[v2] dmaengine: rcar-dmac: don't use DMAC error interrupt

Message ID 8736xrkrci.wl%kuninori.morimoto.gx@renesas.com
State Changes Requested
Headers show
Series
  • [v2] dmaengine: rcar-dmac: don't use DMAC error interrupt
Related show

Commit Message

Kuninori Morimoto June 13, 2018, 8:49 a.m.
From: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>

Current rcar-dmac is using DMAC error interrupt which will handle all
channel's error. But in this design, error handling itself will be
issue if user want to use virtualization, multi OS, etc.
This patch removes current DMAC error interrupt handling, and handle it
on each channel interrupt handler.

Signed-off-by: Magnus Damm <damm+renesas@opensource.se>
[Kuninori: updated patch to adjust DMACHCR/DMAOR]
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Tested-by: Nguyen Viet Dung <nv-dung@jinso.co.jp>
---
v1 -> v2

 - add [Kuninori: xxx] line on git log
 - don't remove DT property
 - tidyup rcar_dmac_isr_channel()

 .../devicetree/bindings/dma/renesas,rcar-dmac.txt  |  1 +
 drivers/dma/sh/rcar-dmac.c                         | 72 +++++++---------------
 2 files changed, 23 insertions(+), 50 deletions(-)

Comments

Geert Uytterhoeven June 13, 2018, 11:36 a.m. | #1
Hi Morimoto-san,

Thanks for your patch!

On Wed, Jun 13, 2018 at 10:49 AM Kuninori Morimoto
<kuninori.morimoto.gx@renesas.com> wrote:
> From: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
>
> Current rcar-dmac is using DMAC error interrupt which will handle all
> channel's error. But in this design, error handling itself will be
> issue if user want to use virtualization, multi OS, etc.
> This patch removes current DMAC error interrupt handling, and handle it
> on each channel interrupt handler.

Just wondering, what's the usage pattern where this causes a problem?
Do you plan to export part of a SYS-DMAC instance to a guest?
How to maintain host/guest isolation?

> --- a/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> +++ b/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> @@ -38,6 +38,7 @@ Required Properties:
>  - interrupt-names: one entry for the error interrupt, named "error", plus one
>    entry per channel, named "ch%u", where %u is the channel number ranging from
>    zero to the number of channels minus one.
> +  # "error" interrupt will be ignored, so far

Again, DT describes hardware, not software policy.
So please drop all changes to the bindings.

With that fixed:
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Thanks!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Vinod Koul June 14, 2018, 8:34 a.m. | #2
On 13-06-18, 08:49, Kuninori Morimoto wrote:
> 
> From: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> 
> Current rcar-dmac is using DMAC error interrupt which will handle all
> channel's error. But in this design, error handling itself will be
> issue if user want to use virtualization, multi OS, etc.
> This patch removes current DMAC error interrupt handling, and handle it
> on each channel interrupt handler.

So what happens when you are not in virtualization, multi OS..? Who
handles this interrupt?

