Em 05-07-2012 11:35, Antti Palosaari escreveu:
> On 07/05/2012 05:32 PM, Mauro Carvalho Chehab wrote:
>> Em 18-05-2012 15:47, Thomas Mair escreveu:
>>> +static const int reg_mask[32] = {
>>> + 0x00000001,
>>> + 0x00000003,
>>> + 0x00000007,
>>> + 0x0000000f,
>>> + 0x0000001f,
>>> + 0x0000003f,
>>> + 0x0000007f,
>>> + 0x000000ff,
>>> + 0x000001ff,
>>> + 0x000003ff,
>>> + 0x000007ff,
>>> + 0x00000fff,
>>> + 0x00001fff,
>>> + 0x00003fff,
>>> + 0x00007fff,
>>> + 0x0000ffff,
>>> + 0x0001ffff,
>>> + 0x0003ffff,
>>> + 0x0007ffff,
>>> + 0x000fffff,
>>> + 0x001fffff,
>>> + 0x003fffff,
>>> + 0x007fffff,
>>> + 0x00ffffff,
>>> + 0x01ffffff,
>>> + 0x03ffffff,
>>> + 0x07ffffff,
>>> + 0x0fffffff,
>>> + 0x1fffffff,
>>> + 0x3fffffff,
>>> + 0x7fffffff,
>>> + 0xffffffff
>>> +};
>>
>> It would be better to use a macro here like:
>>
>> #define REG_MASK(b) ((1 << ((b) + 1)) -1)
>>
>> Even better, you could use the bitops.h BIT() macro:
>>
>> #define REG_MASK(b) (BIT(b + 1) - 1)
>
> I said also that once for Thomas during review but he didn't changed it :)
As those findings are minor ones, I'll just apply the patch series
and add a patch replacing reg_mask table by a macro like above.
>>> +static int rtl2832_read_signal_strength(struct dvb_frontend *fe, u16 *strength)
>>> +{
>>> + *strength = 0;
>>> + return 0;
>>> +}
>>
>> Why to implement the above, if they're doing nothing?
>
> Other your findings were correct but for that I would like to comment.
>
> Have you ever tested what happens you lest those without stub implementation? IIRC ugly errors are seen for example zap and femon outputs. Some kind of DVB-core changes are needed. And IIRC there was some error code defined too for API - but such code does not exists.
>
I'll keep those stubs for now, but we should really fix the core and not allow/add
crap things like that.
Regards,
Mauro
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
