Re: [2.6 patch] fix fallout from the pxa-regs.h #define move

[Russell King - ARM Linux <linux@xxxxxxxxxxxxxxxx>]

Okay, now that everything pending has been merged...

On Sat, Jul 19, 2008 at 07:46:28PM +0300, Adrian Bunk wrote:
> diff --git a/drivers/watchdog/sa1100_wdt.c b/drivers/watchdog/sa1100_wdt.c
> index 34a2b3b..40b724d 100644
> --- a/drivers/watchdog/sa1100_wdt.c
> +++ b/drivers/watchdog/sa1100_wdt.c
> @@ -29,6 +29,7 @@
>  
>  #ifdef CONFIG_ARCH_PXA
>  #include <asm/arch/pxa-regs.h>
> +#include <asm/arch/pxa2xx-regs.h>
>  #endif
>  
>  #include <asm/hardware.h>
>  #include <asm/hardware.h>

This needs to be looked at more closely.  The fact that it requires
pxa2xx-regs makes this driver PXA2 only, something which needs to be
reviewed.

If the only reason for the driver being PXA2 only is that its touching
RCSR, then that needs to be addressed.

> diff --git a/drivers/input/keyboard/tosakbd.c b/drivers/input/keyboard/tosakbd.c
> index 94e444b..008ed26 100644
> --- a/drivers/input/keyboard/tosakbd.c
> +++ b/drivers/input/keyboard/tosakbd.c
> @@ -21,6 +21,7 @@
>  
>  #include <asm/arch/gpio.h>
>  #include <asm/arch/tosa.h>
> +#include <asm/arch/pxa2xx-regs.h>
>  
>  #define KB_ROWMASK(r)		(1 << (r))
>  #define SCANCODE(r, c)		(((r)<<4) + (c) + 1)

Builds fine here - maybe resolved by the remainder of the PXA development
being merged?

> diff --git a/arch/arm/mach-pxa/spitz.c b/arch/arm/mach-pxa/spitz.c
> index e7d0fcd..5ce19d4 100644
> --- a/arch/arm/mach-pxa/spitz.c
> +++ b/arch/arm/mach-pxa/spitz.c
> @@ -46,6 +46,7 @@
>  #include <asm/arch/akita.h>
>  #include <asm/arch/spitz.h>
>  #include <asm/arch/sharpsl.h>
> +#include <asm/arch/pxa27x-udc.h>
>  
>  #include <asm/mach/sharpsl_param.h>
>  #include <asm/hardware/scoop.h>

Result of a badly placed commit which was then reverted, but the pxa tree
continuing to carry the commit (resulting in the pxa tree being buildable
but when merged into mainline, it breaking.)  Resolved by remainder of
the PXA development being merged and this fixed inside a merge commit.

So the only one remaining is the sa1100 watchdog, which as I say needs
closer inspection.

One of the reasons for moving PXA2 register definitions to pxa2xx-regs.h
is to help identify drivers which shouldn't be PXA2 specific, so adding
this include to them as a "fix" is not necessarily the right solution.

-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:        http://www.arm.linux.org.uk/mailinglists/faq.php
Etiquette:  http://www.arm.linux.org.uk/mailinglists/etiquette.php