> 
> Signed-off-by: Magnus Damm <damm+renesas@opensource.se>
> [Kuninori: updated patch to adjust DMACHCR/DMAOR]
> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> Tested-by: Nguyen Viet Dung <nv-dung@jinso.co.jp>
> ---
> v1 -> v2
> 
>  - add [Kuninori: xxx] line on git log
>  - don't remove DT property
>  - tidyup rcar_dmac_isr_channel()
> 
>  .../devicetree/bindings/dma/renesas,rcar-dmac.txt  |  1 +
>  drivers/dma/sh/rcar-dmac.c                         | 72 +++++++---------------
>  2 files changed, 23 insertions(+), 50 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt b/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> index b1ba6395..b8e1c13 100644
> --- a/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> +++ b/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
> @@ -38,6 +38,7 @@ Required Properties:
>  - interrupt-names: one entry for the error interrupt, named "error", plus one
>    entry per channel, named "ch%u", where %u is the channel number ranging from
>    zero to the number of channels minus one.
> +  # "error" interrupt will be ignored, so far
>  
>  - clock-names: "fck" for the functional clock
>  - clocks: a list of phandle + clock-specifier pairs, one for each entry
> diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c
> index 2a2ccd9..279c930 100644
> --- a/drivers/dma/sh/rcar-dmac.c
> +++ b/drivers/dma/sh/rcar-dmac.c
> @@ -431,7 +431,8 @@ static void rcar_dmac_chan_start_xfer(struct rcar_dmac_chan *chan)
>  		chcr |= RCAR_DMACHCR_DPM_DISABLED | RCAR_DMACHCR_IE;
>  	}
>  
> -	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr | RCAR_DMACHCR_DE);
> +	rcar_dmac_chan_write(chan, RCAR_DMACHCR,
> +			     chcr | RCAR_DMACHCR_DE | RCAR_DMACHCR_CAIE);
>  }
>  
>  static int rcar_dmac_init(struct rcar_dmac *dmac)
> @@ -783,7 +784,8 @@ static void rcar_dmac_chan_halt(struct rcar_dmac_chan *chan)
>  	u32 chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR);
>  
>  	chcr &= ~(RCAR_DMACHCR_DSE | RCAR_DMACHCR_DSIE | RCAR_DMACHCR_IE |
> -		  RCAR_DMACHCR_TE | RCAR_DMACHCR_DE);
> +		  RCAR_DMACHCR_TE | RCAR_DMACHCR_DE |
> +		  RCAR_DMACHCR_CAE | RCAR_DMACHCR_CAIE);
>  	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr);
>  	rcar_dmac_chcr_de_barrier(chan);
>  }
> @@ -812,12 +814,7 @@ static void rcar_dmac_chan_reinit(struct rcar_dmac_chan *chan)
>  	}
>  }
>  
> -static void rcar_dmac_stop(struct rcar_dmac *dmac)
> -{
> -	rcar_dmac_write(dmac, RCAR_DMAOR, 0);
> -}
> -
> -static void rcar_dmac_abort(struct rcar_dmac *dmac)
> +static void rcar_dmac_stop_all_chan(struct rcar_dmac *dmac)
>  {
>  	unsigned int i;
>  
> @@ -829,11 +826,10 @@ static void rcar_dmac_abort(struct rcar_dmac *dmac)
>  		spin_lock(&chan->lock);
>  		rcar_dmac_chan_halt(chan);
>  		spin_unlock(&chan->lock);
> -
> -		rcar_dmac_chan_reinit(chan);
>  	}
>  }
>  
> +
>  /* -----------------------------------------------------------------------------
>   * Descriptors preparation
>   */
> @@ -1522,11 +1518,18 @@ static irqreturn_t rcar_dmac_isr_channel(int irq, void *dev)
>  	u32 mask = RCAR_DMACHCR_DSE | RCAR_DMACHCR_TE;
>  	struct rcar_dmac_chan *chan = dev;
>  	irqreturn_t ret = IRQ_NONE;
