[RFC] introduce ioremap() and dev_ioremap_resource()

[Antony Pavlov]

On Fri, 19 Dec 2014 07:45:42 +0100
Sascha Hauer <s.hauer at pengutronix.de> wrote:

> On Fri, Dec 19, 2014 at 01:32:37AM +0300, Antony Pavlov wrote:
> > The commit
> > 
> >     commit 0d7a21334536cb36b3c9b64d868acc55aec41332
> >     Author: Antony Pavlov <antonynpavlov at gmail.com>
> >     Date:   Wed Sep 10 11:42:19 2014 +0400
> > 
> >         MIPS: dts: use physical addresses (as Linux does)
> > 
> >         With IOMEM() adapted for MIPS we can use physical addresses
> >         in device tree reg property.
> > 
> > is premature. Current device tree files unusable on real hardware.
> >                           I'M SO SORRY!
> > The patch was tested on qemu, but qemu malta board is tolerant
> > of using physical addresses for accesing to device. Real hardware
> > can throw an exception in this situation.
> > 
> > Additional physical address to virtual address translation on MIPS is needed.
> > 
> > This commit demonstrate a possible (but not very good) solution.
> > This solution is copied-and-pasted from linux kernel.
> > In linux drivers use devm_ioremap_resource() function, so
> > the patch just replicates similar dev_ioremap_resource() function.
> > 
> > Changing dev_request_mem_region() to dev_ioremap_resource()
> > is not a very good idea because there are too many dev_request_mem_region()
> > in barebox sources
> > 
> >     barebox$ git grep dev_request_mem_region | wc -l
> >     197
> > 
> > moreover dev_ioremap_resource() has the second 'struct resource *res'
> > argument, so additional dev_get_resource() is needed.
> > 
> > Can we insert ioremap() into dev_request_mem_region() directly?
> > (we also can use already existion IOMEM() macro instead of linux' ioremap()).
> 
> I'm not sure if it's a good idea to put that behind some standard
> lookingioremap() call, because its behaviour is not standard.
> 
> I'm also fine with adding some
> 
> #ifdef CONFIG_MIPS
> 	return mips_iomem(res->start);
> #else
> 	return (void __force __iomem *)res->start;
> #endif
> 
> At least this makes explicit that MIPS has a very special handling.

We already have this in include/common.h

#if defined(CONFIG_MIPS)
#include <asm/addrspace.h>

#define IOMEM(addr) ((void __force __iomem *)CKSEG1ADDR(addr))
#else
#define IOMEM(addr) ((void __force __iomem *)(addr))
#endif

So can I use just

    return IOMEM(res->start);

in dev_request_mem_region()?

-- 
Best regards,
  Antony Pavlov