?

What's the solution for a platform that an ABI break is not okay and
can't deal with the fallout?

> And in case of AT91 the new kernel will simply work with the old DT
> and just emit the same warning vs. the IRQF_NO_SUSPEND mismatch. Older
> kernels won't work with a new DT, but that's about it.
>
> Aside of that, I'm quite amused about your DT update worries. DTs
> break with every kernel version on very popular platforms in very
> weird and subtle ways.

Someone has to. It is a problem that we need to get better at
preventing. It's an ABI. I don't think I need to explain what that
means.

Certainly there are folks who would just prefer DT to be a kernel data
structure so we don't have to worry about this ABI nonsense.

Rob