[PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Shubhrajyoti D]

In case there are dma errors currently the driver exits.
Make the spi driver fall back to pio mode in case of dma errors.

If the DMA engine is not selected the driver
exits.This patch makes the spi fall back to pio in that case.

Also adds a field dma_unusable to struct omap2_mcspi.
 
Signed-off-by: Shubhrajyoti D <shubhrajyoti@ti.com>
---
 drivers/spi/spi-omap2-mcspi.c |   21 +++++++++++++--------
 1 files changed, 13 insertions(+), 8 deletions(-)

diff --git a/drivers/spi/spi-omap2-mcspi.c b/drivers/spi/spi-omap2-mcspi.c
index bc47781..f243a39 100644
--- a/drivers/spi/spi-omap2-mcspi.c
+++ b/drivers/spi/spi-omap2-mcspi.c
@@ -129,6 +129,7 @@ struct omap2_mcspi {
 	struct omap2_mcspi_dma	*dma_channels;
 	struct device		*dev;
 	struct omap2_mcspi_regs ctx;
+	int dma_unusable;
 };
 
 struct omap2_mcspi_cs {
@@ -845,7 +846,7 @@ static int omap2_mcspi_setup(struct spi_device *spi)
 	if (!mcspi_dma->dma_rx || !mcspi_dma->dma_tx) {
 		ret = omap2_mcspi_request_dma(spi);
 		if (ret < 0)
-			return ret;
+			mcspi->dma_unusable = 1;
 	}
 
 	ret = omap2_mcspi_enable_clocks(mcspi);
@@ -956,7 +957,8 @@ static void omap2_mcspi_work(struct omap2_mcspi *mcspi, struct spi_message *m)
 				__raw_writel(0, cs->base
 						+ OMAP2_MCSPI_TX0);
 
-			if (m->is_dma_mapped || t->len >= DMA_MIN_BYTES)
+			if (!mcspi->dma_unusable && (m->is_dma_mapped ||
+						t->len >= DMA_MIN_BYTES))
 				count = omap2_mcspi_txrx_dma(spi, t);
 			else
 				count = omap2_mcspi_txrx_pio(spi, t);
@@ -1030,7 +1032,8 @@ static int omap2_mcspi_transfer_one_message(struct spi_master *master,
 			return -EINVAL;
 		}
 
-		if (m->is_dma_mapped || len < DMA_MIN_BYTES)
+		if (mcspi->dma_unusable || m->is_dma_mapped ||
+					len < DMA_MIN_BYTES)
 			continue;
 
 		if (tx_buf != NULL) {
@@ -1054,6 +1057,7 @@ static int omap2_mcspi_transfer_one_message(struct spi_master *master,
 				return -EINVAL;
 			}
 		}
+
 	}
 
 	omap2_mcspi_work(mcspi, m);
@@ -1216,9 +1220,12 @@ static int __devinit omap2_mcspi_probe(struct platform_device *pdev)
 		mcspi->dma_channels[i].dma_tx_sync_dev = dma_res->start;
 	}
 
-	if (status < 0)
-		goto dma_chnl_free;
-
+	if (status < 0) {
+		dev_err(&pdev->dev, "cannot get DMA channel switching to pio\n");
+		mcspi->dma_unusable = 1;
+		status = 0;
+		kfree(mcspi->dma_channels);
+	}
 	pm_runtime_use_autosuspend(&pdev->dev);
 	pm_runtime_set_autosuspend_delay(&pdev->dev, SPI_AUTOSUSPEND_TIMEOUT);
 	pm_runtime_enable(&pdev->dev);
@@ -1229,14 +1236,12 @@ static int __devinit omap2_mcspi_probe(struct platform_device *pdev)
 	status = spi_register_master(master);
 	if (status < 0)
 		goto err_spi_register;
-
 	return status;
 
 err_spi_register:
 	spi_master_put(master);
 disable_pm:
 	pm_runtime_disable(&pdev->dev);
-dma_chnl_free:
 	kfree(mcspi->dma_channels);
 free_master:
 	kfree(master);
-- 
1.7.5.4

