[PATCH v5 2/7] mailbox: arm_mhu: add driver for ARM MHU controller

[Jassi Brar]

On 5 February 2015 at 17:38, Russell King - ARM Linux
<linux at arm.linux.org.uk> wrote:
> On Thu, Feb 05, 2015 at 05:32:39PM +0530, Jassi Brar wrote:
>> On Feb 5, 2015 5:13 PM, "Arnd Bergmann" <arnd at arndb.de> wrote:
>> >
>> > >
>> > > I know typedef's are frowned upon, but how bad is the following option?
>> > >        typedef void*  mbox_data_info
>> > >        int mbox_send_message(struct mbox_chan *chan, mbox_data_info
>> data);
>> >
>> > I don't see how that would help.
>> >
>> If it's abuse because the argument is a void*... What if we called it
>> mbox_data_info?  To say platforms are free to pass data as a pointer or a
>> value :)
>
> Using a typedef really doesn't change anything.  If you read the kernel
> coding style, you'll realise that typedefs are a reason to reject
> patches - especially to use them in the way you are proposing.
>
I started with "I know typedef's are frowned upon". But thanks for reminding me.

Instead of a void*, we could say a client passes data to the
controller via an argument that is sizeof(void *) big, so we
facilitate both usages. Anyways, I get the message that that is not
acceptable.

> Try the solution I suggested (which I notice was seemingly totally
> ignored.)
>
My client drivers work in blocking mode, so I don't suffer. For the
platform that passes u32 in async mode, we'll let it manage its own
circular buffer scheme as you suggested.

Thanks.