Re: [PATCH V3 06/10] MIPS: lantiq: add NOR flash support

[Sergei Shtylyov <sshtylyov@mvista.com>]

I wrote:

>> NOR flash is attached to the same EBU (External Bus Unit) as PCI. As 
>> described
>> in the PCI patch, the EBU is a little buggy, resulting in the upper 
>> and lower
>> 16 bit of the data on a 32 bit read are swapped. (essentially we have 
>> a addr^=2)

>    "Are" not needed.

>> To work around this we do a addr^=2 during the probe. Once probed we 
>> adapt
>> cfi->addr_unlock1 and cfi->addr_unlock2 to represent the endianess bug.

>> Signed-off-by: John Crispin<blogic@openwrt.org>
>> Signed-off-by: Ralph Hempel<ralph.hempel@lantiq.com>
>> Cc: David Woodhouse<dwmw2@infradead.org>
>> Cc: Daniel Schwierzeck<daniel.schwierzeck@googlemail.com>
>> Cc: linux-mips@linux-mips.org
>> Cc: linux-mtd@lists.infradead.org
> [...]

>> diff --git a/drivers/mtd/maps/lantiq.c b/drivers/mtd/maps/lantiq.c
>> new file mode 100644
>> index 0000000..674be0a
>> --- /dev/null
>> +++ b/drivers/mtd/maps/lantiq.c
>> @@ -0,0 +1,190 @@
[...]
>> +void
>> +ltq_copy_from(struct map_info *map, void *to,
>  > +    unsigned long from, ssize_t len)

>    Shouldn't it be static?

>> +{
>> +    unsigned char *p;
>> +    unsigned char *to_8;
>> +    unsigned long flags;
>> +
>> +    spin_lock_irqsave(&ebu_lock, flags);
>> +    from = (unsigned long) (map->virt + from);

>    Why not:

>     from += (unsigned long) map->virt;

> like you do in ltq_copy_to()?

>> +    p = (unsigned char *) from;

>    Could be done in initializer, like in ltq_copy_to().

>> +    to_8 = (unsigned char *) to;
>> +    while (len--)
>> +        *to_8++ = *p++;

>    BTW, you could use memcpy_fromio().

    Actually not, as on MIPS it's implemented via memcpy(). On ARM it doesn 
byte-by-byte copying -- that's why I remembered about it...

WBR, Sergei