Re: [PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Tony Lindgren]

* Shubhrajyoti D <shubhrajyoti@ti.com> [120724 23:26]:
> In case there are dma errors currently the driver exits.
> Make the spi driver fall back to pio mode in case of dma errors.
> 
> If the DMA engine is not selected the driver
> exits.This patch makes the spi fall back to pio in that case.
> 
> Also adds a field dma_unusable to struct omap2_mcspi.
>  
> Signed-off-by: Shubhrajyoti D <shubhrajyoti@ti.com>
> ---
>  drivers/spi/spi-omap2-mcspi.c |   21 +++++++++++++--------
>  1 files changed, 13 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/spi/spi-omap2-mcspi.c b/drivers/spi/spi-omap2-mcspi.c
> index bc47781..f243a39 100644
> --- a/drivers/spi/spi-omap2-mcspi.c
> +++ b/drivers/spi/spi-omap2-mcspi.c
> @@ -129,6 +129,7 @@ struct omap2_mcspi {
>  	struct omap2_mcspi_dma	*dma_channels;
>  	struct device		*dev;
>  	struct omap2_mcspi_regs ctx;
> +	int dma_unusable;
>  };

Don't you need to check separately for rx and tx dma? There's a slight
chance that you get a channel for one but not for the other..

Tony

Re: [PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Shubhrajyoti]

On Tuesday 07 August 2012 01:17 PM, Tony Lindgren wrote:
>>  };
> Don't you need to check separately for rx and tx dma? There's a slight
> chance that you get a channel for one but not for the other..
In that case I treat it as non usable and fall back to pio.

Are you suggesting that let one channel be dma and only the failed one pio?

Re: [PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Tony Lindgren]

* Shubhrajyoti <shubhrajyoti@ti.com> [120807 04:21]:
> On Tuesday 07 August 2012 01:17 PM, Tony Lindgren wrote:
> >>  };
> > Don't you need to check separately for rx and tx dma? There's a slight
> > chance that you get a channel for one but not for the other..
> In that case I treat it as non usable and fall back to pio.

OK that should work too.
 
> Are you suggesting that let one channel be dma and only the failed one pio?

I guess both are doable. For reduced CPU load using DMA where possible
of course is the best way to go.

Tony

Re: [PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Felipe Balbi]

[-- Attachment #1: Type: text/plain, Size: 749 bytes --]

Hi,

On Tue, Aug 07, 2012 at 05:02:27AM -0700, Tony Lindgren wrote:
> * Shubhrajyoti <shubhrajyoti@ti.com> [120807 04:21]:
> > On Tuesday 07 August 2012 01:17 PM, Tony Lindgren wrote:
> > >>  };
> > > Don't you need to check separately for rx and tx dma? There's a slight
> > > chance that you get a channel for one but not for the other..
> > In that case I treat it as non usable and fall back to pio.
> 
> OK that should work too.
>  
> > Are you suggesting that let one channel be dma and only the failed one pio?
> 
> I guess both are doable. For reduced CPU load using DMA where possible
> of course is the best way to go.

indeed. if you can get DMA for TX but not for RX, then at least TX can
go via DMA ;-)

-- 
balbi

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

RE: [PATCH] spi: omap2-mcspi: In case of dma errors fall back to pio

[Philip, Avinash]

On Wed, Jul 25, 2012 at 11:50:29, Datta, Shubhrajyoti wrote:
> In case there are dma errors currently the driver exits.
> Make the spi driver fall back to pio mode in case of dma errors.
> 
> If the DMA engine is not selected the driver
> exits.This patch makes the spi fall back to pio in that case.
> 
> Also adds a field dma_unusable to struct omap2_mcspi.
>  
> Signed-off-by: Shubhrajyoti D <shubhrajyoti-l0cyMroinI0@public.gmane.org>

I have tested this patch for SPI PIO mode on AM335x EVM.

Acked-by: Philip, Avinash <avinashphilip-l0cyMroinI0@public.gmane.org>


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/