On 6 August 2015 at 13:00, Lars-Peter Clausen <lars@xxxxxxxxxx> wrote:
> On 08/06/2015 12:42 PM, Peter Meerwald wrote:
>> Hello,
>>
>>> I am sure at least some of you were faced to split a u16 or u32 value
>>> in u8 array without being affected by host endianness. My search
>>> turned empty for a standardized function for this task, but this seems
>>> quite improbable. My task is to send 16 bit LE data through i2c.
>>>
>>> I am looking for something like:
>>> static void u16_to_u8_array(u16 value, u8 *array)
>>> {
>>> *array = cpu_to_le16(value) >> 8;
>>> *(++array) = (u8) cpu_to_le16(value);
>>> }
>>
>> how about
>>
>> static void u16_to_le16_array(u16 value, __le16 *array)
>> {
>> *array++ = cpu_to_le16(value);
>> }
>>
>> and calling it with a cast if need be, such as
>>
>> char *u8array;
>> u16_to_le16_array(123, (__le16 *) u8array)
>>
>> an issue could be alignment of the u8array
>
> There is put_unaligned_le16() and friends for this situation where the
> target buffer is just a generic bytestream. This is architecture optimized,
> so if the architecture supports unaligned access it will just do a store, if
> it doesn't it will split the operation into multiple stores.
>
>>
>> I'd try to avoid u8 altogether and work with le16 as a datatype
>
> That seems to be the best solution in this case.
So i2c_master_send takes a char buf as input, and if I would send a le16
array[2] then the cast would nicely fit values it into that.
Thanks for the put_unaligned_le16() hint - exactly what I wanted, but
will investigate the best solution in this case as it seems to require a lot
less lines.
--
To unsubscribe from this list: send the line "unsubscribe linux-iio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
