On Monday 10 March 2014 20:16:25 Geert Uytterhoeven wrote:

On Mon, Mar 10, 2014 at 7:59 PM, Arnd Bergmann [off-list ref] wrote:

quoted

quoted

arch/avr32/include/asm/io.h:#define IO_SPACE_LIMIT    0xffffffff
arch/frv/include/asm/io.h:#define IO_SPACE_LIMIT      0xffffffff

They have no MMU, and the code relies on the port number to match
both the virtual and the physical address. You could be right about
these, but I would guess that the code also needs some other
changes if we want to make it work on nommu kernels. It also depends
on whether the I/O bus address is the same as the CPU address, or
if it starts at bus address 0.

quoted

quoted

arch/m68k/include/asm/io_no.h:#define IO_SPACE_LIMIT 0xffffffff

This looks like a mistake, it should be smaller

io_no.h is for nommu.

Ah, I missed that. In that case I assume it doesn't matter because
the only m68k with PCI is M54xx and that always has an MMU.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html