Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1357668 > unrolled thread

[PATCH v2 4/5] mmc: davinci: don't use dma platform resources

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH v2 4/5] mmc: davinci: don't use dma platform resources David Lechner <david@lechnology.com> - 2016-03-15 00:00 +0100
    Re: [PATCH v2 4/5] mmc: davinci: don't use dma platform resources Sekhar Nori <nsekhar@ti.com> - 2016-03-15 09:20 +0100
    Re: [PATCH v2 4/5] mmc: davinci: don't use dma platform resources Peter Ujfalusi <peter.ujfalusi@ti.com> - 2016-03-15 10:00 +0100
    Re: [PATCH v2 4/5] mmc: davinci: don't use dma platform resources Peter Ujfalusi <peter.ujfalusi@ti.com> - 2016-03-15 10:00 +0100
      Re: [PATCH v2 4/5] mmc: davinci: don't use dma platform resources David Lechner <david@lechnology.com> - 2016-03-15 18:20 +0100
        Re: [PATCH v2 4/5] mmc: davinci: don't use dma platform resources Peter Ujfalusi <peter.ujfalusi@ti.com> - 2016-03-16 09:50 +0100

#1357668 — [PATCH v2 4/5] mmc: davinci: don't use dma platform resources

The davinci arch now has dma_slave_map tables for dma resources, so it is
no longer necessary to pass dma resources through the platform device.

Signed-off-by: David Lechner <david@lechnology.com>
---

v2 changes: Remove platform_get_resource completly instead of just ignoring it.


 drivers/mmc/host/davinci_mmc.c | 19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)

diff --git a/drivers/mmc/host/davinci_mmc.c b/drivers/mmc/host/davinci_mmc.c
index 8d10a92..dc96401 100644
--- a/drivers/mmc/host/davinci_mmc.c
+++ b/drivers/mmc/host/davinci_mmc.c
@@ -202,7 +202,6 @@ struct mmc_davinci_host {
 	u32 buffer_bytes_left;
 	u32 bytes_left;
 
-	u32 rxdma, txdma;
 	struct dma_chan *dma_tx;
 	struct dma_chan *dma_rx;
 	bool use_dma;
@@ -520,16 +519,14 @@ static int __init davinci_acquire_dma_channels(struct mmc_davinci_host *host)
 	dma_cap_set(DMA_SLAVE, mask);
 
 	host->dma_tx =
-		dma_request_slave_channel_compat(mask, edma_filter_fn,
-				&host->txdma, mmc_dev(host->mmc), "tx");
+		dma_request_slave_channel(mmc_dev(host->mmc), "tx");
 	if (!host->dma_tx) {
 		dev_err(mmc_dev(host->mmc), "Can't get dma_tx channel\n");
 		return -ENODEV;
 	}
 
 	host->dma_rx =
-		dma_request_slave_channel_compat(mask, edma_filter_fn,
-				&host->rxdma, mmc_dev(host->mmc), "rx");
+		dma_request_slave_channel(mmc_dev(host->mmc), "rx");
 	if (!host->dma_rx) {
 		dev_err(mmc_dev(host->mmc), "Can't get dma_rx channel\n");
 		r = -ENODEV;
@@ -1251,18 +1248,6 @@ static int __init davinci_mmcsd_probe(struct platform_device *pdev)
 	host = mmc_priv(mmc);
 	host->mmc = mmc;	/* Important */
 
-	r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
-	if (!r)
-		dev_warn(&pdev->dev, "RX DMA resource not specified\n");
-	else
-		host->rxdma = r->start;
-
-	r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
-	if (!r)
-		dev_warn(&pdev->dev, "TX DMA resource not specified\n");
-	else
-		host->txdma = r->start;
-
 	host->mem_res = mem;
 	host->base = devm_ioremap(&pdev->dev, mem->start, mem_size);
 	if (!host->base)
-- 
1.9.1

[toc] | [next] | [standalone]

#1357884

+ Peter

On Tuesday 15 March 2016 04:24 AM, David Lechner wrote:
> The davinci arch now has dma_slave_map tables for dma resources, so it is
> no longer necessary to pass dma resources through the platform device.
> 
> Signed-off-by: David Lechner <david@lechnology.com>

Looks good to me.

