Re: [PATCH v11 4/5] powerpc: Add flexcan device support for p1010rdb.

[Wolfgang Grandegger <wg@grandegger.com>]

On 08/11/2011 05:56 AM, Robin Holt wrote:
> On Wed, Aug 10, 2011 at 08:16:33PM +0200, Wolfgang Grandegger wrote:
>> On 08/10/2011 07:01 PM, Kumar Gala wrote:
>>>
>>> On Aug 10, 2011, at 11:27 AM, Robin Holt wrote:
>>>
>>>> I added a simple clock source for the p1010rdb so the flexcan driver
>>>> could determine a clock frequency.  The p1010 flexcan device only has
>>>> an oscillator of system bus frequency divided by 2.
>>>>
>>>> Signed-off-by: Robin Holt <holt@sgi.com>
>>>> Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>,
>>>> Acked-by: Wolfgang Grandegger <wg@grandegger.com>,
>>>> Cc: U Bhaskar-B22300 <B22300@freescale.com>
>>>> Cc: socketcan-core@lists.berlios.de,
>>>> Cc: netdev@vger.kernel.org,
>>>> Cc: PPC list <linuxppc-dev@lists.ozlabs.org>
>>>> Cc: Kumar Gala <galak@kernel.crashing.org>
>>>> ---
>>>> arch/powerpc/platforms/85xx/Kconfig    |    2 +
>>>> arch/powerpc/platforms/85xx/Makefile   |    2 +
>>>> arch/powerpc/platforms/85xx/clock.c    |   52 ++++++++++++++++++++++++++++++++
>>>> arch/powerpc/platforms/85xx/p1010rdb.c |    8 +++++
>>>> 4 files changed, 64 insertions(+), 0 deletions(-)
>>>> create mode 100644 arch/powerpc/platforms/85xx/clock.c
>>>
>>> I dont understand how mpc85xx_clk_functions() ends up being associated with the frequency the flexcan is running at.
>>
>> The function mpc85xx_clk_get_rate() returns "fsl_get_sys_freq() / 2" for
>> Flexcan devices.
>>
>>> This either seems to global or I'm missing something.
>>
>> This patch extends the existing Flexcan platform driver for ARM for the
>> PowerPC using the device tree. Due to the nice integration of the device
>> tree (of-platform) into the platform driver and devices, the difference
>> are quite small (see patches 1..3). Apart from the endianess issue, only
>> the clock needs to be handled in a common way. As ARM already uses the
>> clk interface, we found it straight-forward to implement it for the
>> P1010, or more general for the 85xx, as well, instead of using an
>> additional helper function.
>>
>>> I still think the clk / freq info should be in the device tree and handled in the driver and NOT arch/powerpc platform code.
>>
>> If I understand you correctly, you want the boot-loader to provide the
>> relevant information by fixing up the device tree, which then can be
>> handled arch-independently by the driver, right?
> 
> Marc and Wolfgang,
> 
> This is a very early swag at this which I worked up while driving home from dinner
> this evening.  It works with my current config, but that includes at least one
> bogus patch.  I will have to do more testing tomorrow.  For now, it is something
> to ponder.

Yes, that's what Kumar is proposing. Fine for me with the P1010. It just
needs some proper U-Boot implementation.

> Thanks,
> Robin
> ------------------------------------------------------------------------
> 
> flexcan: Prefer device tree clock frequency if available.
> 
> If our CAN device's device tree node has a clock-frequency property,
> then use that value for the can devices clock frequency.  If not, fall
> back to asking the platform/mach code for the clock frequency associated
> with the flexcan device.
> 
> Too-early-to-be-signed-off-by: Robin Holt <holt@sgi.com>
> 
> diff --git a/drivers/net/can/flexcan.c b/drivers/net/can/flexcan.c
> index 662f832..d6a87c9 100644
> --- a/drivers/net/can/flexcan.c
> +++ b/drivers/net/can/flexcan.c
> @@ -33,6 +33,7 @@
>  #include <linux/kernel.h>
>  #include <linux/list.h>
>  #include <linux/module.h>
> +#include <linux/of.h>
>  #include <linux/platform_device.h>
>  
>  #define DRV_NAME			"flexcan"
> @@ -929,12 +930,25 @@ static int __devinit flexcan_probe(struct platform_device *pdev)
>  	void __iomem *base;
>  	resource_size_t mem_size;
>  	int err, irq;
> +	u32 clock_freq = 0;
>  
> -	clk = clk_get(&pdev->dev, NULL);
> -	if (IS_ERR(clk)) {
> -		dev_err(&pdev->dev, "no clock defined\n");
> -		err = PTR_ERR(clk);
> -		goto failed_clock;
> +	if (pdev->dev.of_node) {
> +		u32 *clock_freq_p;

Should be:

		const u32 *clock_freq_p;

> +
> +		clk = NULL;
> +		clock_freq_p = (u32 *)of_get_property(pdev->dev.of_node, "clock-frequency", NULL);

No cast please.

> +		if (clock_freq_p)
> +			clock_freq = *clock_freq_p;
> +	}

Wolfgang.