> +	bool reinit = false;
>  	u32 chcr;
>  
>  	spin_lock(&chan->lock);
>  
>  	chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR);
> +	if (chcr & RCAR_DMACHCR_CAE) {
> +		rcar_dmac_chan_halt(chan);
> +		reinit = true;
> +		goto spin_lock_end;
> +	}
> +
>  	if (chcr & RCAR_DMACHCR_TE)
>  		mask |= RCAR_DMACHCR_DE;
>  	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr & ~mask);
> @@ -1539,8 +1542,16 @@ static irqreturn_t rcar_dmac_isr_channel(int irq, void *dev)
>  	if (chcr & RCAR_DMACHCR_TE)
>  		ret |= rcar_dmac_isr_transfer_end(chan);
>  
> +spin_lock_end:
>  	spin_unlock(&chan->lock);
>  
> +	if (reinit) {
> +		dev_err(chan->chan.device->dev, "Channel Address Error\n");
> +
> +		rcar_dmac_chan_reinit(chan);
> +		ret = IRQ_HANDLED;
> +	}
> +
>  	return ret;
>  }
>  
> @@ -1597,24 +1608,6 @@ static irqreturn_t rcar_dmac_isr_channel_thread(int irq, void *dev)
>  	return IRQ_HANDLED;
>  }
>  
> -static irqreturn_t rcar_dmac_isr_error(int irq, void *data)
> -{
> -	struct rcar_dmac *dmac = data;
> -
> -	if (!(rcar_dmac_read(dmac, RCAR_DMAOR) & RCAR_DMAOR_AE))
> -		return IRQ_NONE;
> -
> -	/*
> -	 * An unrecoverable error occurred on an unknown channel. Halt the DMAC,
> -	 * abort transfers on all channels, and reinitialize the DMAC.
> -	 */
> -	rcar_dmac_stop(dmac);
> -	rcar_dmac_abort(dmac);
> -	rcar_dmac_init(dmac);
> -
> -	return IRQ_HANDLED;
> -}
> -
>  /* -----------------------------------------------------------------------------
>   * OF xlate and channel filter
>   */
> @@ -1784,8 +1777,6 @@ static int rcar_dmac_probe(struct platform_device *pdev)
>  	struct rcar_dmac *dmac;
>  	struct resource *mem;
>  	unsigned int i;
> -	char *irqname;
> -	int irq;
>  	int ret;
>  
>  	dmac = devm_kzalloc(&pdev->dev, sizeof(*dmac), GFP_KERNEL);
> @@ -1824,17 +1815,6 @@ static int rcar_dmac_probe(struct platform_device *pdev)
>  	if (IS_ERR(dmac->iomem))
>  		return PTR_ERR(dmac->iomem);
>  
> -	irq = platform_get_irq_byname(pdev, "error");
> -	if (irq < 0) {
> -		dev_err(&pdev->dev, "no error IRQ specified\n");
> -		return -ENODEV;
> -	}
> -
> -	irqname = devm_kasprintf(dmac->dev, GFP_KERNEL, "%s:error",
> -				 dev_name(dmac->dev));
> -	if (!irqname)
> -		return -ENOMEM;
> -
>  	/* Enable runtime PM and initialize the device. */
>  	pm_runtime_enable(&pdev->dev);
>  	ret = pm_runtime_get_sync(&pdev->dev);
> @@ -1885,14 +1865,6 @@ static int rcar_dmac_probe(struct platform_device *pdev)
>  			goto error;
>  	}
>  
> -	ret = devm_request_irq(&pdev->dev, irq, rcar_dmac_isr_error, 0,
> -			       irqname, dmac);
> -	if (ret) {
> -		dev_err(&pdev->dev, "failed to request IRQ %u (%d)\n",
> -			irq, ret);
> -		return ret;
> -	}
> -
>  	/* Register the DMAC as a DMA provider for DT. */
>  	ret = of_dma_controller_register(pdev->dev.of_node, rcar_dmac_of_xlate,
>  					 NULL);
> @@ -1932,7 +1904,7 @@ static void rcar_dmac_shutdown(struct platform_device *pdev)
>  {
>  	struct rcar_dmac *dmac = platform_get_drvdata(pdev);
>  
> -	rcar_dmac_stop(dmac);
> +	rcar_dmac_stop_all_chan(dmac);
>  }
>  
>  static const struct of_device_id rcar_dmac_of_ids[] = {
> -- 
> 1.9.1
Kuninori Morimoto June 15, 2018, 12:24 a.m. | #3
Hi Vinod

