Re: [PATCH] iomap: make PCI iomap stuff excluded when PCI isn't configured

[Al Viro <viro@ftp.linux.org.uk>]

On Sat, Feb 10, 2007 at 10:50:33PM +0000, Al Viro wrote:
> We don't even require ioremap().  What the hell could it possibly mean on
> e.g. UML?  The same goes for DMA.  And yes, of course it means that a lot
> of drivers are not buildable on it - hardly a surprise, that...

FWIW, the current picture wrt io-related stuff looks so:

alpha frv i386 ia64 mips parisc powerpc sparc64 x86_64: full set
s390: no ioport_map, no ioread*/iowrite*, no port IO except for (in|out)b(_p|)
sparc: insl() and friends are not there unless we have PCI, ioremap() is
not exported unless on PCI.
um: odd stubs for inb()/outb(), no other port IO; no ioremap(), no
memcpy_..io(), for some reason read[bwlq]()/write[bwlq]() are there,
but never used.  No ioread*/iowrite*.
arm: depends on subarchitecture.  E.g. 32bit port access might be a sham
(causes fun problems with e.g. modular ide-core on some subarchitectures -
it doesn't _use_ insl(), but the reference is there no matter what, so the
lack of export => missing symbol).
m68k: as if.  No ioread/iowrite, ioremap() only for hp300, availability of
any IO accessor depends on config in rather weird ways.
m32r: no ioread/iowrite/ioport_map
the rest of embedded stuff: no idea, never played with those targets.