Skip to contentSkip to conversation

ARM: OMAP3630: Add generic machine descriptor

RFC PATCH In review Nishanth Menon 5 participants 20 messages Sep 20, 2013 – Oct 8, 2013

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 ARM: OMAP3630: Add generic machine descriptor

Add generic machine which can handle inits for omap3630. Currently beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early

This defeats the purpose of having SoC specific init_early handlers,
example: clock nodes are not the same between 3430 and 3630.

---

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

 arch/arm/mach-omap2/board-generic.c |   18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
arch/arm/mach-omap2/board-generic.c +18 -0
index 39c7838..cd85b36 100644
@@ -147,6 +147,24 @@ DT_MACHINE_START(OMAP3_GP_DT, "Generic OMAP3-GP (Flattened Device Tree)")
 	.dt_compat	= omap3_gp_boards_compat,
 	.restart	= omap3xxx_restart,
 MACHINE_END
+
+static const char *omap3630_gp_boards_compat[] __initdata = {
+	"ti,omap3-beagle-xm",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP3630_GP_DT, "Generic OMAP3630-GP (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_secure_sync32k_timer_init,
+	.dt_compat	= omap3630_gp_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
 #endif
 
 #ifdef CONFIG_SOC_AM33XX
Signature · 1 line
1.7.9.5
#1 lore raw

Hi,

Quoted 4 lines · Nishanth Menon

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

Yes, this would be better, but you can do add a DT_MACHINE as in this patch but have ti,omap3630 as the dt_compat table. Then there's no need to add runtime checks.

-Olof

#2 lore raw

Hi,

Quoted 9 lines · Olof Johansson
Quoted 4 lines · Nishanth Menon

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

Yes, this would be better, but you can do add a DT_MACHINE as in this patch but have ti,omap3630 as the dt_compat table. Then there's no need to add runtime checks.

I was going to reply that adding of_machine_is_compatible("ti,omap3630") would help in some situations, but guess it's already tainted ;-)

cheers

Signature · 8 lines
balbi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20130920/8516cdaf/attachment.sig>
#3 lore raw
Quoted 9 lines · Felipe Balbi

Hi,

Quoted 6 lines · Olof Johansson
Quoted 4 lines · Nishanth Menon

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

Quoted 6 lines
Quoted 3 lines

Yes, this would be better, but you can do add a DT_MACHINE as in this patch but have ti,omap3630 as the dt_compat table. Then there's no need to add runtime checks.

I was going to reply that adding of_machine_is_compatible("ti,omap3630") would help in some situations, but guess it's already tainted ;-)

Oh, if it's just a few checks, then by all means go down that route. I didn't look at the code to see how much it would be.

But if a new DT_MACHINE is added, then it should definitely be based on ti,omap3630 instead of listing all the boards.

-Olof

#4 lore raw

Hi,

Quoted 11 lines · Olof Johansson
Quoted 11 lines · Olof Johansson
Quoted 9 lines · Nishanth Menon
Quoted 4 lines

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

Yes, this would be better, but you can do add a DT_MACHINE as in this patch but have ti,omap3630 as the dt_compat table. Then there's no need to add runtime checks.

Quoted 8 lines
Quoted 2 lines

I was going to reply that adding of_machine_is_compatible("ti,omap3630") would help in some situations, but guess it's already tainted ;-)

Oh, if it's just a few checks, then by all means go down that route. I didn't look at the code to see how much it would be.

But if a new DT_MACHINE is added, then it should definitely be based on ti,omap3630 instead of listing all the boards.

the idea was to CPU compatible property to conditionally enable known erratas workarounds. In some cases, Revision register can't be trusted, so instead of creating per-errata DT properties (since that'd be describing the SW, in a way), I thought of using of_machine_is_compatible() checks, but that assumes CPU compatible is "correct".

cheers

Signature · 8 lines
balbi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20130920/9cfa9366/attachment.sig>
#5 lore raw
Quoted 22 lines · Felipe Balbi

Hi,

Quoted 19 lines · Olof Johansson
Quoted 7 lines · Olof Johansson
Quoted 6 lines · Nishanth Menon
Quoted 4 lines

An alternative approach may be to (for all SoCs):
1. define every SoC entry - ti,omap3430 ti,omap3630...
2. have a generic omap3_init which uses "if (of_machine_is_compatible("ti,omap3630"))" to invoke the appropriate omap3xxx_init_early.

