Re: [patch 2/2] [media] ds3000: off by one in ds3000_read_snr()

[walter harms <wharms@bfs.de>]

Am 19.01.2012 10:33, schrieb Dan Carpenter:
> On Wed, Jan 18, 2012 at 06:06:46PM +0100, walter harms wrote:
>>
>>
>> Am 17.01.2012 08:30, schrieb Dan Carpenter:
>>> This is a static checker patch and I don't have the hardware to test
>>> this, so please review it carefully.  The dvbs2_snr_tab[] array has 80
>>> elements so when we cap it at 80, that's off by one.  I would have
>>> assumed that the test was wrong but in the lines right before we have
>>> the same test but use "snr_reading - 1" as the array offset.  I've done
>>> the same thing here.
>>>
>>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>
>>> diff --git a/drivers/media/dvb/frontends/ds3000.c b/drivers/media/dvb/frontends/ds3000.c
>>> index af65d01..3f5ae0a 100644
>>> --- a/drivers/media/dvb/frontends/ds3000.c
>>> +++ b/drivers/media/dvb/frontends/ds3000.c
>>> @@ -681,7 +681,7 @@ static int ds3000_read_snr(struct dvb_frontend *fe, u16 *snr)
>>>  			snr_reading = dvbs2_noise_reading / tmp;
>>>  			if (snr_reading > 80)
>>>  				snr_reading = 80;
>>> -			*snr = -(dvbs2_snr_tab[snr_reading] / 1000);
>>> +			*snr = -(dvbs2_snr_tab[snr_reading - 1] / 1000);
>>>  		}
>>>  		dprintk("%s: raw / cooked = 0x%02x / 0x%04x\n", __func__,
>>>  				snr_reading, *snr);
>>
>> hi dan,
>>
>> perhaps it is more useful to do it in the check above ?
> 
> It looks like the check is correct but we need to shift all the
> values by one.  Again, I don't have this hardware, I'm just going by
> the context.
> 
I do not have the hardware either so this is pure theoretical.

Access to the data field depends on the value of dvbs2_noise_reading/tmp
even when the data are reasonable like 50/100 snr_reading would become 0
and the index suddenly is -1.

just my 2 cents.

re,
 wh


>> thinking about that why not replace the number (80) with ARRAY_SIZE() ?
> 
> That would be a cleanup, yes but it could go in a separate patch.
> 
> regards,
> dan carpenter
>