ARM: dt-binding: fix GIC example

PATCH New Lucas Stach 3 participants 9 messages Mar 28, 2014 – Apr 11, 2014

Subsystems (2)

Device tree bindings Arm i.MX and Layerscape

lore

Share & tools

lore

Lucas Stach

author

2014-03-28 16:25 UTC #0 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

From: Tim Harvey <[email protected]>

When using interrupt-maps, the size of a map entry is #address-cells + address-cells should be 0 as this is not used.

As #address-cells = <0> is the same as not specifying the property, simply remove it.

Signed-off-by Tim Harvey <[email protected]>

Signed-off-by Lucas Stach <[email protected]>

---
v2 lst:
- completely remove the property

---
 Documentation/devicetree/bindings/arm/gic.txt | 1 -
 1 file changed, 1 deletion(-)

Documentation/devicetree/bindings/arm/gic.txt +0 -1

index bae0d87a38b2..42ab3b28a266 100644
@@ -55,7 +55,6 @@ Example:
 	intc: interrupt-controller at fff11000 {
 		compatible = "arm,cortex-a9-gic";
 		#interrupt-cells = <3>;
-		#address-cells = <1>;
 		interrupt-controller;
 		reg = <0xfff11000 0x1000>,
 		      <0xfff10100 0x100>;

Signature · 1 line

1.9.0

Lucas Stach

author

2014-03-28 16:25 UTC #1 lore raw

PATCH 2/2 ARM: dts: imx: drop invalid size and address cells properties

Those two properties should have been set to zero, which is the same as not specifying them.

Having address-cells set to 1 causes OF interrupt mapping routines to add 1 to the interrupt-cells property and as result fail because all calculations are off by one.

Signed-off-by Lucas Stach <[email protected]>

---
 arch/arm/boot/dts/imx6qdl.dtsi | 2 --
 arch/arm/boot/dts/imx6sl.dtsi  | 2 --
 2 files changed, 4 deletions(-)

arch/arm/boot/dts/imx6qdl.dtsi +0 -2

index fb28b2ecb1db..31bd844798a7 100644
@@ -38,8 +38,6 @@
 	intc: interrupt-controller at 00a01000 {
 		compatible = "arm,cortex-a9-gic";
 		#interrupt-cells = <3>;
-		#address-cells = <1>;
-		#size-cells = <1>;
 		interrupt-controller;
 		reg = <0x00a01000 0x1000>,
 		      <0x00a00100 0x100>;

arch/arm/boot/dts/imx6sl.dtsi +0 -2

index 28558f1aaf2d..276425089ea0 100644
@@ -44,8 +44,6 @@
 	intc: interrupt-controller at 00a01000 {
 		compatible = "arm,cortex-a9-gic";
 		#interrupt-cells = <3>;
-		#address-cells = <1>;
-		#size-cells = <1>;
 		interrupt-controller;
 		reg = <0x00a01000 0x1000>,
 		      <0x00a00100 0x100>;

Signature · 1 line

1.9.0

Shawn Guo

Applied

2014-04-03 06:47 UTC #2 lore raw

PATCH 2/2 ARM: dts: imx: drop invalid size and address cells properties

Quoted 9 lines · Lucas Stach

Those two properties should have been set to zero, which is the same as not specifying them.
Having address-cells set to 1 causes OF interrupt mapping routines to add 1 to the interrupt-cells property and as result fail because all calculations are off by one.
Signed-off-by Lucas Stach <[email protected]>

Applied, thanks.

Rob Herring

2014-04-10 15:21 UTC #3 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

In reply to Lucas Stach

Quoted 3 lines · Lucas Stach

From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +

Plus what? You mean ", and "?

Quoted 7 lines

address-cells should be 0 as this is not used.
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>

Did this get applied? I can't seem to find patch 2.

Rob

Shawn Guo

2014-04-11 01:16 UTC #4 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

Quoted 9 lines · Rob Herring

Quoted 3 lines · Lucas Stach
From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +
Plus what? You mean ", and "?
Quoted 2 lines
address-cells should be 0 as this is not used.

Quoted 7 lines

Quoted 5 lines
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>
Did this get applied? I can't seem to find patch 2.

No. We expect this patch to be applied on DT tree, while the patch #2 [1] is going through IMX tree.

Shawn

[1] http://www.spinics.net/lists/arm-kernel/msg318810.html

Lucas Stach

author

2014-04-11 10:10 UTC #5 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

In reply to Rob Herring

Hi Rob,

Am Donnerstag, den 10.04.2014, 10:21 -0500 schrieb Rob Herring:

Quoted 7 lines · Lucas Stach

Quoted 3 lines
From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +
Plus what? You mean ", and "?

Urgh, I obviously messed this up while editing the commit message in the last minute. It should have been something along the lines of:

When using interrupt-maps, the size of a map entry is #address-cells + #interrupt-cells. address-cells should be 0 as this offset is not used in the GIC.

As #address-cells = <0> is the same as not specifying the property, simply remove it.

Can you edit this while applying, or should I send a new patch?

Quoted 10 lines

Quoted 7 lines
address-cells should be 0 as this is not used.
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>
Did this get applied? I can't seem to find patch 2.

As Shawn pointed out patch 2 got applied through the imx tree as it fixes a real bug there.

