On 4/30/2012 9:01 PM, Lars-Peter Clausen wrote:
> On 04/30/2012 09:26 PM, joseph daniel wrote:
>> Signed-off-by: joseph daniel<[email protected]>
>> ---
>> Hi Lars,
>> Thanks for review. how about the below change?
>> drivers/staging/iio/meter/ade7854-i2c.c | 6 +++++-
>> 1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/staging/iio/meter/ade7854-i2c.c b/drivers/staging/iio/meter/ade7854-i2c.c
>> index 1e1faa0..52bdb05 100644
>> --- a/drivers/staging/iio/meter/ade7854-i2c.c
>> +++ b/drivers/staging/iio/meter/ade7854-i2c.c
>> @@ -181,6 +181,7 @@ static int ade7854_i2c_read_reg_32(struct device *dev,
>> {
>> struct iio_dev *indio_dev = dev_get_drvdata(dev);
>> struct ade7854_state *st = iio_priv(indio_dev);
>> + uint32_t value;
>> int ret;
>>
>> mutex_lock(&st->buf_lock);
>> @@ -195,7 +196,10 @@ static int ade7854_i2c_read_reg_32(struct device *dev,
>> if (ret)
>> goto out;
>>
>> - *val = (st->rx[0]<< 24) | (st->rx[1]<< 16) | (st->rx[2]<< 8) | st->rx[3];
>> + memcpy(&value, st->rx, sizeof(uint32_t));
>> +
> Uhm, yes, you are right st->rx is unaligned. The memcpy is not necessary though
> if you use get_unaligned_be32. Sorry for the pointer to the wrong function.
Or we could just force the alignment of st->rx? Might not be worth
bothering.
>
> Btw. there are a few other locations in this driver which could use the
> put_unaligned_be* and get_unaligned_be* functions. Would be great if you could
> convert these as well.
>
>> + *val = be32_to_cpu(value);
>> +
>> out:
>> mutex_unlock(&st->buf_lock);
>> return ret;
On 05/01/2012 09:56 AM, Jonathan Cameron wrote:
> On 4/30/2012 9:01 PM, Lars-Peter Clausen wrote:
>> On 04/30/2012 09:26 PM, joseph daniel wrote:
>>> Signed-off-by: joseph daniel<[email protected]>
>>> ---
>>> Hi Lars,
>>> Thanks for review. how about the below change?
>>> drivers/staging/iio/meter/ade7854-i2c.c | 6 +++++-
>>> 1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/staging/iio/meter/ade7854-i2c.c
>>> b/drivers/staging/iio/meter/ade7854-i2c.c
>>> index 1e1faa0..52bdb05 100644
>>> --- a/drivers/staging/iio/meter/ade7854-i2c.c
>>> +++ b/drivers/staging/iio/meter/ade7854-i2c.c
>>> @@ -181,6 +181,7 @@ static int ade7854_i2c_read_reg_32(struct device
>>> *dev,
>>> {
>>> struct iio_dev *indio_dev = dev_get_drvdata(dev);
>>> struct ade7854_state *st = iio_priv(indio_dev);
>>> + uint32_t value;
>>> int ret;
>>>
>>> mutex_lock(&st->buf_lock);
>>> @@ -195,7 +196,10 @@ static int ade7854_i2c_read_reg_32(struct device
>>> *dev,
>>> if (ret)
>>> goto out;
>>>
>>> - *val = (st->rx[0]<< 24) | (st->rx[1]<< 16) | (st->rx[2]<< 8)
>>> | st->rx[3];
>>> + memcpy(&value, st->rx, sizeof(uint32_t));
>>> +
>> Uhm, yes, you are right st->rx is unaligned. The memcpy is not
>> necessary though
>> if you use get_unaligned_be32. Sorry for the pointer to the wrong
>> function.
> Or we could just force the alignment of st->rx? Might not be worth
> bothering.
Yes, but be32_to_cpu wants a be32 and there is another functions which only
does a 16 bit access on rx, so we'd need to cast and things get messy. Also for
tx we store first a 16bit value and then a 32bit value and while it is possible
to get this properly aligned this gets a bit messy too. I think we are fine
with {get,set}_unaligned for now.