2017-07-20 15:52:44

by Arnd Bergmann

[permalink] [raw]
Subject: [PATCH 04/12] ARM: sa1100/pxa: fix MTD_XIP build

In commit 3169663ac5902 "ARM: sa11x0/pxa: convert OS timer registers
to IOMEM", the definition of the OSCR macro was changed to be an
__iomem pointer, but the same register is also used by the XIP
code. This patch does the corresponding change here as well.

On PXA, the IRQ register definitions were removed even earlier, in
commit 5d284e353eb1 ("ARM: pxa: avoid accessing interrupt registers
directly"). This patch unfortunately brings some of that back. An
earlier version of my patch moved the code into an external function,
which could not work for CONFIG_XIP_KERNEL+CONFIG_MTD_XIP, so this
restores something close to the original code.

Link: http://lists.infradead.org/pipermail/linux-arm-kernel/2014-March/241716.html
Signed-off-by: Arnd Bergmann <[email protected]>
Cc: Russell King <[email protected]>
---
---
arch/arm/mach-pxa/include/mach/mtd-xip.h | 10 +++++++---
arch/arm/mach-sa1100/include/mach/mtd-xip.h | 4 ++--
2 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/arch/arm/mach-pxa/include/mach/mtd-xip.h b/arch/arm/mach-pxa/include/mach/mtd-xip.h
index 990d2bf2fb45..f0d40bd36bac 100644
--- a/arch/arm/mach-pxa/include/mach/mtd-xip.h
+++ b/arch/arm/mach-pxa/include/mach/mtd-xip.h
@@ -17,11 +17,15 @@

#include <mach/regs-ost.h>

-#define xip_irqpending() (ICIP & ICMR)
+/* restored July 2017, this did not build since 2011! */
+
+#define ICIP io_p2v(0x40d00000)
+#define ICMR io_p2v(0x40d00004)
+#define xip_irqpending() readl(ICIP) & readl(ICMR)

/* we sample OSCR and convert desired delta to usec (1/4 ~= 1000000/3686400) */
-#define xip_currtime() (OSCR)
-#define xip_elapsed_since(x) (signed)((OSCR - (x)) / 4)
+#define xip_currtime() readl(OSCR)
+#define xip_elapsed_since(x) (signed)((readl(OSCR) - (x)) / 4)

/*
* xip_cpu_idle() is used when waiting for a delay equal or larger than
diff --git a/arch/arm/mach-sa1100/include/mach/mtd-xip.h b/arch/arm/mach-sa1100/include/mach/mtd-xip.h
index b3d684098fbf..cb76096a2e36 100644
--- a/arch/arm/mach-sa1100/include/mach/mtd-xip.h
+++ b/arch/arm/mach-sa1100/include/mach/mtd-xip.h
@@ -20,7 +20,7 @@
#define xip_irqpending() (ICIP & ICMR)

/* we sample OSCR and convert desired delta to usec (1/4 ~= 1000000/3686400) */
-#define xip_currtime() (OSCR)
-#define xip_elapsed_since(x) (signed)((OSCR - (x)) / 4)
+#define xip_currtime() readl_relaxed(OSCR)
+#define xip_elapsed_since(x) (signed)((readl_relaxed(OSCR) - (x)) / 4)

#endif /* __ARCH_SA1100_MTD_XIP_H__ */
--
2.9.0


2017-07-23 15:53:28

by Robert Jarzmik

[permalink] [raw]
Subject: Re: [PATCH 04/12] ARM: sa1100/pxa: fix MTD_XIP build

Arnd Bergmann <[email protected]> writes:

Hi Arnd,

> -#define xip_irqpending() (ICIP & ICMR)
> +/* restored July 2017, this did not build since 2011! */
> +
> +#define ICIP io_p2v(0x40d00000)
> +#define ICMR io_p2v(0x40d00004)
Okay, I suppose the IO mapping is guaranteed to work, ie. io_p2v() is behaving
correctly when xip_irqpending() is used, right ? I'm not challenging this, I'm
just ensuring this _could_ work (if anybody had the silly idea to make it work
again, and I admit I don't have that much courage).

> +#define xip_irqpending() readl(ICIP) & readl(ICMR)
This is not strictly equivalent to (ICIP & ICMR), I would have put for priority
reasons :
+#define xip_irqpending() (readl(ICIP) & readl(ICMR))

..zip..

Cheers.

--
Robert

2017-07-25 15:14:59

by Arnd Bergmann

[permalink] [raw]
Subject: Re: [PATCH 04/12] ARM: sa1100/pxa: fix MTD_XIP build

On Sun, Jul 23, 2017 at 5:53 PM, Robert Jarzmik <[email protected]> wrote:
> Arnd Bergmann <[email protected]> writes:
>
> Hi Arnd,
>
>> -#define xip_irqpending() (ICIP & ICMR)
>> +/* restored July 2017, this did not build since 2011! */
>> +
>> +#define ICIP io_p2v(0x40d00000)
>> +#define ICMR io_p2v(0x40d00004)
> Okay, I suppose the IO mapping is guaranteed to work, ie. io_p2v() is behaving
> correctly when xip_irqpending() is used, right ? I'm not challenging this, I'm
> just ensuring this _could_ work (if anybody had the silly idea to make it work
> again, and I admit I don't have that much courage).

Good thinking ;-)

I double-checked this and found that it is correct, and we still map that memory
in the same place in pxa_map_io().

>> +#define xip_irqpending() readl(ICIP) & readl(ICMR)
> This is not strictly equivalent to (ICIP & ICMR), I would have put for priority
> reasons :
> +#define xip_irqpending() (readl(ICIP) & readl(ICMR))

Ok, I've changed that and will resend.

Arnd