Re: [PATCH V2] dma: add common of_dma_slave_xlate()

[Lars-Peter Clausen <lars@metafoo.de>]

On 11/26/2013 02:25 AM, Stephen Warren wrote:
[...]
> +struct dma_chan *of_dma_slave_xlate(struct of_phandle_args *dma_spec,
> +				    struct of_dma *ofdma)
> +{
> +	struct of_dma_slave_xlate_info *info = ofdma->of_dma_data;
> +	struct dma_chan *candidate, *chan;
> +	int ret;
> +
> +retry:
> +	candidate = NULL;
> +
> +	/*
> +	 * walk the list of channels registered with the current instance and
> +	 * find one that is currently unused
> +	 */
> +	list_for_each_entry(chan, &info->device->channels, device_node)
> +		if (chan->client_count == 0) {
> +			candidate = chan;
> +			break;
> +		}
> +
> +	if (!candidate)
> +		return NULL;
> +
> +	/*
> +	 * dma_get_slave_channel will return NULL if we lost a race between
> +	 * the lookup and the reservation
> +	 */
> +	chan = dma_get_slave_channel(candidate);
> +	if (!chan)
> +		goto retry;

I think it will be better to implement this functionality in the core. This
means we can hold the dma_list_mutex and don't have to do this retry loop
and do not have to peek at the client_count field in non-core code.
Something like dma_get_free_slave_channel(), which would call
private_candidate() followed by dma_chan_get().

> +
> +	if (info->post_alloc) {
> +		ret = info->post_alloc(chan, dma_spec);

If you need to do something at the end of the function I think it is nicer
to just wrap this function with your own function instead of adding a callback.

- Lars