Acked-by: Sekhar Nori <nsekhar@ti.com>

Thanks,
Sekhar

> ---
> 
> v2 changes: Remove platform_get_resource completly instead of just ignoring it.
> 
> 
>  drivers/mmc/host/davinci_mmc.c | 19 ++-----------------
>  1 file changed, 2 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/mmc/host/davinci_mmc.c b/drivers/mmc/host/davinci_mmc.c
> index 8d10a92..dc96401 100644
> --- a/drivers/mmc/host/davinci_mmc.c
> +++ b/drivers/mmc/host/davinci_mmc.c
> @@ -202,7 +202,6 @@ struct mmc_davinci_host {
>  	u32 buffer_bytes_left;
>  	u32 bytes_left;
>  
> -	u32 rxdma, txdma;
>  	struct dma_chan *dma_tx;
>  	struct dma_chan *dma_rx;
>  	bool use_dma;
> @@ -520,16 +519,14 @@ static int __init davinci_acquire_dma_channels(struct mmc_davinci_host *host)
>  	dma_cap_set(DMA_SLAVE, mask);
>  
>  	host->dma_tx =
> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
> -				&host->txdma, mmc_dev(host->mmc), "tx");
> +		dma_request_slave_channel(mmc_dev(host->mmc), "tx");
>  	if (!host->dma_tx) {
>  		dev_err(mmc_dev(host->mmc), "Can't get dma_tx channel\n");
>  		return -ENODEV;
>  	}
>  
>  	host->dma_rx =
> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
> -				&host->rxdma, mmc_dev(host->mmc), "rx");
> +		dma_request_slave_channel(mmc_dev(host->mmc), "rx");
>  	if (!host->dma_rx) {
>  		dev_err(mmc_dev(host->mmc), "Can't get dma_rx channel\n");
>  		r = -ENODEV;
> @@ -1251,18 +1248,6 @@ static int __init davinci_mmcsd_probe(struct platform_device *pdev)
>  	host = mmc_priv(mmc);
>  	host->mmc = mmc;	/* Important */
>  
> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
> -	if (!r)
> -		dev_warn(&pdev->dev, "RX DMA resource not specified\n");
> -	else
> -		host->rxdma = r->start;
> -
> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
> -	if (!r)
> -		dev_warn(&pdev->dev, "TX DMA resource not specified\n");
> -	else
> -		host->txdma = r->start;
> -
>  	host->mem_res = mem;
>  	host->base = devm_ioremap(&pdev->dev, mem->start, mem_size);
>  	if (!host->base)
> 

[toc] | [prev] | [next] | [standalone]

#1357915

On 03/15/16 10:54, Peter Ujfalusi wrote:
> On 03/15/16 00:54, David Lechner wrote:
>> The davinci arch now has dma_slave_map tables for dma resources, so it is
>> no longer necessary to pass dma resources through the platform device.
>>
>> Signed-off-by: David Lechner <david@lechnology.com>
>> ---
>>
>> v2 changes: Remove platform_get_resource completly instead of just ignoring it.
>>
>>
>>  drivers/mmc/host/davinci_mmc.c | 19 ++-----------------
>>  1 file changed, 2 insertions(+), 17 deletions(-)
>>
>> diff --git a/drivers/mmc/host/davinci_mmc.c b/drivers/mmc/host/davinci_mmc.c
>> index 8d10a92..dc96401 100644
>> --- a/drivers/mmc/host/davinci_mmc.c
>> +++ b/drivers/mmc/host/davinci_mmc.c
>> @@ -202,7 +202,6 @@ struct mmc_davinci_host {
>>  	u32 buffer_bytes_left;
>>  	u32 bytes_left;
>>  
>> -	u32 rxdma, txdma;
>>  	struct dma_chan *dma_tx;
>>  	struct dma_chan *dma_rx;
>>  	bool use_dma;
>> @@ -520,16 +519,14 @@ static int __init davinci_acquire_dma_channels(struct mmc_davinci_host *host)
>>  	dma_cap_set(DMA_SLAVE, mask);
>>  
>>  	host->dma_tx =
>> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
>> -				&host->txdma, mmc_dev(host->mmc), "tx");
>> +		dma_request_slave_channel(mmc_dev(host->mmc), "tx");
> 
> you would need to use dma_request_chan() to be able to rely on the legacy
> channel mapping.

In other words: w/o dma_request_chan() legacy boot will not work as you will
not get the DMA channel.

> I have staged commits for converting all daVinci and OMAP drivers, I'll attach
> the patch I have for davinci-mmc for reference.
> When we convert to use the dma_request_chan() we can handle deferred probing
> also...
> 
> 
> 
>>  	if (!host->dma_tx) {
>>  		dev_err(mmc_dev(host->mmc), "Can't get dma_tx channel\n");
>>  		return -ENODEV;
>>  	}
>>  
>>  	host->dma_rx =
>> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
>> -				&host->rxdma, mmc_dev(host->mmc), "rx");
>> +		dma_request_slave_channel(mmc_dev(host->mmc), "rx");
>>  	if (!host->dma_rx) {
>>  		dev_err(mmc_dev(host->mmc), "Can't get dma_rx channel\n");
>>  		r = -ENODEV;
>> @@ -1251,18 +1248,6 @@ static int __init davinci_mmcsd_probe(struct platform_device *pdev)
>>  	host = mmc_priv(mmc);
>>  	host->mmc = mmc;	/* Important */
>>  
>> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
>> -	if (!r)
>> -		dev_warn(&pdev->dev, "RX DMA resource not specified\n");
>> -	else
>> -		host->rxdma = r->start;
>> -
>> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
>> -	if (!r)
>> -		dev_warn(&pdev->dev, "TX DMA resource not specified\n");
>> -	else
>> -		host->txdma = r->start;
>> -
>>  	host->mem_res = mem;
>>  	host->base = devm_ioremap(&pdev->dev, mem->start, mem_size);
>>  	if (!host->base)
>>
> 
> 


