Re: [PATCH try#2] [MTD/MAPS] Blackfin BF5xx Maps: Handle the case where flash memory and ethernet mac/phy are mapped onto the same async bank

["Mike Frysinger" <vapier.adi@gmail.com>]

On Mon, May 12, 2008 at 8:26 AM, Jörn Engel <joern@logfs.org> wrote:
> On Sun, 3 February 2008 14:32:35 +0800, Bryan Wu wrote:
>> +static void bf5xx_copy_from(struct map_info *map, void *to,
unsigned long from, ssize_t len)
>> +{
>> +     unsigned long i;
>> +     map_word test;
>> +
>> +     if ((unsigned long)to & 0x1) {
>> +             for (i = 0; i < len / 2 * 2; i += 2) {
>> +                     test = bf5xx_read(map, from + i);
>> +                     put_unaligned(test.x[0], (__le16 *)(to + i));
>> +             }
>> +     } else {
>> +             for (i = 0; i < len / 2 * 2; i += 2) {
>> +                     test = bf5xx_read(map, from + i);
>> +                     *((u16*)(to + i)) = test.x[0];
>                        ^^^^^^^^^^^^^^^^^
>
> I'm not particularly fond of such constructs, maybe because I have such
> a hard time understanding what the code actually does.
>
> How about:
>        u16 *foo = to;
>
>        for (i = 0; i < (len & ~1); i++) {
>                test = bf5xx_read(map, from + 2 * i);
>                foo[i] = test.x[0];
>        }
>
> With perhaps a better name than "foo" and the same conversion in the
> rest of the function as well?

indeed.  Bryan: i can take care of this as there's a few other pieces
i want to fiddle first (like getting rid of "bf5xx")
-mike