On Thu, Oct 30, 2014 at 3:48 AM, Arnd Bergmann <arnd@xxxxxxxx> wrote:
> On Thursday 30 October 2014 11:43:00 Thomas Gleixner wrote:
>> On Thu, 30 Oct 2014, Arnd Bergmann wrote:
>>
>> > On Wednesday 29 October 2014 19:17:55 Kevin Cernekee wrote:
>> > > Defining these macros way down in arch/sh/.../irq.c doesn't cause
>> > > kernel/irq/generic-chip.c to use them. As far as I can tell this code
>> > > has no effect.
>> > >
>> > > Signed-off-by: Kevin Cernekee <cernekee@xxxxxxxxx>
>> >
>> > Actually it overrides the 32-bit accessors with 16-bit accessors,
>> > which does seem intentional and certainly has an effect.
>>
>> Not really. Neither arch/sh/boards/mach-se/7343/irq.c nor
>> arch/sh/boards/mach-se/7722/irq.c actually use
>> irq_reg_readl/writel. They simply define it.
>
> Ah, that makes things easier. I looked at the commits that introduced
> them, and even then they were unused. Probably an artifact from an
> earlier version of the patch which did not get merged.
I suspect that the intention was to put them into the machine's
<irq.h> so that generic-chip.c would pick them up. The sh irq.c files
do call ioread16/iowrite16 themselves, and like bcm7120-l2, they
utilize irq_gc_mask_{set,clr}_bit from generic-chip.c.
This may be something for the maintainers (Paul?) to double-check, but
if the submission worked as-is, maybe overriding the I/O accessors
wasn't actually needed. Or maybe there are subtle and unfortunate
side effects on adjacent registers.