Quoted 6 lines
Quoted 3 lines

Yes, this would be better, but you can do add a DT_MACHINE as in this patch but have ti,omap3630 as the dt_compat table. Then there's no need to add runtime checks.

I was going to reply that adding of_machine_is_compatible("ti,omap3630") would help in some situations, but guess it's already tainted ;-)

Oh, if it's just a few checks, then by all means go down that route. I didn't look at the code to see how much it would be.

But if a new DT_MACHINE is added, then it should definitely be based on ti,omap3630 instead of listing all the boards.

This was more in the direction I was hoping to go.

Quoted 7 lines

the idea was to CPU compatible property to conditionally enable known erratas workarounds. In some cases, Revision register can't be trusted, so instead of creating per-errata DT properties (since that'd be describing the SW, in a way), I thought of using of_machine_is_compatible() checks, but that assumes CPU compatible is "correct".

I think the quirk handling is part of what Tony is attempting, and the definition of these might help, I think..

Signature · 2 lines
Regards,
Nishanth Menon
#6 lore raw

cover ARM: dts: OMAP: standardize SoC specific bindings

This is a respin of [1] based on of Olof's comments to introduce a generic SoC binding. This standardizes the binding definitions for SoCs based on existing "implied" bindings and based on existing usage in arch/arm/mach-omap2/soc.h. Eventually we should be able to get rid of soc_is_xyz() functions and allow machine descriptors to seamlessly handle the deltas.

The series is based on [2] and was triggered primarily due to the bug seen with [3] - clock dts conversion.

Nishanth Menon (2):
Documentation: dt: OMAP: standardize SoC naming definition
ARM: dts: omap3-beagle: use 3630 definitions

 .../devicetree/bindings/arm/omap/omap.txt          |   45 ++++++++++++++++++++
 arch/arm/boot/dts/omap3-beagle-xm.dts              |    2 +-
 arch/arm/mach-omap2/board-generic.c                |   23 ++++++++++
 3 files changed, 69 insertions(+), 1 deletion(-)

[1] https://patchwork.kernel.org/patch/2919661/
[2] https://git.kernel.org/cgit/linux/kernel/git/bcousson/linux-omap-dt.git/log/?h=for_3.13/dts
[3] http://marc.info/?t=138009899400001&r=1&w=2

Signature · 1 line
1.7.9.5
#7 lore raw

PATCH 1/2 Documentation: dt: OMAP: standardize SoC naming definition

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

Eventually, we will have descriptors match only with SoC types and should not contain anything specific to board handling.

---
 .../devicetree/bindings/arm/omap/omap.txt          |   45 ++++++++++++++++++++
 arch/arm/mach-omap2/board-generic.c                |   23 ++++++++++
 2 files changed, 68 insertions(+)
Documentation/devicetree/bindings/arm/omap/omap.txt +45 -0
index 91b7049..28b20a2 100644
@@ -30,6 +30,51 @@ spinlock at 1 {
     ti,hwmods = "spinlock";
 };
 
+SoC Type(optional):
+- ti,gp	- General Purpose devices
+- ti,hs	- High Security devices
+
+SoC Families:
+
+- OMAP2 generic - defaults to OMAP2420
+  compatible = "ti,omap2"
+- OMAP3 generic - defaults to OMAP3430
+  compatible = "ti,omap3"
+- OMAP4 generic - defaults to OMAP4430
+  compatible = "ti,omap4"
+- OMAP5 generic - defaults to OMAP5430
+  compatible = "ti,omap5"
+- DRA7 generic - defaults to DRA742
+  compatible = "ti,dra7"
+- AM43x generic - defaults to AM437x
+  compatible = "ti,am43"
+
+SoCs:
+
+- OMAP2420
+  compatible = "ti,omap2420", "ti,omap2"
+- OMAP2430
+  compatible = "ti,omap2430", "ti,omap2"
+
+- OMAP3430
+  compatible = "ti,omap343x", "ti,omap3"
+- OMAP3630
+  compatible = "ti,omap363x", "ti,omap3"
+- AM33xx
+  compatible = "ti,am33xx", "ti,omap3"
+
+- OMAP4430
+  compatible = "ti,omap443x", "ti,omap4"
+- OMAP4460
+  compatible = "ti,omap446x", "ti,omap4"
+
+- OMAP5430
+  compatible = "ti,omap5430", "ti,omap5"
+- OMAP5432
+  compatible = "ti,omap5432", "ti,omap5"
+
+- DRA742
+  compatible = "ti,dra7xx", "ti,dra7"
 
