[PATCH] OMAP1: clock_data: use runtime cpu / machine checks

[PATCH] OMAP1: clock_data: use runtime cpu / machine checks

[Janusz Krzysztofik]

Otherwise multi-omap1 configurations may set wrong clock speed.

Created and tested against l-o master on Amstrad Delta.

Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
---
Tested with omap1_defconfig after dropping 
CONFIG_OMAP_CLOCKS_SET_BY_BOOTLOADER=y and those of 12MHz xtal based clock 
speeds which are too high for my board.

 arch/arm/mach-omap1/clock_data.c |   17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

--- git/arch/arm/mach-omap1/clock_data.c.orig	2010-12-09 23:07:35.000000000 +0100
+++ git/arch/arm/mach-omap1/clock_data.c	2010-12-11 21:05:52.000000000 +0100
@@ -823,12 +823,10 @@ int __init omap1_clk_init(void)
 			crystal_type = info->system_clock_type;
 	}
 
-#if defined(CONFIG_ARCH_OMAP730) || defined(CONFIG_ARCH_OMAP850)
-	ck_ref.rate = 13000000;
-#elif defined(CONFIG_ARCH_OMAP16XX)
-	if (crystal_type == 2)
+	if (cpu_is_omap7xx())
+		ck_ref.rate = 13000000;
+	if (cpu_is_omap16xx() && crystal_type == 2)
 		ck_ref.rate = 19200000;
-#endif
 
 	pr_info("Clocks: ARM_SYSST: 0x%04x DPLL_CTL: 0x%04x ARM_CKCTL: "
 		"0x%04x\n", omap_readw(ARM_SYSST), omap_readw(DPLL_CTL),
@@ -883,10 +881,11 @@ int __init omap1_clk_init(void)
 	       ck_dpll1.rate / 1000000, (ck_dpll1.rate / 100000) % 10,
 	       arm_ck.rate / 1000000, (arm_ck.rate / 100000) % 10);
 
-#if defined(CONFIG_MACH_OMAP_PERSEUS2) || defined(CONFIG_MACH_OMAP_FSAMPLE)
-	/* Select slicer output as OMAP input clock */
-	omap_writew(omap_readw(OMAP7XX_PCC_UPLD_CTRL) & ~0x1, OMAP7XX_PCC_UPLD_CTRL);
-#endif
+	if (machine_is_omap_perseus2() || machine_is_omap_fsample()) {
+		/* Select slicer output as OMAP input clock */
+		omap_writew(omap_readw(OMAP7XX_PCC_UPLD_CTRL) & ~0x1,
+				OMAP7XX_PCC_UPLD_CTRL);
+	}
 
 	/* Amstrad Delta wants BCLK high when inactive */
 	if (machine_is_ams_delta())

[PATCH] OMAP1: clock_data: use runtime cpu / machine checks

[Paul Walmsley]

Hello Janusz, Cory,

On Sat, 11 Dec 2010, Janusz Krzysztofik wrote:

> Otherwise multi-omap1 configurations may set wrong clock speed.
> 
> Created and tested against l-o master on Amstrad Delta.
> 
> Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>

Thanks, this is a significant step forward in building a good 
omap1_defconfig.  Cory, do you mind testing this out?  If it works for you 
also, let's plan to take this for 2.6.38.

However:

> Tested with omap1_defconfig after dropping 
> CONFIG_OMAP_CLOCKS_SET_BY_BOOTLOADER=y and those of 12MHz xtal based 
> clock speeds which are too high for my board.

This is also key.  Until the board files can somehow indicate which rate 
sets are usable for their boards at runtime -- which is what the 
CONFIG_OMAP_ARM* KConfig options used to do at KConfig-time -- I don't 
think omap1_defconfig is possible yet.

Volunteers to tackle that project are gladly welcome :-)


- Paul

[PATCH] OMAP1: clock_data: use runtime cpu / machine checks

[Cory Maccarrone]

On Mon, Dec 13, 2010 at 1:24 AM, Paul Walmsley <paul@pwsan.com> wrote:
>
> Hello Janusz, Cory,
>
> On Sat, 11 Dec 2010, Janusz Krzysztofik wrote:
>
> > Otherwise multi-omap1 configurations may set wrong clock speed.
> >
> > Created and tested against l-o master on Amstrad Delta.
> >
> > Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
>
> Thanks, this is a significant step forward in building a good
> omap1_defconfig. ?Cory, do you mind testing this out? ?If it works for you
> also, let's plan to take this for 2.6.38.
>

Sure, no problem, I'll add it to my testing queue.

- Cory

[PATCH] OMAP1: clock_data: use runtime cpu / machine checks

[Cory Maccarrone]

On Mon, Dec 13, 2010 at 1:24 AM, Paul Walmsley <paul@pwsan.com> wrote:
> Hello Janusz, Cory,
>
> On Sat, 11 Dec 2010, Janusz Krzysztofik wrote:
>
>> Otherwise multi-omap1 configurations may set wrong clock speed.
>>
>> Created and tested against l-o master on Amstrad Delta.
>>
>> Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
>
> Thanks, this is a significant step forward in building a good
> omap1_defconfig. ?Cory, do you mind testing this out? ?If it works for you
> also, let's plan to take this for 2.6.38.
>
> However:
>
>> Tested with omap1_defconfig after dropping
>> CONFIG_OMAP_CLOCKS_SET_BY_BOOTLOADER=y and those of 12MHz xtal based
>> clock speeds which are too high for my board.
>
> This is also key. ?Until the board files can somehow indicate which rate
> sets are usable for their boards at runtime -- which is what the
> CONFIG_OMAP_ARM* KConfig options used to do at KConfig-time -- I don't
> think omap1_defconfig is possible yet.
>
> Volunteers to tackle that project are gladly welcome :-)
>
>
> - Paul
>

This doesn't seem to work for me.  Using my own config, it'll boot
fine, but the omap1_defconfig, even after making the above changes,
still fails to boot.  I'll see if I can figure out what makes it stop.

- Cory