[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[Ezequiel Garcia]

In order to remove the following ugly message:

  BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space

the iotable mappings should be re-located inside the vmalloc
region. Such move was introduced at commit:

commit 0536bdf33faff4d940ac094c77998cfac368cfff
Author: Nicolas Pitre <nicolas.pitre@linaro.org>
Date:   Thu Aug 25 00:35:59 2011 -0400

    ARM: move iotable mappings within the vmalloc region

While at it, let's add some nicer defines to make the code
more readable.

Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
---
 arch/arm/mach-pxa/generic.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
index 4225417..7c01095 100644
--- a/arch/arm/mach-pxa/generic.c
+++ b/arch/arm/mach-pxa/generic.c
@@ -24,6 +24,7 @@
 #include <mach/hardware.h>
 #include <asm/mach/map.h>
 #include <asm/mach-types.h>
+#include <asm/pgtable.h>
 
 #include <mach/reset.h>
 #include <mach/smemc.h>
@@ -77,6 +78,11 @@ EXPORT_SYMBOL(get_clk_frequency_khz);
  * Note: virtual 0xfffe0000-0xffffffff is reserved for the vector table
  *       and cache flush area.
  */
+
+#define UNCACHED_PHY	0x00000000
+#define UNCACHED_SIZE	SZ_1M
+#define UNCACHED_VIRT	(VMALLOC_END - UNCACHED_SIZE)
+
 static struct map_desc common_io_desc[] __initdata = {
   	{	/* Devs */
 		.virtual	=  0xf2000000,
@@ -84,9 +90,9 @@ static struct map_desc common_io_desc[] __initdata = {
 		.length		= 0x02000000,
 		.type		= MT_DEVICE
 	}, {	/* UNCACHED_PHYS_0 */
-		.virtual	= 0xff000000,
-		.pfn		= __phys_to_pfn(0x00000000),
-		.length		= 0x00100000,
+		.virtual	= UNCACHED_VIRT,
+		.pfn		= __phys_to_pfn(UNCACHED_PHY),
+		.length		= UNCACHED_SIZE,
 		.type		= MT_DEVICE
 	}
 };
-- 
1.8.1.5

[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[Russell King - ARM Linux]

On Sun, Nov 24, 2013 at 09:36:11AM -0300, Ezequiel Garcia wrote:
> In order to remove the following ugly message:
> 
>   BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space
> 
> the iotable mappings should be re-located inside the vmalloc
> region. Such move was introduced at commit:
> 
> commit 0536bdf33faff4d940ac094c77998cfac368cfff
> Author: Nicolas Pitre <nicolas.pitre@linaro.org>
> Date:   Thu Aug 25 00:35:59 2011 -0400
> 
>     ARM: move iotable mappings within the vmalloc region
> 
> While at it, let's add some nicer defines to make the code
> more readable.
> 
> Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>

NAK.  I'm guessing that you haven't properly tested this.  Grep for
UNCACHED_PHYS_0, and what you'll find is that your change will break
standby and suspend on all PXA platforms.

[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[Ezequiel Garcia]

On Sun, Nov 24, 2013 at 12:41:25PM +0000, Russell King - ARM Linux wrote:
> On Sun, Nov 24, 2013 at 09:36:11AM -0300, Ezequiel Garcia wrote:
> > In order to remove the following ugly message:
> > 
> >   BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space
> > 
> > the iotable mappings should be re-located inside the vmalloc
> > region. Such move was introduced at commit:
> > 
> > commit 0536bdf33faff4d940ac094c77998cfac368cfff
> > Author: Nicolas Pitre <nicolas.pitre@linaro.org>
> > Date:   Thu Aug 25 00:35:59 2011 -0400
> > 
> >     ARM: move iotable mappings within the vmalloc region
> > 
> > While at it, let's add some nicer defines to make the code
> > more readable.
> > 
> > Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
> > Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> 
> NAK.  I'm guessing that you haven't properly tested this.  Grep for
> UNCACHED_PHYS_0, and what you'll find is that your change will break
> standby and suspend on all PXA platforms.

Right, I fail to grep about this properly, sorry about that.
Anyway, after grepping for 'UNCACHED' and doing some testings
I found a few interesting things.

First, UNCACHED_PHYS_0 seems to be used by PXA27xx and PXA25xx. The
board I have here is PXA3xx, so there's no way I can test anything.

Question: Should we add an #ifdef around the mapping, or is that
just pure churn?

Second, suspend/resume seems broken on my CM-X300 board, or at least
the wake-up. It enters suspend/standby but then never seem to wake.
Maybe I'm not pushing the right button, but according to the reference
guide the SW2 (connected to EXTWAKE#) should do the trick.

