Re: [PATCH] spi: spi-rspi: add dmaengine supporting

["Shimoda, Yoshihiro" <yoshihiro.shimoda.uh-zM6kxYcvzFBBDgjK7y7TUQ@public.gmane.org>]

2012/04/20 10:36, Paul Mundt wrote:
> On Thu, Apr 19, 2012 at 07:22:53PM +0900, Shimoda, Yoshihiro wrote:
>> +static struct dma_async_tx_descriptor *rspi_dma_prep_sg(struct scatterlist *sg,
>> +				void *buf, unsigned len, struct dma_chan *chan,
>> +				enum dma_transfer_direction dir)
>> +{
>> +	sg_init_table(sg, 1);
>> +	sg_set_buf(sg, buf, len);
>> +	sg_dma_len(sg) = len,
>> +	dma_map_sg(chan->device->dev, sg, 1, dir);
>> +	return dmaengine_prep_slave_sg(chan, sg, 1, dir,
>> +				       DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
>> +}
>> +
> ..
> 
>> +	desc = rspi_dma_prep_sg(&sg, buf, len, rspi->chan_tx, DMA_TO_DEVICE);
>> +	if (!desc) {
>> +		ret = -EIO;
>> +		goto error;
>> +	}
>> +
> ..
>> +error:
>> +	if (rspi->dma_width_16bit)
>> +		kfree(buf);
>> +
>> +	return ret;
>> +}
>> +
> The sg list handling is unbalanced. Specifically you always map the
> scatterlist with dma_map_sg() but you have no corresponding
> dma_unmap_sg() anywhere, either in the error path or your regular exit
> path.
> 

Thank you for the point. I will fix this.

Best regards,
Yoshihiro Shimoda

------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2