Thank you for your feedback

> > From: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> > 
> > Current rcar-dmac is using DMAC error interrupt which will handle all
> > channel's error. But in this design, error handling itself will be
> > issue if user want to use virtualization, multi OS, etc.
> > This patch removes current DMAC error interrupt handling, and handle it
> > on each channel interrupt handler.
> 
> So what happens when you are not in virtualization, multi OS..? Who
> handles this interrupt?

Sorry for non clear English
It has 2types of interrupt.
 1) IRQ for Error
 2) IRQ for each channels

If error happens on some channels,
the error IRQ was handled by 1), and "all" channels will be restarted.
Thus, this patch removes 1), and handles each error on 2).

Best regards
---
Kuninori Morimoto
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Kuninori Morimoto June 15, 2018, 12:33 a.m. | #4
Hi Geert

Thank you for your feedback

> > Current rcar-dmac is using DMAC error interrupt which will handle all
> > channel's error. But in this design, error handling itself will be
> > issue if user want to use virtualization, multi OS, etc.
> > This patch removes current DMAC error interrupt handling, and handle it
> > on each channel interrupt handler.
> 
> Just wondering, what's the usage pattern where this causes a problem?
> Do you plan to export part of a SYS-DMAC instance to a guest?

"virtualization, multi OS" part is maybe misleading.
It will be problem on non virtualization, too.

> How to maintain host/guest isolation?

This is another topic/issue, I think

Best regards
---
Kuninori Morimoto
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Kuninori Morimoto June 15, 2018, 12:35 a.m. | #5
Hi, again

> > > Current rcar-dmac is using DMAC error interrupt which will handle all
> > > channel's error. But in this design, error handling itself will be
> > > issue if user want to use virtualization, multi OS, etc.
> > > This patch removes current DMAC error interrupt handling, and handle it
> > > on each channel interrupt handler.
> > 
> > Just wondering, what's the usage pattern where this causes a problem?
> > Do you plan to export part of a SYS-DMAC instance to a guest?
> 
> "virtualization, multi OS" part is maybe misleading.
> It will be problem on non virtualization, too.
> 
> > How to maintain host/guest isolation?
> 
> This is another topic/issue, I think

I will post v3 patch, soon


Best regards
---
Kuninori Morimoto
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt b/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
index b1ba6395..b8e1c13 100644
--- a/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
+++ b/Documentation/devicetree/bindings/dma/renesas,rcar-dmac.txt
@@ -38,6 +38,7 @@  Required Properties:
 - interrupt-names: one entry for the error interrupt, named "error", plus one
   entry per channel, named "ch%u", where %u is the channel number ranging from
   zero to the number of channels minus one.
+  # "error" interrupt will be ignored, so far
 
 - clock-names: "fck" for the functional clock
 - clocks: a list of phandle + clock-specifier pairs, one for each entry
diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c
index 2a2ccd9..279c930 100644
--- a/drivers/dma/sh/rcar-dmac.c
+++ b/drivers/dma/sh/rcar-dmac.c
@@ -431,7 +431,8 @@  static void rcar_dmac_chan_start_xfer(struct rcar_dmac_chan *chan)
 		chcr |= RCAR_DMACHCR_DPM_DISABLED | RCAR_DMACHCR_IE;
 	}
 
-	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr | RCAR_DMACHCR_DE);
+	rcar_dmac_chan_write(chan, RCAR_DMACHCR,
+			     chcr | RCAR_DMACHCR_DE | RCAR_DMACHCR_CAIE);
 }
 
 static int rcar_dmac_init(struct rcar_dmac *dmac)
@@ -783,7 +784,8 @@  static void rcar_dmac_chan_halt(struct rcar_dmac_chan *chan)
 	u32 chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR);
 
 	chcr &= ~(RCAR_DMACHCR_DSE | RCAR_DMACHCR_DSIE | RCAR_DMACHCR_IE |
-		  RCAR_DMACHCR_TE | RCAR_DMACHCR_DE);
+		  RCAR_DMACHCR_TE | RCAR_DMACHCR_DE |
+		  RCAR_DMACHCR_CAE | RCAR_DMACHCR_CAIE);
 	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr);
 	rcar_dmac_chcr_de_barrier(chan);
 }
@@ -812,12 +814,7 @@  static void rcar_dmac_chan_reinit(struct rcar_dmac_chan *chan)
 	}
 }
 
-static void rcar_dmac_stop(struct rcar_dmac *dmac)
-{
-	rcar_dmac_write(dmac, RCAR_DMAOR, 0);
-}
-
-static void rcar_dmac_abort(struct rcar_dmac *dmac)
+static void rcar_dmac_stop_all_chan(struct rcar_dmac *dmac)
 {
 	unsigned int i;
 
@@ -829,11 +826,10 @@  static void rcar_dmac_abort(struct rcar_dmac *dmac)
 		spin_lock(&chan->lock);
 		rcar_dmac_chan_halt(chan);
 		spin_unlock(&chan->lock);
-
-		rcar_dmac_chan_reinit(chan);
 	}
 }
 