Ideas?
-- 
Ezequiel Garc?a, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com

[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[Nicolas Pitre]

On Sun, 24 Nov 2013, Ezequiel Garcia wrote:

> In order to remove the following ugly message:
> 
>   BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space
> 
> the iotable mappings should be re-located inside the vmalloc
> region. Such move was introduced at commit:
> 
> commit 0536bdf33faff4d940ac094c77998cfac368cfff
> Author: Nicolas Pitre <nicolas.pitre@linaro.org>
> Date:   Thu Aug 25 00:35:59 2011 -0400
> 
>     ARM: move iotable mappings within the vmalloc region
> 
> While at it, let's add some nicer defines to make the code
> more readable.
> 
> Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
> Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> ---
>  arch/arm/mach-pxa/generic.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
> index 4225417..7c01095 100644
> --- a/arch/arm/mach-pxa/generic.c
> +++ b/arch/arm/mach-pxa/generic.c
> @@ -24,6 +24,7 @@
>  #include <mach/hardware.h>
>  #include <asm/mach/map.h>
>  #include <asm/mach-types.h>
> +#include <asm/pgtable.h>
>  
>  #include <mach/reset.h>
>  #include <mach/smemc.h>
> @@ -77,6 +78,11 @@ EXPORT_SYMBOL(get_clk_frequency_khz);
>   * Note: virtual 0xfffe0000-0xffffffff is reserved for the vector table
>   *       and cache flush area.
>   */
> +
> +#define UNCACHED_PHY	0x00000000
> +#define UNCACHED_SIZE	SZ_1M
> +#define UNCACHED_VIRT	(VMALLOC_END - UNCACHED_SIZE)
> +
>  static struct map_desc common_io_desc[] __initdata = {
>    	{	/* Devs */
>  		.virtual	=  0xf2000000,
> @@ -84,9 +90,9 @@ static struct map_desc common_io_desc[] __initdata = {
>  		.length		= 0x02000000,
>  		.type		= MT_DEVICE
>  	}, {	/* UNCACHED_PHYS_0 */
> -		.virtual	= 0xff000000,
> -		.pfn		= __phys_to_pfn(0x00000000),
> -		.length		= 0x00100000,
> +		.virtual	= UNCACHED_VIRT,
> +		.pfn		= __phys_to_pfn(UNCACHED_PHY),
> +		.length		= UNCACHED_SIZE,
>  		.type		= MT_DEVICE
>  	}
>  };

You might wish to change the virtual address for this mapping, but 
nowhere in your patch you are changing users of that virtual mapping.


Nicolas

[PATCH] ARM: pxa: Move iotable mapping inside vmalloc region

[Ezequiel Garcia]

On Sun, Nov 24, 2013 at 06:26:45PM -0500, Nicolas Pitre wrote:
> On Sun, 24 Nov 2013, Ezequiel Garcia wrote:
> 
> > In order to remove the following ugly message:
> > 
> >   BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space
> > 
> > the iotable mappings should be re-located inside the vmalloc
> > region. Such move was introduced at commit:
> > 
> > commit 0536bdf33faff4d940ac094c77998cfac368cfff
> > Author: Nicolas Pitre <nicolas.pitre@linaro.org>
> > Date:   Thu Aug 25 00:35:59 2011 -0400
> > 
> >     ARM: move iotable mappings within the vmalloc region
> > 
> > While at it, let's add some nicer defines to make the code
> > more readable.
> > 
> > Cc: Nicolas Pitre <nicolas.pitre@linaro.org>
> > Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> > ---
> >  arch/arm/mach-pxa/generic.c | 12 +++++++++---
> >  1 file changed, 9 insertions(+), 3 deletions(-)
> > 
> > diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
> > index 4225417..7c01095 100644
> > --- a/arch/arm/mach-pxa/generic.c
> > +++ b/arch/arm/mach-pxa/generic.c
> > @@ -24,6 +24,7 @@
> >  #include <mach/hardware.h>
> >  #include <asm/mach/map.h>
> >  #include <asm/mach-types.h>
> > +#include <asm/pgtable.h>
> >  
> >  #include <mach/reset.h>
> >  #include <mach/smemc.h>
> > @@ -77,6 +78,11 @@ EXPORT_SYMBOL(get_clk_frequency_khz);
> >   * Note: virtual 0xfffe0000-0xffffffff is reserved for the vector table
> >   *       and cache flush area.
> >   */
> > +
> > +#define UNCACHED_PHY	0x00000000
> > +#define UNCACHED_SIZE	SZ_1M
> > +#define UNCACHED_VIRT	(VMALLOC_END - UNCACHED_SIZE)
> > +
> >  static struct map_desc common_io_desc[] __initdata = {
> >    	{	/* Devs */
> >  		.virtual	=  0xf2000000,
> > @@ -84,9 +90,9 @@ static struct map_desc common_io_desc[] __initdata = {
> >  		.length		= 0x02000000,
> >  		.type		= MT_DEVICE
> >  	}, {	/* UNCACHED_PHYS_0 */
> > -		.virtual	= 0xff000000,
> > -		.pfn		= __phys_to_pfn(0x00000000),
> > -		.length		= 0x00100000,
> > +		.virtual	= UNCACHED_VIRT,
> > +		.pfn		= __phys_to_pfn(UNCACHED_PHY),
> > +		.length		= UNCACHED_SIZE,
> >  		.type		= MT_DEVICE
> >  	}
> >  };
> 
> You might wish to change the virtual address for this mapping, but 
> nowhere in your patch you are changing users of that virtual mapping.
> 

Yeah, this patch is pure crap. I'll re-work it and submit a new
(hopefully not stupid) version.

Thanks for taking a look!
-- 
Ezequiel Garc?a, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com