Regards,
Lucas

Signature · 4 lines

Pengutronix e.K.                           | Lucas Stach                 |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-5076 |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Rob Herring

2014-04-11 13:20 UTC #6 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

Quoted 13 lines · Lucas Stach

Hi Rob,
Am Donnerstag, den 10.04.2014, 10:21 -0500 schrieb Rob Herring:
Quoted 7 lines · Lucas Stach
Quoted 3 lines
From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +
Plus what? You mean ", and "?
Urgh, I obviously messed this up while editing the commit message in the last minute. It should have been something along the lines of:

Quoted 8 lines

When using interrupt-maps, the size of a map entry is #address-cells + #interrupt-cells. address-cells should be 0 as this offset is not used in the GIC.
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Can you edit this while applying, or should I send a new patch?

I'll fix it up and apply.

Quoted 6 lines

Quoted 6 lines
Quoted 6 lines
address-cells should be 0 as this is not used.
As #address-cells = <0> is the same as not specifying the property, simply remove it.

Quoted 7 lines

Quoted 5 lines
Quoted 2 lines
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>
Did this get applied? I can't seem to find patch 2.
As Shawn pointed out patch 2 got applied through the imx tree as it fixes a real bug there.

It was highlighted that other instances beyond iMX were wrong. Please fix those as well.

Rob

Lucas Stach

author

2014-04-11 13:23 UTC #7 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

Am Freitag, den 11.04.2014, 08:20 -0500 schrieb Rob Herring:

Quoted 18 lines · Lucas Stach

Quoted 17 lines
Hi Rob,
Am Donnerstag, den 10.04.2014, 10:21 -0500 schrieb Rob Herring:
Quoted 7 lines · Lucas Stach
Quoted 3 lines
From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +
Plus what? You mean ", and "?
Urgh, I obviously messed this up while editing the commit message in the last minute. It should have been something along the lines of:
When using interrupt-maps, the size of a map entry is #address-cells + #interrupt-cells. address-cells should be 0 as this offset is not used in the GIC.

Quoted 7 lines

Quoted 4 lines
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Can you edit this while applying, or should I send a new patch?
I'll fix it up and apply.

Thanks.

Quoted 9 lines

Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
address-cells should be 0 as this is not used.
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>

Quoted 7 lines

Quoted 4 lines
Quoted 2 lines
Did this get applied? I can't seem to find patch 2.
As Shawn pointed out patch 2 got applied through the imx tree as it fixes a real bug there.
It was highlighted that other instances beyond iMX were wrong. Please fix those as well.

Oh, I wasn't aware of that. Will send patches.

Regards,
Lucas

Signature · 4 lines

Pengutronix e.K.                           | Lucas Stach                 |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-5076 |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Rob Herring

2014-04-11 13:40 UTC #8 lore raw

PATCH 1/2 ARM: dt-binding: fix GIC example

Quoted 39 lines · Lucas Stach

Am Freitag, den 11.04.2014, 08:20 -0500 schrieb Rob Herring:
Quoted 18 lines · Lucas Stach
Quoted 17 lines
Hi Rob,
Am Donnerstag, den 10.04.2014, 10:21 -0500 schrieb Rob Herring:
Quoted 7 lines · Lucas Stach
Quoted 3 lines
From: Tim Harvey <[email protected]>
When using interrupt-maps, the size of a map entry is #address-cells +
Plus what? You mean ", and "?
Urgh, I obviously messed this up while editing the commit message in the last minute. It should have been something along the lines of:
When using interrupt-maps, the size of a map entry is #address-cells + #interrupt-cells. address-cells should be 0 as this offset is not used in the GIC.
Quoted 7 lines
Quoted 4 lines
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Can you edit this while applying, or should I send a new patch?
I'll fix it up and apply.
Thanks.
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 8 lines
address-cells should be 0 as this is not used.
As #address-cells = <0> is the same as not specifying the property, simply remove it.
Signed-off-by Tim Harvey <[email protected]>
Signed-off-by Lucas Stach <[email protected]>
Did this get applied? I can't seem to find patch 2.

Quoted 7 lines

Quoted 5 lines
Quoted 2 lines
As Shawn pointed out patch 2 got applied through the imx tree as it fixes a real bug there.
It was highlighted that other instances beyond iMX were wrong. Please fix those as well.
Oh, I wasn't aware of that. Will send patches.

It was raised here: http://www.spinics.net/lists/linux-pci/msg29768.html

Here's a somewhat accurate list:

$ git grep -A8 'arm,.*gic' origin/master  -- arch/arm/boot/dts | grep
'address.*1'
origin/master:arch/arm/boot/dts/ecx-2000.dts-
#address-cells = <1>;
origin/master:arch/arm/boot/dts/highbank.dts-
#address-cells = <1>;
origin/master:arch/arm/boot/dts/imx6qdl.dtsi-           #address-cells = <1>;
origin/master:arch/arm/boot/dts/imx6sl.dtsi-            #address-cells = <1>;
origin/master:arch/arm/boot/dts/keystone.dtsi-          #address-cells = <1>;
origin/master:arch/arm/boot/dts/r8a7740.dtsi-           #address-cells = <1>;
origin/master:arch/arm/boot/dts/sh73a0.dtsi-            #address-cells = <1>;
origin/master:arch/arm/boot/dts/socfpga.dtsi-           #address-cells = <1>;
origin/master:arch/arm/boot/dts/ste-dbx5x0.dtsi-
 #address-cells = <1>;
origin/master:arch/arm/boot/dts/vf610.dtsi-
 #address-cells = <1>;
origin/master:arch/arm/boot/dts/zynq-7000.dtsi-
#address-cells = <1>;

Rob