 Boards:
 
arch/arm/mach-omap2/board-generic.c +23 -0
index 39c7838..c78b070 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
 	NULL,
 };
 
@@ -129,6 +130,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap363x",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP36xx_DT, "Generic OMAP363x (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
+
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",
@@ -171,6 +190,8 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP4
 static const char *omap4_boards_compat[] __initdata = {
 	"ti,omap4",
+	"ti,omap4430",
+	"ti,omap4460",
 	NULL,
 };
 
@@ -191,6 +212,8 @@ MACHINE_END
 #ifdef CONFIG_SOC_OMAP5
 static const char *omap5_boards_compat[] __initdata = {
 	"ti,omap5",
+	"ti,omap5430",
+	"ti,omap5432",
 	NULL,
 };
 
Signature · 1 line
1.7.9.5
maintainer
#8 lore raw

Re: [PATCH 1/2] Documentation: dt: OMAP: standardize SoC naming definition

Hi,

Few comments below.

Quoted 4 lines · Nishanth Menon

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

It seems that we should try to queue this as a fix to avoid debugging it multiple times as it's actually quite easy to hit this one.

So maybe use a subject along lines of:

"ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree"

And also include these warnings you can get:

omap_hwmod: uart4: cannot clk_get main_clk uart4_fck
omap_hwmod: uart4: cannot _init_clocks

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2434 _init+0x6c/0x80()

omap_hwmod: uart4: couldn't init clocks
...

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126 _enable+0x254/0x280()

omap_hwmod: timer12: enabled state can only be entered from initialized, idle, or disabled state
...

WARNING: CPU: 0 PID: 46 at arch/arm/mach-omap2/omap_hwmod.c:2224 _idle+0xd4/0xf8()

omap_hwmod: timer12: idle state can only be entered from enabled state

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126 _enable+0x254/0x280()

omap_hwmod: uart4: enabled state can only be entered from initialized, idle, or disabled state

And also describe that the outcome is that the system fails to boot.

Quoted 43 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +37 -0
@@ -30,6 +30,51 @@ spinlock at 1 {
     ti,hwmods = "spinlock";
 };
 
+SoC Type(optional):
+- ti,gp	- General Purpose devices
+- ti,hs	- High Security devices
+
+SoC Families:
+
+- OMAP2 generic - defaults to OMAP2420
+  compatible = "ti,omap2"
+- OMAP3 generic - defaults to OMAP3430
+  compatible = "ti,omap3"
+- OMAP4 generic - defaults to OMAP4430
+  compatible = "ti,omap4"
+- OMAP5 generic - defaults to OMAP5430
+  compatible = "ti,omap5"
+- DRA7 generic - defaults to DRA742
+  compatible = "ti,dra7"
+- AM43x generic - defaults to AM437x
+  compatible = "ti,am43"
+
+SoCs:
+
+- OMAP2420
+  compatible = "ti,omap2420", "ti,omap2"
+- OMAP2430
+  compatible = "ti,omap2430", "ti,omap2"
+
+- OMAP3430
+  compatible = "ti,omap343x", "ti,omap3"
+- OMAP3630
+  compatible = "ti,omap363x", "ti,omap3"
+- AM33xx
+  compatible = "ti,am33xx", "ti,omap3"
+
+- OMAP4430
+  compatible = "ti,omap443x", "ti,omap4"
+- OMAP4460
+  compatible = "ti,omap446x", "ti,omap4"

Documentation/devicetree/bindings/arm/omap/omap.txt spinlock at 1 { · line 79

Quoted 10 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +8 -0
+
+- OMAP5430
+  compatible = "ti,omap5430", "ti,omap5"
+- OMAP5432
+  compatible = "ti,omap5432", "ti,omap5"
+
+- DRA742
+  compatible = "ti,dra7xx", "ti,dra7"
 
 Boards:

And I would leave the documentation parts out of the fix as we're pretty late into the -rc cycle.

Quoted 10 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +2 -0
+++ b/arch/arm/mach-omap2/board-generic.c
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
 	NULL,
 };
 

Documentation/devicetree/bindings/arm/omap/omap.txt DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)") · line 136

Quoted 7 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +4 -0
@@ -129,6 +130,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap363x",
+	NULL,
+};

Why not make it just "ti,omap36xx"? This also applies to 3703 where the 3 is missing as it does not have the DSP.

