[PATCH] ARM: shmobile: fix MMCIF IRQ assignment on kzm9g

[PATCH] ARM: shmobile: fix MMCIF IRQ assignment on kzm9g

[Guennadi Liakhovetski]

Error and operation IRQs are swapped in kzm9g platform data, this patch
fixes their order. The only problem this bug causes is the confusing IRQ 
count in /proc/interrupts output, otherwise functionality is unaffected, 
since the driver doesn't really differentiate between the two IRQs.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---

The error is easy to see in /proc/interrupts output, kota2 has the correct 
IRQ assignment, ag5evm has the same problem, as kzm9g, so, it might need 
fixing too. I certainly could cook up a similar trivial patch for it, but 
someone will have to test it.

 arch/arm/mach-shmobile/board-kzm9g.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-shmobile/board-kzm9g.c b/arch/arm/mach-shmobile/board-kzm9g.c
index 29cca80..21c90c4 100644
--- a/arch/arm/mach-shmobile/board-kzm9g.c
+++ b/arch/arm/mach-shmobile/board-kzm9g.c
@@ -346,11 +346,11 @@ static struct resource sh_mmcif_resources[] = {
 		.flags	= IORESOURCE_MEM,
 	},
 	[1] = {
-		.start	= gic_spi(141),
+		.start	= gic_spi(140),
 		.flags	= IORESOURCE_IRQ,
 	},
 	[2] = {
-		.start	= gic_spi(140),
+		.start	= gic_spi(141),
 		.flags	= IORESOURCE_IRQ,
 	},
 };
-- 
1.7.2.5

Re: [PATCH] ARM: shmobile: fix MMCIF IRQ assignment on kzm9g

[Simon Horman]

On Wed, Sep 19, 2012 at 12:52:26AM +0200, Guennadi Liakhovetski wrote:
> Error and operation IRQs are swapped in kzm9g platform data, this patch
> fixes their order. The only problem this bug causes is the confusing IRQ 
> count in /proc/interrupts output, otherwise functionality is unaffected, 
> since the driver doesn't really differentiate between the two IRQs.
> 
> Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>

Thanks Guennadi,

as it happens I merged a very similar (same?) patch into the
fixes branch of my renesas tree yesterday. It has been accepted
by the arm-soc people and should appear in 3.6.

Re: [PATCH] ARM: shmobile: fix MMCIF IRQ assignment on kzm9g

[Guennadi Liakhovetski]

Hi Simon

On Wed, 19 Sep 2012, Simon Horman wrote:

> On Wed, Sep 19, 2012 at 12:52:26AM +0200, Guennadi Liakhovetski wrote:
> > Error and operation IRQs are swapped in kzm9g platform data, this patch
> > fixes their order. The only problem this bug causes is the confusing IRQ 
> > count in /proc/interrupts output, otherwise functionality is unaffected, 
> > since the driver doesn't really differentiate between the two IRQs.
> > 
> > Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
> 
> Thanks Guennadi,
> 
> as it happens I merged a very similar (same?) patch into the
> fixes branch of my renesas tree yesterday. It has been accepted
> by the arm-soc people and should appear in 3.6.

Good, just wondering what about ag5evm, it does seem to have the same 
problem. As mentioned before, this is just a "cosmetic" fix, so, up to you 
to decide when and how and if you'd like to fix it.

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

Re: [PATCH] ARM: shmobile: fix MMCIF IRQ assignment on kzm9g

[Simon Horman]

On Wed, Sep 19, 2012 at 11:08:52PM +0200, Guennadi Liakhovetski wrote:
> Hi Simon
> 
> On Wed, 19 Sep 2012, Simon Horman wrote:
> 
> > On Wed, Sep 19, 2012 at 12:52:26AM +0200, Guennadi Liakhovetski wrote:
> > > Error and operation IRQs are swapped in kzm9g platform data, this patch
> > > fixes their order. The only problem this bug causes is the confusing IRQ 
> > > count in /proc/interrupts output, otherwise functionality is unaffected, 
> > > since the driver doesn't really differentiate between the two IRQs.
> > > 
> > > Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
> > 
> > Thanks Guennadi,
> > 
> > as it happens I merged a very similar (same?) patch into the
> > fixes branch of my renesas tree yesterday. It has been accepted
> > by the arm-soc people and should appear in 3.6.
> 
> Good, just wondering what about ag5evm, it does seem to have the same 
> problem. As mentioned before, this is just a "cosmetic" fix, so, up to you 
> to decide when and how and if you'd like to fix it.

Good point.

Could you prepare a patch just for ag5evm for me to queue up?