[PATCH v4] dmaengine: qcom-bam-dma: Add pm_runtime support

[PATCH v4] dmaengine: qcom-bam-dma: Add pm_runtime support

[Pramod Gurav]

Adds pm_runtime support for BAM DMA so that clock is enabled only
when there is a transaction going on to help save power.

Signed-off-by: Pramod Gurav <pramod.gurav@linaro.org>
---
Changes in v3:
- pm_runtime_get_sync returns < 0 on error hence chech for return < 0
- Move pm_runtime_get_sync/put in bam_start_dma
- Tested large size file than tested with previous patch

Changes in v2:
- Handled return values of pm_runtime_get_sync to return error
- Rework remove function

 drivers/dma/qcom/bam_dma.c | 110 ++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 109 insertions(+), 1 deletion(-)

diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
index 969b481..4754891 100644
--- a/drivers/dma/qcom/bam_dma.c
+++ b/drivers/dma/qcom/bam_dma.c
@@ -48,6 +48,7 @@
 #include <linux/of_dma.h>
 #include <linux/clk.h>
 #include <linux/dmaengine.h>
+#include <linux/pm_runtime.h>
 
 #include "../dmaengine.h"
 #include "../virt-dma.h"
@@ -58,6 +59,8 @@ struct bam_desc_hw {
 	__le16 flags;
 };
 
+#define BAM_DMA_AUTOSUSPEND_DELAY 100
+
 #define DESC_FLAG_INT BIT(15)
 #define DESC_FLAG_EOT BIT(14)
 #define DESC_FLAG_EOB BIT(13)
@@ -527,12 +530,17 @@ static void bam_free_chan(struct dma_chan *chan)
 	struct bam_device *bdev = bchan->bdev;
 	u32 val;
 	unsigned long flags;
+	int ret;
+
+	ret = pm_runtime_get_sync(bdev->dev);
+	if (ret < 0)
+		return;
 
 	vchan_free_chan_resources(to_virt_chan(chan));
 
 	if (bchan->curr_txd) {
 		dev_err(bchan->bdev->dev, "Cannot free busy channel\n");
-		return;
+		goto err;
 	}
 
 	spin_lock_irqsave(&bchan->vc.lock, flags);
@@ -550,6 +558,10 @@ static void bam_free_chan(struct dma_chan *chan)
 
 	/* disable irq */
 	writel_relaxed(0, bam_addr(bdev, bchan->id, BAM_P_IRQ_EN));
+
+err:
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
 }
 
 /**
@@ -696,11 +708,18 @@ static int bam_pause(struct dma_chan *chan)
 	struct bam_chan *bchan = to_bam_chan(chan);
 	struct bam_device *bdev = bchan->bdev;
 	unsigned long flag;
+	int ret;
+
+	ret = pm_runtime_get_sync(bdev->dev);
+	if (ret < 0)
+		return ret;
 
 	spin_lock_irqsave(&bchan->vc.lock, flag);
 	writel_relaxed(1, bam_addr(bdev, bchan->id, BAM_P_HALT));
 	bchan->paused = 1;
 	spin_unlock_irqrestore(&bchan->vc.lock, flag);
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
 
 	return 0;
 }
@@ -715,11 +734,18 @@ static int bam_resume(struct dma_chan *chan)
 	struct bam_chan *bchan = to_bam_chan(chan);
 	struct bam_device *bdev = bchan->bdev;
 	unsigned long flag;
+	int ret;
+
+	ret = pm_runtime_get_sync(bdev->dev);
+	if (ret < 0)
+		return ret;
 
 	spin_lock_irqsave(&bchan->vc.lock, flag);
 	writel_relaxed(0, bam_addr(bdev, bchan->id, BAM_P_HALT));
 	bchan->paused = 0;
 	spin_unlock_irqrestore(&bchan->vc.lock, flag);
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
 
 	return 0;
 }
@@ -795,6 +821,7 @@ static irqreturn_t bam_dma_irq(int irq, void *data)
 {
 	struct bam_device *bdev = data;
 	u32 clr_mask = 0, srcs = 0;
+	int ret;
 
 	srcs |= process_channel_irqs(bdev);
 
@@ -802,6 +829,10 @@ static irqreturn_t bam_dma_irq(int irq, void *data)
 	if (srcs & P_IRQ)
 		tasklet_schedule(&bdev->task);
 
+	ret = pm_runtime_get_sync(bdev->dev);
+	if (ret < 0)
+		return ret;
+
 	if (srcs & BAM_IRQ) {
 		clr_mask = readl_relaxed(bam_addr(bdev, 0, BAM_IRQ_STTS));
 
@@ -814,6 +845,9 @@ static irqreturn_t bam_dma_irq(int irq, void *data)
 		writel_relaxed(clr_mask, bam_addr(bdev, 0, BAM_IRQ_CLR));
 	}
 