Documentation/devicetree/bindings/arm/omap/omap.txt DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)") · line 137

Quoted 1 line
Documentation/devicetree/bindings/arm/omap/omap.txt +1 -0
+DT_MACHINE_START(OMAP36xx_DT, "Generic OMAP363x (Flattened Device Tree)")

Upper case OMAP36XX_DT here please like the others have.

Documentation/devicetree/bindings/arm/omap/omap.txt DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)") · line 145

Quoted 8 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +8 -0
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,

Looks like this version has the correct init_time function too :)

Documentation/devicetree/bindings/arm/omap/omap.txt DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)") · line 148

Quoted 3 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +3 -0
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END

This looks correct to me.

Quoted 13 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +2 -0
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",
@@ -171,6 +190,8 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP4
 static const char *omap4_boards_compat[] __initdata = {
 	"ti,omap4",
+	"ti,omap4430",
+	"ti,omap4460",
 	NULL,
 };
 

Documentation/devicetree/bindings/arm/omap/omap.txt MACHINE_END · line 218

Quoted 7 lines
Documentation/devicetree/bindings/arm/omap/omap.txt +2 -0
@@ -191,6 +212,8 @@ MACHINE_END
 #ifdef CONFIG_SOC_OMAP5
 static const char *omap5_boards_compat[] __initdata = {
 	"ti,omap5",
+	"ti,omap5430",
+	"ti,omap5432",
 	NULL,
 };

I would leave these parts for later, maybe with the documentation changes?

Regards,

Tony

#9 lore raw

PATCH 2/2 ARM: dts: omap3-beagle: use 3630 definitions

In reply to Nishanth Menon

beagle-xm currently would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. So add compatiblity for 3630 to allow match

---
 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0f7cfc5..2079e22 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3-beagle", "ti,omap3";
 
 	cpus {
 		cpu at 0 {
Signature · 1 line
1.7.9.5
maintainer
#10 lore raw

Re: [PATCH 2/2] ARM: dts: omap3-beagle: use 3630 definitions

Quoted 14 lines · Nishanth Menon

beagle-xm currently would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. So add compatiblity for 3630 to allow match

---
 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +0 -0
index 0f7cfc5..2079e22 100644

arch/arm/boot/dts/omap3-beagle-xm.dts line 17

Quoted 8 lines
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3-beagle", "ti,omap3";
 
 	cpus {
 		cpu at 0 {

This compatible string looks hacky.. How about just make it

"ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3";

How about just leave out "ti,omap3-beagle" here?

And I would fold this change into the fix part of your first patch in this series.

Regards,

Tony

#11 lore raw

PATCH v2 1/1 ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

With clock node dts conversion, we get the following warnings as a result and 3630 based platforms fails to boot (uart4 clocks are only present in OMAP3630 and not present in OMAP3430):

...
omap_hwmod: uart4: cannot clk_get main_clk uart4_fck
omap_hwmod: uart4: cannot _init_clocks

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2434
_init+0x6c/0x80()

omap_hwmod: uart4: couldn't init clocks
...

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: timer12: enabled state can only be entered from initialized, idle, or disabled state
...

WARNING: CPU: 0 PID: 46 at arch/arm/mach-omap2/omap_hwmod.c:2224
_idle+0xd4/0xf8()

omap_hwmod: timer12: idle state can only be entered from enabled state

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: uart4: enabled state can only be entered from initialized, idle, or disabled state

So, add specific compatiblity for 3630 to allow match for Beagle-XM platform.

---

Changes in V2 (since v1):
- based on v3.12-rc4 tag
- Update based on Tony's review comments.

V1: http://marc.info/?l=linux-omap&m=138117342909375&w=2

RFC: https://patchwork.kernel.org/patch/2919661/

 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 arch/arm/mach-omap2/board-generic.c   |   19 +++++++++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0c514dc..02dd4a9 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3";
 
 	cpus {
 		cpu at 0 {
arch/arm/mach-omap2/board-generic.c +19 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
 	NULL,
 };
 
@@ -129,6 +130,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap36xx",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP36XX_DT, "Generic OMAP363x (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
+
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",
Signature · 1 line
1.7.9.5
#12 lore raw

Re: [PATCH V2] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

On 10/07/2013 03:30 PM, Nishanth Menon wrote:
[...]

arch/arm/boot/dts/omap3-beagle-xm.dts line 14

Quoted 10 lines
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0c514dc..02dd4a9 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3";

^^^ yikes - should have been ti,omap36xx -> apologies, my fix was not committed in local branch :( - will update and resend - grr..

