Re: [PATCH 2/3 v2] ARM: bcm2835: Add the Raspberry Pi firmware driver

["Noralf Trønnes" <noralf@tronnes.org>]

Den 28.05.2015 23:17, skrev Stephen Warren:
> On 05/13/2015 01:00 PM, Eric Anholt wrote:
>> This gives us a function for making mailbox property channel requests
>> of the firmware, which is most notable in that it will let us get and
>> set clock rates.
> ...
>> Note that I don't think I've done what srwarren wanted for
>> -EPROBE_DEFER, because I'm not clear what he wants.  I think he might
>> just be asking for a function that does:
>>
>> /*
>>   * Returns 0 if the firmware device is probed and available, otherwise
>>   * -EPROBE_DEFER.
>>   */
>> int rpi_firmware_get(struct device_node *firmware_node)
>> {
>> 	struct platform_device *pdev = of_find_device_by_node(of_node);
>> 	if (!platform_get_drvdata(pdev))

of_find_device_by_node() can return NULL if the device can't be found.
platform_get_drvdata() can't handle a NULL pointer.

>> 		return -EPROBE_DEFER;
>> 	return 0;
>> }
>> EXPORT_SYMBOL(rpi_firmware_get)
>>
>> If that's all, I'm happy to add it.
> Yes, there definitely needs to be something that clients can call at
> probe() time to make sure the firmware driver is there already. That
> check is quite different from actually sending a request to the FW, so
> I'd certainly expect a separate function for that.

A try_module_get() here would make sure this module won't go away while a
client relies on it. For that we would need a rpi_firmware_put() as well.