+
 /* -----------------------------------------------------------------------------
  * Descriptors preparation
  */
@@ -1522,11 +1518,18 @@  static irqreturn_t rcar_dmac_isr_channel(int irq, void *dev)
 	u32 mask = RCAR_DMACHCR_DSE | RCAR_DMACHCR_TE;
 	struct rcar_dmac_chan *chan = dev;
 	irqreturn_t ret = IRQ_NONE;
+	bool reinit = false;
 	u32 chcr;
 
 	spin_lock(&chan->lock);
 
 	chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR);
+	if (chcr & RCAR_DMACHCR_CAE) {
+		rcar_dmac_chan_halt(chan);
+		reinit = true;
+		goto spin_lock_end;
+	}
+
 	if (chcr & RCAR_DMACHCR_TE)
 		mask |= RCAR_DMACHCR_DE;
 	rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr & ~mask);
@@ -1539,8 +1542,16 @@  static irqreturn_t rcar_dmac_isr_channel(int irq, void *dev)
 	if (chcr & RCAR_DMACHCR_TE)
 		ret |= rcar_dmac_isr_transfer_end(chan);
 
+spin_lock_end:
 	spin_unlock(&chan->lock);
 
+	if (reinit) {
+		dev_err(chan->chan.device->dev, "Channel Address Error\n");
+
+		rcar_dmac_chan_reinit(chan);
+		ret = IRQ_HANDLED;
+	}
+
 	return ret;
 }
 
@@ -1597,24 +1608,6 @@  static irqreturn_t rcar_dmac_isr_channel_thread(int irq, void *dev)
 	return IRQ_HANDLED;
 }
 
-static irqreturn_t rcar_dmac_isr_error(int irq, void *data)
-{
-	struct rcar_dmac *dmac = data;
-
-	if (!(rcar_dmac_read(dmac, RCAR_DMAOR) & RCAR_DMAOR_AE))
-		return IRQ_NONE;
-
-	/*
-	 * An unrecoverable error occurred on an unknown channel. Halt the DMAC,
-	 * abort transfers on all channels, and reinitialize the DMAC.
-	 */
-	rcar_dmac_stop(dmac);
-	rcar_dmac_abort(dmac);
-	rcar_dmac_init(dmac);
-
-	return IRQ_HANDLED;
-}
-
 /* -----------------------------------------------------------------------------
  * OF xlate and channel filter
  */
@@ -1784,8 +1777,6 @@  static int rcar_dmac_probe(struct platform_device *pdev)
 	struct rcar_dmac *dmac;
 	struct resource *mem;
 	unsigned int i;
-	char *irqname;
-	int irq;
 	int ret;
 
 	dmac = devm_kzalloc(&pdev->dev, sizeof(*dmac), GFP_KERNEL);
@@ -1824,17 +1815,6 @@  static int rcar_dmac_probe(struct platform_device *pdev)
 	if (IS_ERR(dmac->iomem))
 		return PTR_ERR(dmac->iomem);
 
-	irq = platform_get_irq_byname(pdev, "error");
-	if (irq < 0) {
-		dev_err(&pdev->dev, "no error IRQ specified\n");
-		return -ENODEV;
-	}
-
-	irqname = devm_kasprintf(dmac->dev, GFP_KERNEL, "%s:error",
-				 dev_name(dmac->dev));
-	if (!irqname)
-		return -ENOMEM;
-
 	/* Enable runtime PM and initialize the device. */
 	pm_runtime_enable(&pdev->dev);
 	ret = pm_runtime_get_sync(&pdev->dev);
@@ -1885,14 +1865,6 @@  static int rcar_dmac_probe(struct platform_device *pdev)
 			goto error;
 	}
 
-	ret = devm_request_irq(&pdev->dev, irq, rcar_dmac_isr_error, 0,
-			       irqname, dmac);
-	if (ret) {
-		dev_err(&pdev->dev, "failed to request IRQ %u (%d)\n",
-			irq, ret);
-		return ret;
-	}
-
 	/* Register the DMAC as a DMA provider for DT. */
 	ret = of_dma_controller_register(pdev->dev.of_node, rcar_dmac_of_xlate,
 					 NULL);
@@ -1932,7 +1904,7 @@  static void rcar_dmac_shutdown(struct platform_device *pdev)
 {
 	struct rcar_dmac *dmac = platform_get_drvdata(pdev);
 
-	rcar_dmac_stop(dmac);
+	rcar_dmac_stop_all_chan(dmac);
 }
 
 static const struct of_device_id rcar_dmac_of_ids[] = {