[PATCH 03/17] mmc: mmci: Mask IRQs for all variants during runtime suspend

[Ulf Hansson]

On 18 February 2014 17:46, Russell King - ARM Linux
<linux at arm.linux.org.uk> wrote:
> On Tue, Feb 18, 2014 at 05:36:52PM +0100, Ulf Hansson wrote:
>> If we add SDIO irq support to mmci in future; parts of that
>> implementation includes a re-route of DAT1 to a GPIO irq when entering
>> runtime suspend state. The mmci HW will in runtime suspend state, not
>> be responsible for handling irqs, which is the same as of today.
>
> Note that the "irq thread" in sdio_irq is scheduled for destruction -
> it's buggy when the system is under high load and a driver claims a
> SDIO irq.  Sched people hate it too...

We kind of realized that too, in the early days for ux500. At that
point we converted it to a use a dedicated work queue, which improved
the behaviour, don't remember the details why, but maybe I should
collect the patches and send them out. :-)

Finally we moved to use a separate GPIO irq as a dedicated SDIO irq
line for our cw1200 wlan chip, mostly due to gain a bit of
performance. That happened to simplified the mmci part as well, which
was welcome.

>
> --
> FTTC broadband for 0.8mile line: 5.8Mbps down 500kbps up.  Estimation
> in database were 13.1 to 19Mbit for a good line, about 7.5+ for a bad.
> Estimate before purchase was "up to 13.2Mbit".