Re: [RFC PATCH 2/9] memory: tegra: add mc flush support

[Vince Hsu <vinceh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>]

On 03/03/2015 04:14 PM, Alexandre Courbot wrote:
> On Tue, Mar 3, 2015 at 5:09 PM, Vince Hsu <vinceh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org> wrote:
>> On 03/03/2015 04:03 PM, Alexandre Courbot wrote:
>>> On Mon, Mar 2, 2015 at 6:29 PM, Alexandre Courbot <gnurou-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>>> wrote:
>>>>>> These functions are identical, excepted for the callback they are
>>>>>> invoking. Could you merge the common part into a function that returns
>>>>>> the right client to call the callback on, or ERR_PTR(-EINVAL) in case
>>>>>> of failure?
>>>>> I couldn't think of a clever way to do this. Any ideas? :)
>>>> How about something like this (warning: might now be that great,
>>>> untested):
>>>>
>>>> /* Have this in your .h and use it in your tegra_mc_ops struct */
>>>> typedef int (*mc_op)(struct tegra_mc *mc,
>>>>                const struct tegra_mc_hotreset *hotreset)
>>> This type should be named tegra_mc_op, since the header that defines
>>> it is in include/linux.
>> Can we just leave it in this C file? I see no reason to place it in some
>> other
>> header file. :)
> Can you also move tegra_mc_ops's definition into the C file? If so I
> agree. Otherwise I prefer to make sure that the same type is used
> everywhere.
No. Let's keep that in the include/soc/tegra/mc.h.

Thanks,
Vince