Re: [PATCH 0/3] PCI: designware: Fixing MSI handling flow

[Trent Piepho <tpiepho@impinj.com>]

On Mon, 2018-12-10 at 18:31 +0000, Marc Zyngier wrote:
> 
> > I had concerns about what appears to be an unnecessary extra lock taken
> > before handling an interrupt and enabling all MSIs even if nothing has
> > tried to enable them.
> 
> Regarding the lock: I'm quite puzzled that you consider it
> "unnecessary", given that all the DWC callbacks expect such a locking. I
> suspect you are considering from a pure performance angle, and I'd
> suggest that you post numbers showing the unacceptable overhead of an
> otherwise uncontended lock.

The difference is that the other callbacks can be called outside the
path of handling an interrupt.  Any thread could possibly make a call
to mask the interrupt at any time, so the lock is needed.

But the ack is only called in the path of handling an irq.  That's why
its different.  There's already a system insuring that callback is not
called reentrantly.  If the lock was ever taken when it was attempted
to be locked, then we'd already be broken.

> As for enabling all MSIs upfront, same thing. Please demonstrate how
> harmful it is, given that they are all masked by default, consistently
> with what other interrupt controllers are doing.

Without any information on the controller, who knows what the
differences between a masked and enabled vs a disabled msi are?  I
don't know there is a difference, but on the other hand, you don't know
there isn't.  It's like a dozen lines of code to not change the
behavior.