[PATCH v6 1/3] Fix ioport_map() for !CONFIG_GENERIC_IOMAP cases.

[Russell King - ARM Linux]

On Fri, Mar 07, 2014 at 01:37:38AM +0100, Arnd Bergmann wrote:
> On Thursday 06 March 2014, Russell King - ARM Linux wrote:
> > On Wed, Mar 05, 2014 at 11:49:08AM +0000, Liviu Dudau wrote:
> > > The inline version of ioport_map() that gets used when !CONFIG_GENERIC_IOMAP
> > > is wrong. It returns a mapped (i.e. virtual) address that can start from
> > > zero and completely ignores the PCI_IOBASE and IO_SPACE_LIMIT that most
> > > architectures that use !CONFIG_GENERIC_MAP define.
> > 
> > What value does PCI_IOBASE and IO_SPACE_LIMIT have on other architectures
> > who make use of asm-generic/io.h ?
> > 
> > $ git grep asm-generic/io.h arch/
> > arch/arc/include/asm/io.h:#include <asm-generic/io.h>
> 
> PCI support hasn't been upstreamed.
> 
> > arch/blackfin/include/asm/io.h:#include <asm-generic/io.h>
> > arch/metag/include/asm/io.h:#include <asm-generic/io.h>
> 
> No PCI support
> 
> > arch/microblaze/include/asm/io.h:/* from asm-generic/io.h */
> 
> PCI_IOBASE=0, IO_SPACE_LIMIT=0xffffffff, so no change.

Seems to define _IO_BASE not PCI_IOBASE.

> > arch/openrisc/include/asm/io.h:#include <asm-generic/io.h>
> 
> No PCI support
> 
> > arch/s390/include/asm/io.h:#include <asm-generic/io.h>
> 
> s390 supports PCI but no I/O space
> 
> > arch/score/include/asm/io.h:#include <asm-generic/io.h>
> 
> No PCI support
> 
> > arch/unicore32/include/asm/io.h:#include <asm-generic/io.h>
> 
> unicore32 is broken currently, the patch fixes it.
> 
> > arch/xtensa/include/asm/io.h:#include <asm-generic/io.h>
> 
> PCI_IOBASE=0, IO_SPACE_LIMIT=0xffffffff, so no change.

Doesn't appear to define PCI_IOBASE.

Maybe there's other patches required for these?

> For most of these, I assume we actually want to remove support
> for inb/outb as they don't support I/O space accesses. The other
> ones look correct to me.

Right, so:

#ifdef CONFIG_HAS_IOPORT
#ifndef CONFIG_GENERIC_IOMAP
static inline void __iomem *ioport_map(unsigned long port, unsigned int nr)
{
        return (void __iomem *) port;
}

changing that to include PCI_IOBASE in there will result in a build
failure if the C compiler sees that.  In other words, when HAS_IOPORT=y
and GENERIC_IOMAP=n.

HAS_IOPORT is set when HAS_IOMEM is also set and NO_IOPORT unset.

It looks to me like blackfin doesn't set NO_IOPORT nor NO_IOMEM, so
this would have HAS_IOPORT set, and from what I can see doesn't set
GENERIC_IOMAP.  So, this change probably breaks blackfin.

I haven't looked deeply at the others.

-- 
FTTC broadband for 0.8mile line: now at 9.7Mbps down 460kbps up... slowly
improving, and getting towards what was expected from it.