Re: [PATCH] DMA: PL330: Add runtime pm support

[Vinod Koul]

On Wed, 2012-10-17 at 09:42 +0530, Inderpal Singh wrote:
> At the pl330's probe point the device is already in runtime resume state.
> Hence to manage the device with runtime, the probe should do pm_runtime_put
> and remove should do pm_runtime_get to balance with probe.
> 
> And in between, the device is being get/put at alloc_chan_resources and
> free_chan_resources.
> 
> Signed-off-by: Inderpal Singh <inderpal.singh@xxxxxxxxxx>
> ---
> This patch is based on slave-dma's next branch and on top
> of the clean up patches at [1].
> 
> [1] https://lkml.org/lkml/2012/10/5/169
> 
>  drivers/dma/pl330.c |   10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/drivers/dma/pl330.c b/drivers/dma/pl330.c
> index 2ee1538..5ae19ea 100644
> --- a/drivers/dma/pl330.c
> +++ b/drivers/dma/pl330.c
> @@ -25,6 +25,7 @@
>  #include <linux/amba/pl330.h>
>  #include <linux/scatterlist.h>
>  #include <linux/of.h>
> +#include <linux/pm_runtime.h>
>  
>  #include "dmaengine.h"
>  #define PL330_MAX_CHAN		8
> @@ -2384,6 +2385,8 @@ static int pl330_alloc_chan_resources(struct dma_chan *chan)
>  	struct dma_pl330_dmac *pdmac = pch->dmac;
>  	unsigned long flags;
>  
> +	pm_runtime_get_sync(pdmac->ddma.dev);
it would be really good if we move this to tx_submit. That way you
resume the dmaengine when actual transactions happen, not when someone
requests a channel.
Yes thats a little more work, though initially we can have this way.
> +
>  	spin_lock_irqsave(&pch->lock, flags);
>  
>  	dma_cookie_init(chan);
> @@ -2392,6 +2395,7 @@ static int pl330_alloc_chan_resources(struct dma_chan *chan)
>  	pch->pl330_chid = pl330_request_channel(&pdmac->pif);
>  	if (!pch->pl330_chid) {
>  		spin_unlock_irqrestore(&pch->lock, flags);
> +		pm_runtime_put(pdmac->ddma.dev);
>  		return -ENOMEM;
>  	}
>  
> @@ -2470,6 +2474,8 @@ static void pl330_free_chan_resources(struct dma_chan *chan)
>  		list_splice_tail_init(&pch->work_list, &pch->dmac->desc_pool);
>  
>  	spin_unlock_irqrestore(&pch->lock, flags);
> +
> +	pm_runtime_put(pch->dmac->ddma.dev);
>  }
>  
>  static enum dma_status
> @@ -2974,6 +2980,8 @@ pl330_probe(struct amba_device *adev, const struct amba_id *id)
>  		pi->pcfg.data_bus_width / 8, pi->pcfg.num_chan,
>  		pi->pcfg.num_peri, pi->pcfg.num_events);
>  
> +	pm_runtime_put(pd->dev);
pm_runtime_put_noidle is better. Yous should also do pm_runtime_allow
here and pm_runtime_forbid in your remove.

> +
>  	return 0;
>  
>  probe_err5:
> @@ -3017,6 +3025,8 @@ static int __devexit pl330_remove(struct amba_device *adev)
>  			return -EBUSY;
>  	}
>  
> +	pm_runtime_get(pdmac->ddma.dev);
> +
>  	dma_async_device_unregister(&pdmac->ddma);
>  
>  	amba_set_drvdata(adev, NULL);


-- 
Vinod Koul
Intel Corp.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/