Signature · 2 lines
Regards,
Nishanth Menon
#13 lore raw

PATCH v3 1/1 ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

In reply to Tony Lindgren

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

With clock node dts conversion, we get the following warnings as a result and 3630 based platforms fails to boot (uart4 clocks are only present in OMAP3630 and not present in OMAP3430):

...
omap_hwmod: uart4: cannot clk_get main_clk uart4_fck
omap_hwmod: uart4: cannot _init_clocks

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2434
_init+0x6c/0x80()

omap_hwmod: uart4: couldn't init clocks
...

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: timer12: enabled state can only be entered from initialized, idle, or disabled state
...

WARNING: CPU: 0 PID: 46 at arch/arm/mach-omap2/omap_hwmod.c:2224
_idle+0xd4/0xf8()

omap_hwmod: timer12: idle state can only be entered from enabled state

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: uart4: enabled state can only be entered from initialized, idle, or disabled state

So, add specific compatiblity for 3630 to allow match for Beagle-XM platform.

---

Changes in V3 (since v2):
- Fix typo in dts :(

V2: https://patchwork.kernel.org/patch/2999111/
- based on v3.12-rc4 tag
- Update based on Tony's review comments.

V1: http://marc.info/?l=linux-omap&m=138117342909375&w=2

RFC: https://patchwork.kernel.org/patch/2919661/

 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 arch/arm/mach-omap2/board-generic.c   |   19 +++++++++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0c514dc..2816bf6 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap36xx", "ti,omap3";
 
 	cpus {
 		cpu at 0 {
arch/arm/mach-omap2/board-generic.c +19 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
 	NULL,
 };
 
@@ -129,6 +130,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap36xx",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP36XX_DT, "Generic OMAP36xx (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
+
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",
Signature · 1 line
1.7.9.5
#14 lore raw

Re: [PATCH V3] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

arch/arm/mach-omap2/board-generic.c MACHINE_END · line 116

Quoted 9 lines · Nishanth Menon
arch/arm/mach-omap2/board-generic.c +1 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
            ^^^^

You used omap36xx everywhere, so I guess this should be "ti,omap34xx"?

-- Sebastian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20131008/75bf54d3/attachment.sig>

#15 lore raw

Re: [PATCH V3] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

Quoted 11 lines · Sebastian Reichel

arch/arm/mach-omap2/board-generic.c MACHINE_END · line 116

Quoted 9 lines · Nishanth Menon
arch/arm/mach-omap2/board-generic.c +1 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
            ^^^^
Quoted 2 lines

You used omap36xx everywhere, so I guess this should be "ti,omap34xx"?

3630 has at least two variants:
http://www.ti.com/product/omap3621
http://www.ti.com/product/omap3630

It got further spun off in 37xx family:
http://www.ti.com/product/dm3730 (DM3730 and DM3725)

3430 is no longer in production - but just had a single version (at least what I can remember).

3530 variant of 3430 on the other hand has a few:
http://processors.wiki.ti.com/index.php/OMAP3_Overview

I am ok to change to ti,omap34xx if folks think that is the right thing to do. Personally, I might prefer to handle 35xx family slightly differently considering deltas.

Signature · 2 lines
Regards,
Nishanth Menon
maintainer
#16 lore raw

Re: [PATCH V3] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

Quoted 25 lines · Nishanth Menon
Quoted 11 lines · Sebastian Reichel

arch/arm/mach-omap2/board-generic.c MACHINE_END · line 116

Quoted 9 lines · Nishanth Menon
arch/arm/mach-omap2/board-generic.c +1 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
            ^^^^
Quoted 2 lines

You used omap36xx everywhere, so I guess this should be "ti,omap34xx"?

3630 has at least two variants:
http://www.ti.com/product/omap3621
http://www.ti.com/product/omap3630

It got further spun off in 37xx family:
http://www.ti.com/product/dm3730 (DM3730 and DM3725)

3430 is no longer in production - but just had a single version (at least what I can remember).

Quoted 7 lines

3530 variant of 3430 on the other hand has a few:
http://processors.wiki.ti.com/index.php/OMAP3_Overview

I am ok to change to ti,omap34xx if folks think that is the right thing to do. Personally, I might prefer to handle 35xx family slightly differently considering deltas.

