Felipe Balbi <balbi@xxxxxx> writes:
> On Thu, Nov 01, 2012 at 11:06:58AM -0400, David Miller wrote:
>> From: Bjørn Mork <bjorn@xxxxxxx>
>> Date: Thu, 01 Nov 2012 14:38:29 +0100
>>
>> > Felipe Balbi <balbi@xxxxxx> writes:
>> >
>> >> btw, only alpha and x86 seem to be providing all of those, but if you
>> >> look into our documentation, it's said that drivers are supposed to use
>> >> write/read{b,w,l,q} for all memory-mapped io. I'd expect all platforms
>> >> to provide those even if just to allow compilation of drivers.
>> >
>> > Only if HAS_IOMEM is true, I believe.
>>
>> This is correct.
>
> fair enough, then we can depend on HAS_IOMEM.
>
> Another question, if HAS_IOMEM is true, is the arch expected to provide
> *all* accessor methods ?
I'll leave this for those with more clue...
But there is a comment in arch/alpha/include/asm/io.h indicating that
writeq/readq support is tested separately, so I guess it must be
optional:
/*
* Some mucking forons use if[n]def writeq to check if platform has it.
* It's a bloody bad idea and we probably want ARCH_HAS_WRITEQ for them
* to play with; for now just use cpp anti-recursion logics and make sure
* that damn thing is defined and expands to itself.
*/
#define writeq writeq
#define readq readq
> And the last one: if the arch provides
> read/write{b,w,l,q}, should it be selecting HAS_IOMEM ??
But that one is easy: HAS_IOMEM is true by default. The arch sets
NO_IOMEM to disable it. Similar to HAS_DMA. So you can be pretty sure
that HAS_IOMEM is selected unless the arch has reasons not to. From
lib/Kconfig:
config HAS_IOMEM
boolean
depends on !NO_IOMEM
select GENERIC_IO
default y
Bjørn
--
To unsubscribe from this list: send the line "unsubscribe linux-arch" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
