Re: [Qemu-devel] [PATCH v2 2/4] Add i.MX I2C controller driver.

[Jean-Christophe DUBOIS <jcd@tribudubois.net>]

On 05/05/2013 01:53 PM, Peter Crosthwaite wrote:
> Hi Peter,
>
> On Sun, May 5, 2013 at 9:41 PM, Peter Maydell <peter.maydell@linaro.org> wrote:
>> On 5 May 2013 11:53, Peter Crosthwaite <peter.crosthwaite@xilinx.com> wrote:
>>> On Sun, May 5, 2013 at 8:47 PM, Peter Maydell <peter.maydell@linaro.org> wrote:
>>>> On 5 May 2013 04:58, Jean-Christophe DUBOIS <jcd@tribudubois.net> wrote:
>>>>> no extract16() macro spotted.
>>>>> Should one be added?
>>>> There's no need for one -- just use extract32. The only
>>>> reason for having a separate extract64 is to avoid doing
>>>> 64 bit arithmetic when we don't have to, I think.
>>>>
>>> perhaps a quick:
>>>
>>> #define extract16(a, b, c) (uint16_t)extract32((uint16_t)(a), (b), (c))
>>>
>>> would keep the 16bit device code explicit and clean? Save on
>>> dummy casts in certain situations as well (but not this one).
>> Hmm, what situations would ever require a cast of the result
>> or the input of extract32?
>>
> The one off the top of my head:
>
> fprintf(stderr, "your 16 field is :%" PRIx16 "\n", extract16(foo, bar, baz));
>
> Otherwise you would have to match PRIx32 to extract 16 which is clumsy.
>
> But aren't there there some other varargs situations that may require
> casts as well?
>
> As for the input cast, I got nothin, I just put it in there for
> completeness. Meet you half way and drop the input cast and keep the
> output?
>
> Regards,
> Peter

I will use extract32() for now.

JC

>
>> -- PMM
>>