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
+
 	return IRQ_HANDLED;
 }
 
@@ -893,6 +927,7 @@ static void bam_start_dma(struct bam_chan *bchan)
 	struct bam_desc_hw *desc;
 	struct bam_desc_hw *fifo = PTR_ALIGN(bchan->fifo_virt,
 					sizeof(struct bam_desc_hw));
+	int ret;
 
 	lockdep_assert_held(&bchan->vc.lock);
 
@@ -904,6 +939,10 @@ static void bam_start_dma(struct bam_chan *bchan)
 	async_desc = container_of(vd, struct bam_async_desc, vd);
 	bchan->curr_txd = async_desc;
 
+	ret = pm_runtime_get_sync(bdev->dev);
+	if (ret < 0)
+		return;
+
 	/* on first use, initialize the channel hardware */
 	if (!bchan->initialized)
 		bam_chan_init_hw(bchan, async_desc->dir);
@@ -946,6 +985,9 @@ static void bam_start_dma(struct bam_chan *bchan)
 	wmb();
 	writel_relaxed(bchan->tail * sizeof(struct bam_desc_hw),
 			bam_addr(bdev, bchan->id, BAM_P_EVNT_REG));
+
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
 }
 
 /**
@@ -970,6 +1012,7 @@ static void dma_tasklet(unsigned long data)
 			bam_start_dma(bchan);
 		spin_unlock_irqrestore(&bchan->vc.lock, flags);
 	}
+
 }
 
 /**
@@ -1213,6 +1256,13 @@ static int bam_dma_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_unregister_dma;
 
+	pm_runtime_irq_safe(&pdev->dev);
+	pm_runtime_set_autosuspend_delay(&pdev->dev, BAM_DMA_AUTOSUSPEND_DELAY);
+	pm_runtime_use_autosuspend(&pdev->dev);
+	pm_runtime_mark_last_busy(&pdev->dev);
+	pm_runtime_set_active(&pdev->dev);
+	pm_runtime_enable(&pdev->dev);
+
 	return 0;
 
 err_unregister_dma:
@@ -1233,6 +1283,8 @@ static int bam_dma_remove(struct platform_device *pdev)
 	struct bam_device *bdev = platform_get_drvdata(pdev);
 	u32 i;
 
+	pm_runtime_force_suspend(&pdev->dev);
+
 	of_dma_controller_free(pdev->dev.of_node);
 	dma_async_device_unregister(&bdev->common);
 
@@ -1260,11 +1312,67 @@ static int bam_dma_remove(struct platform_device *pdev)
 	return 0;
 }
 
+static int bam_dma_runtime_suspend(struct device *dev)
+{
+	struct bam_device *bdev = dev_get_drvdata(dev);
+
+	clk_disable(bdev->bamclk);
+
+	return 0;
+}
+
+static int bam_dma_runtime_resume(struct device *dev)
+{
+	struct bam_device *bdev = dev_get_drvdata(dev);
+	int ret;
+
+	ret = clk_enable(bdev->bamclk);
+	if (ret < 0) {
+		dev_err(dev, "clk_enable failed: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+#ifdef CONFIG_PM_SLEEP
+static int bam_dma_suspend(struct device *dev)
+{
+	struct bam_device *bdev = dev_get_drvdata(dev);
+
+	pm_runtime_force_suspend(dev);
+
+	clk_unprepare(bdev->bamclk);
+
+	return 0;
+}
+
+static int bam_dma_resume(struct device *dev)
+{
+	struct bam_device *bdev = dev_get_drvdata(dev);
+	int ret;
+
+	ret = clk_prepare(bdev->bamclk);
+	if (ret)
+		return ret;
+
+	pm_runtime_force_resume(dev);
+
+	return 0;
+}
+#endif
+
+static const struct dev_pm_ops bam_dma_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(bam_dma_suspend, bam_dma_resume)
+	SET_RUNTIME_PM_OPS(bam_dma_runtime_suspend, bam_dma_runtime_resume,
+				NULL)
+};
+
 static struct platform_driver bam_dma_driver = {
 	.probe = bam_dma_probe,
 	.remove = bam_dma_remove,
 	.driver = {
 		.name = "bam-dma-engine",
+		.pm = &bam_dma_pm_ops,
 		.of_match_table = bam_of_match,
 	},
 };
-- 
1.8.2.1

Re: [PATCH v4] dmaengine: qcom-bam-dma: Add pm_runtime support

[Vinod Koul]

On Fri, Jun 17, 2016 at 03:56:03PM +0530, Pramod Gurav wrote:
> Adds pm_runtime support for BAM DMA so that clock is enabled only
> when there is a transaction going on to help save power.

Applied, thanks

-- 
~Vinod