I've dropped that part as that's not needed for the fix AFAIK. Here's what I've applied and pushed out to omap-for-v3.12/fixes.

Regards,

Tony

From: Nishanth Menon <[email protected]>
Date: Mon, 7 Oct 2013 15:43:49 -0500
Subject: [PATCH] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

With clock node dts conversion, we get the following warnings before system hangs as a result and 3630 based platforms fails to boot (uart4 clocks are only present in OMAP3630 and not present in OMAP3430):

...
omap_hwmod: uart4: cannot clk_get main_clk uart4_fck
omap_hwmod: uart4: cannot _init_clocks

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2434
_init+0x6c/0x80()

omap_hwmod: uart4: couldn't init clocks
...

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: timer12: enabled state can only be entered from initialized, idle, or disabled state
...

WARNING: CPU: 0 PID: 46 at arch/arm/mach-omap2/omap_hwmod.c:2224
_idle+0xd4/0xf8()

omap_hwmod: timer12: idle state can only be entered from enabled state

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: uart4: enabled state can only be entered from initialized, idle, or disabled state

So, add specific compatiblity for 3630 to allow match for Beagle-XM platform.

[tony at atomide.com: left out ti,omap343x, updated comments]
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0c514dc..2816bf6 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap36xx", "ti,omap3";
 
 	cpus {
 		cpu at 0 {
arch/arm/mach-omap2/board-generic.c +18 -0
index 39c7838..87162e1 100644
@@ -129,6 +129,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap36xx",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP36XX_DT, "Generic OMAP36xx (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
+
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",
#17 lore raw

Re: [PATCH V3] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

Quoted 119 lines · Tony Lindgren
Quoted 25 lines · Nishanth Menon
Quoted 11 lines · Sebastian Reichel

arch/arm/mach-omap2/board-generic.c MACHINE_END · line 116

Quoted 9 lines · Nishanth Menon
arch/arm/mach-omap2/board-generic.c +1 -0
index 39c7838..4fe5b9c 100644
@@ -113,6 +113,7 @@ MACHINE_END
 #ifdef CONFIG_ARCH_OMAP3
 static const char *omap3_boards_compat[] __initdata = {
 	"ti,omap3",
+	"ti,omap343x",
            ^^^^
Quoted 2 lines

You used omap36xx everywhere, so I guess this should be "ti,omap34xx"?

3630 has at least two variants:
http://www.ti.com/product/omap3621
http://www.ti.com/product/omap3630

It got further spun off in 37xx family:
http://www.ti.com/product/dm3730 (DM3730 and DM3725)

3430 is no longer in production - but just had a single version (at least what I can remember).

Quoted 7 lines

3530 variant of 3430 on the other hand has a few:
http://processors.wiki.ti.com/index.php/OMAP3_Overview

I am ok to change to ti,omap34xx if folks think that is the right thing to do. Personally, I might prefer to handle 35xx family slightly differently considering deltas.

I've dropped that part as that's not needed for the fix AFAIK. Here's what I've applied and pushed out to omap-for-v3.12/fixes.

Regards,

Tony

From: Nishanth Menon <[email protected]>
Date: Mon, 7 Oct 2013 15:43:49 -0500
Subject: [PATCH] ARM: OMAP3: Fix hardware detection for omap3630 when booted with device tree

SoC family definitions at the moment are reactive to board needs as a result, beagle-xm would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. Obviously, this is the wrong behavior.

With clock node dts conversion, we get the following warnings before system hangs as a result and 3630 based platforms fails to boot (uart4 clocks are only present in OMAP3630 and not present in OMAP3430):

...
omap_hwmod: uart4: cannot clk_get main_clk uart4_fck
omap_hwmod: uart4: cannot _init_clocks

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2434
_init+0x6c/0x80()

omap_hwmod: uart4: couldn't init clocks
...

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: timer12: enabled state can only be entered from initialized, idle, or disabled state
...

WARNING: CPU: 0 PID: 46 at arch/arm/mach-omap2/omap_hwmod.c:2224
_idle+0xd4/0xf8()

omap_hwmod: timer12: idle state can only be entered from enabled state

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2126
_enable+0x254/0x280()

omap_hwmod: uart4: enabled state can only be entered from initialized, idle, or disabled state

So, add specific compatiblity for 3630 to allow match for Beagle-XM platform.

[tony at atomide.com: left out ti,omap343x, updated comments]
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0c514dc..2816bf6 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap36xx", "ti,omap3";
 
 	cpus {
 		cpu at 0 {
arch/arm/mach-omap2/board-generic.c +11 -0
index 39c7838..87162e1 100644
@@ -129,6 +129,24 @@ DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)")
 	.restart	= omap3xxx_restart,
 MACHINE_END
 
+static const char *omap36xx_boards_compat[] __initdata = {
+	"ti,omap36xx",
+	NULL,
+};
+
+DT_MACHINE_START(OMAP36XX_DT, "Generic OMAP36xx (Flattened Device Tree)")
+	.reserve	= omap_reserve,
+	.map_io		= omap3_map_io,
+	.init_early	= omap3630_init_early,
+	.init_irq	= omap_intc_of_init,
+	.handle_irq	= omap3_intc_handle_irq,

arch/arm/mach-omap2/board-generic.c DT_MACHINE_START(OMAP3_DT, "Generic OMAP3 (Flattened Device Tree)") · line 152

Quoted 10 lines
arch/arm/mach-omap2/board-generic.c +7 -0
+	.init_machine	= omap_generic_init,
+	.init_late	= omap3_init_late,
+	.init_time	= omap3_sync32k_timer_init,
+	.dt_compat	= omap36xx_boards_compat,
+	.restart	= omap3xxx_restart,
+MACHINE_END
+
 static const char *omap3_gp_boards_compat[] __initdata = {
 	"ti,omap3-beagle",
 	"timll,omap3-devkit8000",

LGTM. Thanks for doing it.

Signature · 2 lines
Regards,
Nishanth Menon
#18 lore raw

Re: [PATCH 2/2] ARM: dts: omap3-beagle: use 3630 definitions

In reply to Tony Lindgren

Hi,

Quoted 21 lines · Tony Lindgren
Quoted 20 lines · Nishanth Menon

beagle-xm currently would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. So add compatiblity for 3630 to allow match

---
 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
index 0f7cfc5..2079e22 100644
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3-beagle", "ti,omap3";
 
Quoted 8 lines

arch/arm/boot/dts/omap3-beagle-xm.dts line 17

Quoted 2 lines
arch/arm/boot/dts/omap3-beagle-xm.dts +0 -0
 	cpus {
 		cpu at 0 {

This compatible string looks hacky.. How about just make it

"ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3";

How about just leave out "ti,omap3-beagle" here?

ti,omap3-beagle was already part of the original file, we can definitely remove but I'd expect us to maintain support for that string indefinitely should anybody continue to use older dtbs.

Signature · 8 lines
balbi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20131008/618857fb/attachment.sig>
#19 lore raw

Re: [PATCH 2/2] ARM: dts: omap3-beagle: use 3630 definitions

Quoted 29 lines · Felipe Balbi

Hi,

Quoted 26 lines · Tony Lindgren
Quoted 14 lines · Nishanth Menon

beagle-xm currently would matchup with ti,omap3 which invokes omap3430_init_early instead of omap3630_init_early. So add compatiblity for 3630 to allow match

---
 arch/arm/boot/dts/omap3-beagle-xm.dts |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/boot/dts/omap3-beagle-xm.dts +0 -0
index 0f7cfc5..2079e22 100644

arch/arm/boot/dts/omap3-beagle-xm.dts line 17

Quoted 8 lines
arch/arm/boot/dts/omap3-beagle-xm.dts +1 -1
@@ -11,7 +11,7 @@
 
 / {
 	model = "TI OMAP3 BeagleBoard xM";
-	compatible = "ti,omap3-beagle-xm", "ti,omap3-beagle", "ti,omap3";
+	compatible = "ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3-beagle", "ti,omap3";
 
 	cpus {
 		cpu at 0 {

This compatible string looks hacky.. How about just make it

Quoted 8 lines
Quoted 3 lines

"ti,omap3-beagle-xm", "ti,omap363x", "ti,omap3";

How about just leave out "ti,omap3-beagle" here?

ti,omap3-beagle was already part of the original file, we can definitely remove but I'd expect us to maintain support for that string indefinitely should anybody continue to use older dtbs.

it already is supported (we still have a beagleboard) - the current patch rev does nothing to break existing code or bindings.

Signature · 2 lines
Regards,
Nishanth Menon