Re: [PATCH] dma: Revert "dmaengine: mic_x100: add missing spin_unlock"

[Ashutosh Dixit <ashutosh.dixit@intel.com>]

On Wed, Dec 23 2015 at 12:45:31 AM, Saurabh Sengar <saurabh.truth@gmail.com> wrote:
> On 23 December 2015 at 09:05, Ashutosh Dixit <ashutosh.dixit@intel.com> wrote:
>> This reverts commit e958e079e254 ("dmaengine: mic_x100: add missing
>> spin_unlock").
>>
>> The above patch is incorrect. There is nothing wrong with the original
>> code. The spin_lock is acquired in the "prep" functions and released
>> in "submit".
>
> Hi Ashutosh,
>
> If it is need to be released by submit function, we don't require the
> spin_unlock on success case as well.
> am I correct ?

No, you are wrong. In the prep functions, we are not using spin_unlock in
success cases but in failure ones.

>> Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
>> ---
>>  drivers/dma/mic_x100_dma.c | 15 +++++----------
>>  1 file changed, 5 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/dma/mic_x100_dma.c b/drivers/dma/mic_x100_dma.c
>> index cddfa8d..068e920 100644
>> --- a/drivers/dma/mic_x100_dma.c
>> +++ b/drivers/dma/mic_x100_dma.c
>> @@ -317,7 +317,6 @@ mic_dma_prep_memcpy_lock(struct dma_chan *ch, dma_addr_t dma_dest,
>>         struct mic_dma_chan *mic_ch = to_mic_dma_chan(ch);
>>         struct device *dev = mic_dma_ch_to_device(mic_ch);
>>         int result;
>> -       struct dma_async_tx_descriptor *tx = NULL;
>>
>>         if (!len && !flags)
>>                 return NULL;
>> @@ -325,13 +324,10 @@ mic_dma_prep_memcpy_lock(struct dma_chan *ch, dma_addr_t dma_dest,
>>         spin_lock(&mic_ch->prep_lock);
>>         result = mic_dma_do_dma(mic_ch, flags, dma_src, dma_dest, len);
>>         if (result >= 0)
>> -               tx = allocate_tx(mic_ch);
>> -
>> -       if (!tx)
>> -               dev_err(dev, "Error enqueueing dma, error=%d\n", result);
>> -
>> +               return allocate_tx(mic_ch);
>> +       dev_err(dev, "Error enqueueing dma, error=%d\n", result);
>>         spin_unlock(&mic_ch->prep_lock);
>
> This spin_unlock shouldn't be required as explained it is getting
> released by submit function

This is a failure case. Submit will not be called in this case so
spin_unlock needs to be called here.

>> -       return tx;
>> +       return NULL;
>>  }
>>
>>  static struct dma_async_tx_descriptor *
>> @@ -339,14 +335,13 @@ mic_dma_prep_interrupt_lock(struct dma_chan *ch, unsigned long flags)
>>  {
>>         struct mic_dma_chan *mic_ch = to_mic_dma_chan(ch);
>>         int ret;
>> -       struct dma_async_tx_descriptor *tx = NULL;
>>
>>         spin_lock(&mic_ch->prep_lock);
>>         ret = mic_dma_do_dma(mic_ch, flags, 0, 0, 0);
>>         if (!ret)
>> -               tx = allocate_tx(mic_ch);
>> +               return allocate_tx(mic_ch);
>>         spin_unlock(&mic_ch->prep_lock);
>
> and this too ?

Yes, this too.

>> -       return tx;
>> +       return NULL;
>>  }
>>
>>  /* Return the status of the transaction */
>> --
>> 2.0.0.rc3.2.g998f840
>>