-- 
Péter

[toc] | [prev] | [next] | [standalone]

#1357916

[Multipart message — attachments visible in raw view] — view raw

On 03/15/16 00:54, David Lechner wrote:
> The davinci arch now has dma_slave_map tables for dma resources, so it is
> no longer necessary to pass dma resources through the platform device.
> 
> Signed-off-by: David Lechner <david@lechnology.com>
> ---
> 
> v2 changes: Remove platform_get_resource completly instead of just ignoring it.
> 
> 
>  drivers/mmc/host/davinci_mmc.c | 19 ++-----------------
>  1 file changed, 2 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/mmc/host/davinci_mmc.c b/drivers/mmc/host/davinci_mmc.c
> index 8d10a92..dc96401 100644
> --- a/drivers/mmc/host/davinci_mmc.c
> +++ b/drivers/mmc/host/davinci_mmc.c
> @@ -202,7 +202,6 @@ struct mmc_davinci_host {
>  	u32 buffer_bytes_left;
>  	u32 bytes_left;
>  
> -	u32 rxdma, txdma;
>  	struct dma_chan *dma_tx;
>  	struct dma_chan *dma_rx;
>  	bool use_dma;
> @@ -520,16 +519,14 @@ static int __init davinci_acquire_dma_channels(struct mmc_davinci_host *host)
>  	dma_cap_set(DMA_SLAVE, mask);
>  
>  	host->dma_tx =
> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
> -				&host->txdma, mmc_dev(host->mmc), "tx");
> +		dma_request_slave_channel(mmc_dev(host->mmc), "tx");

you would need to use dma_request_chan() to be able to rely on the legacy
channel mapping.
I have staged commits for converting all daVinci and OMAP drivers, I'll attach
the patch I have for davinci-mmc for reference.
When we convert to use the dma_request_chan() we can handle deferred probing
also...



>  	if (!host->dma_tx) {
>  		dev_err(mmc_dev(host->mmc), "Can't get dma_tx channel\n");
>  		return -ENODEV;
>  	}
>  
>  	host->dma_rx =
> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
> -				&host->rxdma, mmc_dev(host->mmc), "rx");
> +		dma_request_slave_channel(mmc_dev(host->mmc), "rx");
>  	if (!host->dma_rx) {
>  		dev_err(mmc_dev(host->mmc), "Can't get dma_rx channel\n");
>  		r = -ENODEV;
> @@ -1251,18 +1248,6 @@ static int __init davinci_mmcsd_probe(struct platform_device *pdev)
>  	host = mmc_priv(mmc);
>  	host->mmc = mmc;	/* Important */
>  
> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
> -	if (!r)
> -		dev_warn(&pdev->dev, "RX DMA resource not specified\n");
> -	else
> -		host->rxdma = r->start;
> -
> -	r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
> -	if (!r)
> -		dev_warn(&pdev->dev, "TX DMA resource not specified\n");
> -	else
> -		host->txdma = r->start;
> -
>  	host->mem_res = mem;
>  	host->base = devm_ioremap(&pdev->dev, mem->start, mem_size);
>  	if (!host->base)
> 


-- 
Péter

[toc] | [prev] | [next] | [standalone]

#1358160

On 03/15/2016 03:54 AM, Peter Ujfalusi wrote:
> On 03/15/16 00:54, David Lechner wrote:
>>
>>   	host->dma_tx =
>> -		dma_request_slave_channel_compat(mask, edma_filter_fn,
>> -				&host->txdma, mmc_dev(host->mmc), "tx");
>> +		dma_request_slave_channel(mmc_dev(host->mmc), "tx");
>
> you would need to use dma_request_chan() to be able to rely on the legacy
> channel mapping.
> I have staged commits for converting all daVinci and OMAP drivers, I'll attach
> the patch I have for davinci-mmc for reference.
> When we convert to use the dma_request_chan() we can handle deferred probing
> also...
>

OK. So I should redo this patch on top of the patch you attached?

[toc] | [prev] | [next] | [standalone]

#1358722

On 03/15/16 19:14, David Lechner wrote:
> On 03/15/2016 03:54 AM, Peter Ujfalusi wrote:
>> On 03/15/16 00:54, David Lechner wrote:
>>>
>>>       host->dma_tx =
>>> -        dma_request_slave_channel_compat(mask, edma_filter_fn,
>>> -                &host->txdma, mmc_dev(host->mmc), "tx");
>>> +        dma_request_slave_channel(mmc_dev(host->mmc), "tx");
>>
>> you would need to use dma_request_chan() to be able to rely on the legacy
>> channel mapping.
>> I have staged commits for converting all daVinci and OMAP drivers, I'll attach
>> the patch I have for davinci-mmc for reference.
>> When we convert to use the dma_request_chan() we can handle deferred probing
>> also...
>>
> 
> OK. So I should redo this patch on top of the patch you attached?

Or replace your patch with mine in your series. In my patch I forgot to remove
the 'u32 rxdma, txdma;' from struct mmc_davinci_host...

-- 
Péter

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web