Re: Why isn't IRQ shared for i2c-ocore

[Lee Jones <lee.jones@linaro.org>]

On Fri, 08 May 2015, Geert Uytterhoeven wrote:

> On Fri, May 8, 2015 at 8:57 AM, Lee Jones <lee.jones@linaro.org> wrote:
> > On Fri, 08 May 2015, Geert Uytterhoeven wrote:
> >
> >> On Thu, May 7, 2015 at 9:01 AM, Lee Jones <lee.jones@linaro.org> wrote:
> >> >> I have a follow up question regarding interrupt. I see many I2C bus drivers
> >> >> request interrupt with flag = 0. Why not using IRQF_SHARED?
> >> >
> >> > Probably because that particular IRQ is only used by the I2C
> >> > Controller.  I'm not exactly sure that you're getting at?  Why do you
> >> > think it should be shared?  You should only flag it as shared if it
> >> > is.
> >>
> >> However, that's something the driver can't know.
> >> Sharing interrupts is an integration property. The same IP core may share its
> >> interrupt on one SoC, and not on another.
> >
> > I guess that would depend on the IP.  If this is part of an MFD, you'd
> > know if you only hand a single interrupt line coming into the chip or
> > not.  If the IP can be moved around (copy & pasted) into different
> > chips, then yes, that might change.
> >
> > How does one share an interrupt with other drivers if all them don't
> > know the IRQ is shared thought?
> 
> All drivers sharing the same interrupt must pass IRQF_SHARED, cfr. the
> checks in __setup_irq().
> 
> Traditionally, PC (ISA) drivers didn't share interrupts, as the ISA bus
> prohibited interrupt sharing.
> Amiga drivers did.
> 
> New drivers should support IRQ sharing.

Precisely, which is why I'm confused by:

  "However, that's something the driver can't know."

-- 
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog