[PATCH 0/4] 52xx fixes for 2.6.20

[PATCH 0/4] 52xx fixes for 2.6.20

[Sylvain Munaut]

Hello Paul,

This is a set of fixes for the various 52xx platforms I'd like to
see in 2.6.20.

- First patch is just an obvious omission in the serial driver.
Certainly not critical but a simple fix nonetheless.

- The second patch corrects an incorrect usage of device_initcall

Note that it's possible other platforms make the same mistake,
a `grep "device_initcall" arch/powerpc/platforms/* -R` gives several
potential hits. I haven't investigated further though as I don't know
those platforms well enough to know if the use is legit or not.

- The two last patches make the assumption the device tree of the
EFIKA has been fixed properly for linux and allow it to boot using
more common code.


	Sylvain

[PATCH 1/4] powerpc/serial: Dispose irq mapping when done in mpc52xx_serial.c

[Sylvain Munaut]

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
---
 drivers/serial/mpc52xx_uart.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/serial/mpc52xx_uart.c b/drivers/serial/mpc52xx_uart.c
index 9d11a75..eef3b02 100644
--- a/drivers/serial/mpc52xx_uart.c
+++ b/drivers/serial/mpc52xx_uart.c
@@ -997,6 +997,9 @@ mpc52xx_uart_of_remove(struct of_device 
 	if (port)
 		uart_remove_one_port(&mpc52xx_uart_driver, port);
 
+	if (port->irq != NO_IRQ)
+		irq_dispose_mapping(port->irq);
+
 	return 0;
 }
 
-- 
1.4.2

[PATCH 2/4] powerpc: 52xx, Don't use device_initcall to probe of_platform_bus

[Sylvain Munaut]

Using device_initcall makes it happen for every platform that
compiles this file in. This is really bad, for obvious reasons.

Instead, we use the .init field of the machine description. If
the platform needs the hook to do something specific it can provides
its own function and call mpc52xx_declare_of_platform_devices from
there. If not, the mpc52xx_declare_of_platform_devices function can
directly be used as the init hook.

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
---
 arch/powerpc/platforms/52xx/lite5200.c       |    1 +
 arch/powerpc/platforms/52xx/mpc52xx_common.c |    7 ++++---
 include/asm-powerpc/mpc52xx.h                |    1 +
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/arch/powerpc/platforms/52xx/lite5200.c b/arch/powerpc/platforms/52xx/lite5200.c
index eaff71e..0f21bab 100644
--- a/arch/powerpc/platforms/52xx/lite5200.c
+++ b/arch/powerpc/platforms/52xx/lite5200.c
@@ -153,6 +153,7 @@ define_machine(lite52xx) {
 	.name 		= "lite52xx",
 	.probe 		= lite52xx_probe,
 	.setup_arch 	= lite52xx_setup_arch,
+	.init		= mpc52xx_declare_of_platform_devices,
 	.init_IRQ 	= mpc52xx_init_irq,
 	.get_irq 	= mpc52xx_get_irq,
 	.show_cpuinfo	= lite52xx_show_cpuinfo,
diff --git a/arch/powerpc/platforms/52xx/mpc52xx_common.c b/arch/powerpc/platforms/52xx/mpc52xx_common.c
index 8331ff4..cc40889 100644
--- a/arch/powerpc/platforms/52xx/mpc52xx_common.c
+++ b/arch/powerpc/platforms/52xx/mpc52xx_common.c
@@ -116,11 +116,12 @@ unmap_regs:
 	if (xlb) iounmap(xlb);
 }
 
-static int __init
+void __init
 mpc52xx_declare_of_platform_devices(void)
 {
 	/* Find every child of the SOC node and add it to of_platform */
-	return of_platform_bus_probe(NULL, NULL, NULL);
+	if (of_platform_bus_probe(NULL, NULL, NULL))
+		printk(KERN_ERR __FILE__ ": "
+			"Error while probing of_platform bus\n");
 }
 
-device_initcall(mpc52xx_declare_of_platform_devices);
diff --git a/include/asm-powerpc/mpc52xx.h b/include/asm-powerpc/mpc52xx.h
index 4a28a85..4560d72 100644
--- a/include/asm-powerpc/mpc52xx.h
+++ b/include/asm-powerpc/mpc52xx.h
@@ -244,6 +244,7 @@ #ifndef __ASSEMBLY__
 extern void __iomem * mpc52xx_find_and_map(const char *);
 extern unsigned int mpc52xx_find_ipb_freq(struct device_node *node);
 extern void mpc52xx_setup_cpu(void);
+extern void mpc52xx_declare_of_platform_devices(void);
 
 extern void mpc52xx_init_irq(void);
 extern unsigned int mpc52xx_get_irq(void);
-- 
1.4.2

[PATCH 3/4] powerpc: Use common 52xx of_platform probe code for EFIKA

[Sylvain Munaut]

Now that the device tree has the good properties, we can
remove all the efika_init code by a single call to common code.

While we're modifying that file, a few whitespaces/alignement/typo
fixes are made (nothing significant).

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
---
 arch/powerpc/platforms/52xx/efika-setup.c |   64 ++++++++---------------------
 1 files changed, 18 insertions(+), 46 deletions(-)

diff --git a/arch/powerpc/platforms/52xx/efika-setup.c b/arch/powerpc/platforms/52xx/efika-setup.c
index 110c980..d61ce84 100644
--- a/arch/powerpc/platforms/52xx/efika-setup.c
+++ b/arch/powerpc/platforms/52xx/efika-setup.c
@@ -2,7 +2,7 @@
  *
  * Efika 5K2 platform setup
  * Some code really inspired from the lite5200b platform.
- * 
+ *
  * Copyright (C) 2006 bplan GmbH
  *
  * This file is licensed under the terms of the GNU General Public License
@@ -81,35 +81,7 @@ #endif
 	efika_pcisetup();
 
 	if (ppc_md.progress)
-		ppc_md.progress("Linux/PPC " UTS_RELEASE " runnung on Efika ;-)\n", 0x0);
-}
-
-static void __init efika_init(void)
-{
-	struct device_node *np;
-	struct device_node *cnp = NULL;
-	const u32 *base;
-
-	/* Find every child of the SOC node and add it to of_platform */
-	np = of_find_node_by_name(NULL, "builtin");
-	if (np) {
-		char name[BUS_ID_SIZE];
-		while ((cnp = of_get_next_child(np, cnp))) {
-			strcpy(name, cnp->name);
-
-			base = get_property(cnp, "reg", NULL);
-			if (base == NULL)
-				continue;
-
-			snprintf(name+strlen(name), BUS_ID_SIZE, "@%x", *base);
-			of_platform_device_create(cnp, name, NULL);
-
-			printk(KERN_INFO EFIKA_PLATFORM_NAME" : Added %s (type '%s' at '%s') to the known devices\n", name, cnp->type, cnp->full_name);
-		}
-	}
-
-	if (ppc_md.progress)
-		ppc_md.progress("  Have fun with your Efika!    ", 0x7777);
+		ppc_md.progress("Linux/PPC " UTS_RELEASE " running on Efika ;-)\n", 0x0);
 }
 
 static int __init efika_probe(void)
@@ -131,20 +103,20 @@ static int __init efika_probe(void)
 
 define_machine(efika)
 {
-	.name = EFIKA_PLATFORM_NAME,
-	.probe = efika_probe,
-	.setup_arch = efika_setup_arch,
-	.init = efika_init,
-	.show_cpuinfo = efika_show_cpuinfo,
-	.init_IRQ = mpc52xx_init_irq,
-	.get_irq = mpc52xx_get_irq,
-	.restart = rtas_restart,
-	.power_off = rtas_power_off,
-	.halt = rtas_halt,
-	.set_rtc_time = rtas_set_rtc_time,
-	.get_rtc_time = rtas_get_rtc_time,
-	.progress = rtas_progress,
-	.get_boot_time = rtas_get_boot_time,
-	.calibrate_decr = generic_calibrate_decr,
-	.phys_mem_access_prot = pci_phys_mem_access_prot,
+	.name			= EFIKA_PLATFORM_NAME,
+	.probe			= efika_probe,
+	.setup_arch		= efika_setup_arch,
+	.init			= mpc52xx_declare_of_platform_devices,
+	.show_cpuinfo		= efika_show_cpuinfo,
+	.init_IRQ		= mpc52xx_init_irq,
+	.get_irq		= mpc52xx_get_irq,
+	.restart		= rtas_restart,
+	.power_off		= rtas_power_off,
+	.halt			= rtas_halt,
+	.set_rtc_time		= rtas_set_rtc_time,
+	.get_rtc_time		= rtas_get_rtc_time,
+	.progress		= rtas_progress,
+	.get_boot_time		= rtas_get_boot_time,
+	.calibrate_decr		= generic_calibrate_decr,
+	.phys_mem_access_prot	= pci_phys_mem_access_prot,
 };
-- 
1.4.2

[PATCH 4/4] powerpc: Restore 'proper' link order in platform

[Sylvain Munaut]

The 52xx was put before CHRP to allow EFIKA to be recognized
properly. Now the efika tree is fixed up in prom_init so
no need for this ugly hack. So we restore the 'normal'
order.

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
---
 arch/powerpc/platforms/Makefile |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/powerpc/platforms/Makefile b/arch/powerpc/platforms/Makefile
index 507d1b9..44d95ea 100644
--- a/arch/powerpc/platforms/Makefile
+++ b/arch/powerpc/platforms/Makefile
@@ -5,9 +5,9 @@ ifeq ($(CONFIG_PPC64),y)
 obj-$(CONFIG_PPC_PMAC)		+= powermac/
 endif
 endif
-obj-$(CONFIG_PPC_MPC52xx)	+= 52xx/
 obj-$(CONFIG_PPC_CHRP)		+= chrp/
 obj-$(CONFIG_4xx)		+= 4xx/
+obj-$(CONFIG_PPC_MPC52xx)	+= 52xx/
 obj-$(CONFIG_PPC_83xx)		+= 83xx/
 obj-$(CONFIG_PPC_85xx)		+= 85xx/
 obj-$(CONFIG_PPC_86xx)		+= 86xx/
-- 
1.4.2

Re: [PATCH 4/4] powerpc: Restore 'proper' link order in platform

[Grant Likely]

On 1/2/07, Sylvain Munaut <tnt@246tnt.com> wrote:
> The 52xx was put before CHRP to allow EFIKA to be recognized
> properly. Now the efika tree is fixed up in prom_init so
> no need for this ugly hack. So we restore the 'normal'
> order.
>
> Signed-off-by: Sylvain Munaut <tnt@246tNt.com>

Acked-by: Grant Likely <grant.likely@secretlab.ca>

Re: [PATCH 3/4] powerpc: Use common 52xx of_platform probe code for EFIKA

[Grant Likely]

On 1/2/07, Sylvain Munaut <tnt@246tnt.com> wrote:
> Now that the device tree has the good properties, we can
> remove all the efika_init code by a single call to common code.
>
> While we're modifying that file, a few whitespaces/alignement/typo
> fixes are made (nothing significant).
>
> Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>

Re: [PATCH 3/4] powerpc: Use common 52xx of_platform probe code for EFIKA

[Paul Mackerras]

Sylvain Munaut writes:

> Now that the device tree has the good properties, we can
> remove all the efika_init code by a single call to common code.

Ummm, but the device tree doesn't have good properties, does it,
without your patch that adds a whole pile of device tree fixups?
I'm reluctant to put that in for 2.6.20 because it's moderately large,
has at least the potential to upset other platforms, and what it fixes
isn't a regression from 2.6.19, and we're already past -rc4.

Your patch 4/4 seems to depend on this one (3/4) so I didn't put it in
for 2.6.20 either.

Paul.

Re: [PATCH 3/4] powerpc: Use common 52xx of_platform probe code for EFIKA

[Sylvain Munaut]

About the serial patch :

> I didn't put this one in for 2.6.20, because it doesn't solve any
> actual problem, and Ben H tells me it might trigger a bug in the
> generic irq code.

The bug it triggered has been fixed eons ago (like last December 22),
in a impressive one-liner patch that has 5 Signed-off/Acked-by ;) ...

But that code is rarely executed as it would imply the PSC can be
unloaded ... (impossible when used as console). So later merge
is OK.



Paul Mackerras wrote:
> Sylvain Munaut writes:
>
>   
>> Now that the device tree has the good properties, we can
>> remove all the efika_init code by a single call to common code.
>>     
>
> Ummm, but the device tree doesn't have good properties, does it,
> without your patch that adds a whole pile of device tree fixups?
> I'm reluctant to put that in for 2.6.20 because it's moderately large,
> has at least the potential to upset other platforms, and what it fixes
> isn't a regression from 2.6.19, and we're already past -rc4.
>   
Ok, let's just delay everything to the next merge window. Anyway most
the drivers won't get merged until then.

I'll redo a nice patchset with all I got when it opens (or if you open a
for-2.6.21 branch ;).


    Sylvain

Re: [PATCH 3/4] powerpc: Use common 52xx of_platform probe code for EFIKA

[Benjamin Herrenschmidt]

On Tue, 2007-01-09 at 23:02 +0100, Sylvain Munaut wrote:
> About the serial patch :
> 
> > I didn't put this one in for 2.6.20, because it doesn't solve any
> > actual problem, and Ben H tells me it might trigger a bug in the
> > generic irq code.
> 
> The bug it triggered has been fixed eons ago (like last December 22),
> in a impressive one-liner patch that has 5 Signed-off/Acked-by ;) ...
> 
> But that code is rarely executed as it would imply the PSC can be
> unloaded ... (impossible when used as console). So later merge
> is OK.

And there is no real problem in this case for not disposing of the
mapping anyway. You'll just get the same mapping again when you reload
it.

Ben.

Re: [PATCH 2/4] powerpc: 52xx, Don't use device_initcall to probe of_platform_bus

[Grant Likely]

On 1/2/07, Sylvain Munaut <tnt@246tnt.com> wrote:
> Using device_initcall makes it happen for every platform that
> compiles this file in. This is really bad, for obvious reasons.
>
> Instead, we use the .init field of the machine description. If
> the platform needs the hook to do something specific it can provides
> its own function and call mpc52xx_declare_of_platform_devices from
> there. If not, the mpc52xx_declare_of_platform_devices function can
> directly be used as the init hook.
>
> Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>

Re: [PATCH 1/4] powerpc/serial: Dispose irq mapping when done in mpc52xx_serial.c

[Grant Likely]

On 1/2/07, Sylvain Munaut <tnt@246tnt.com> wrote:
> Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>

Re: [PATCH 1/4] powerpc/serial: Dispose irq mapping when done in mpc52xx_serial.c

[Paul Mackerras]

Sylvain Munaut writes:

> Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
> ---
>  drivers/serial/mpc52xx_uart.c |    3 +++
>  1 files changed, 3 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/serial/mpc52xx_uart.c b/drivers/serial/mpc52xx_uart.c
> index 9d11a75..eef3b02 100644
> --- a/drivers/serial/mpc52xx_uart.c
> +++ b/drivers/serial/mpc52xx_uart.c
> @@ -997,6 +997,9 @@ mpc52xx_uart_of_remove(struct of_device 
>  	if (port)
>  		uart_remove_one_port(&mpc52xx_uart_driver, port);
>  
> +	if (port->irq != NO_IRQ)
> +		irq_dispose_mapping(port->irq);
> +

I didn't put this one in for 2.6.20, because it doesn't solve any
actual problem, and Ben H tells me it might trigger a bug in the
generic irq code.

Paul.