[RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[Jon Hunter]

From: Jonathan Hunter <jonathanh@nvidia.com>

The Tegra210 ADMA supports two modes for transferring data to a FIFO
which are ...

1. Transfer data to/from the FIFO as soon as a single burst can be
   transferred.
2. Transfer data to/from the FIFO based upon FIFO thresholds, where
   the FIFO threshold is specified in terms on multiple bursts.

Currently, the ADMA driver programs the FIFO threshold values in the
FIFO_CTRL register, but never enables the transfer mode that uses
these threshold values. Given that these have never been used so far,
simplify the ADMA driver by removing the programming of these threshold
values.

Signed-off-by: Jonathan Hunter <jonathanh@nvidia.com>
Acked-by: Thierry Reding <treding@nvidia.com>
---

Resending the patch rebased on top next-20190704. I have added Thierry's
ACK as well.

 drivers/dma/tegra210-adma.c | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/drivers/dma/tegra210-adma.c b/drivers/dma/tegra210-adma.c
index 2805853e963f..d8646a49ba5b 100644
--- a/drivers/dma/tegra210-adma.c
+++ b/drivers/dma/tegra210-adma.c
@@ -42,12 +42,8 @@
 #define ADMA_CH_CONFIG_MAX_BUFS				8
 
 #define ADMA_CH_FIFO_CTRL				0x2c
-#define TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0xf) << 24)
-#define TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0xf) << 16)
 #define TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0xf) << 8)
 #define TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0xf)
-#define TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0x1f) << 24)
-#define TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0x1f) << 16)
 #define TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0x1f) << 8)
 #define TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0x1f)
 
@@ -64,14 +60,10 @@
 
 #define TEGRA_ADMA_BURST_COMPLETE_TIME			20
 
-#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
-				    TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
-				    TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
+#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
 				    TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(3))
 
-#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
-				    TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
-				    TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
+#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
 				    TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(3))
 
 #define ADMA_CH_REG_FIELD_VAL(val, mask, shift)	(((val) & mask) << shift)
-- 
2.17.1

Re: [RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[Sameer Pujar]

On 7/5/2019 2:45 PM, Jon Hunter wrote:
> From: Jonathan Hunter <jonathanh@nvidia.com>
>
> The Tegra210 ADMA supports two modes for transferring data to a FIFO
> which are ...
>
> 1. Transfer data to/from the FIFO as soon as a single burst can be
>     transferred.
> 2. Transfer data to/from the FIFO based upon FIFO thresholds, where
>     the FIFO threshold is specified in terms on multiple bursts.
>
> Currently, the ADMA driver programs the FIFO threshold values in the
> FIFO_CTRL register, but never enables the transfer mode that uses
> these threshold values. Given that these have never been used so far,
> simplify the ADMA driver by removing the programming of these threshold
> values.
>
> Signed-off-by: Jonathan Hunter <jonathanh@nvidia.com>
> Acked-by: Thierry Reding <treding@nvidia.com>
> ---
>
> Resending the patch rebased on top next-20190704. I have added Thierry's
> ACK as well.
>
>   drivers/dma/tegra210-adma.c | 12 ++----------
>   1 file changed, 2 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/dma/tegra210-adma.c b/drivers/dma/tegra210-adma.c
> index 2805853e963f..d8646a49ba5b 100644
> --- a/drivers/dma/tegra210-adma.c
> +++ b/drivers/dma/tegra210-adma.c
> @@ -42,12 +42,8 @@
>   #define ADMA_CH_CONFIG_MAX_BUFS				8
>   
>   #define ADMA_CH_FIFO_CTRL				0x2c
> -#define TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0xf) << 24)
> -#define TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0xf) << 16)
>   #define TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0xf) << 8)
>   #define TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0xf)
> -#define TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0x1f) << 24)
> -#define TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0x1f) << 16)
>   #define TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0x1f) << 8)
>   #define TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0x1f)
>   
> @@ -64,14 +60,10 @@
>   
>   #define TEGRA_ADMA_BURST_COMPLETE_TIME			20
>   
> -#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
> -				    TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
> -				    TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
> +#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
>   				    TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(3))
>   
> -#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
> -				    TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
> -				    TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
> +#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
>   				    TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(3))
>   
>   #define ADMA_CH_REG_FIELD_VAL(val, mask, shift)	(((val) & mask) << shift)
Acked-by: Sameer Pujar <spujar@nvidia.com>

Re: [RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[Vinod Koul]

On 05-07-19, 10:15, Jon Hunter wrote:
> From: Jonathan Hunter <jonathanh@nvidia.com>
> 
> The Tegra210 ADMA supports two modes for transferring data to a FIFO
> which are ...
> 
> 1. Transfer data to/from the FIFO as soon as a single burst can be
>    transferred.
> 2. Transfer data to/from the FIFO based upon FIFO thresholds, where
>    the FIFO threshold is specified in terms on multiple bursts.
> 
> Currently, the ADMA driver programs the FIFO threshold values in the
> FIFO_CTRL register, but never enables the transfer mode that uses
> these threshold values. Given that these have never been used so far,
> simplify the ADMA driver by removing the programming of these threshold
> values.
> 
> Signed-off-by: Jonathan Hunter <jonathanh@nvidia.com>
> Acked-by: Thierry Reding <treding@nvidia.com>
> ---
> 
> Resending the patch rebased on top next-20190704. I have added Thierry's
> ACK as well.

Thanks but this fails as well. I had applied few tegra patches so I
suspect that is causing issues now. It would have been nice to have them
in series.

Would you rebase on
git.kernel.org/pub/scm/linux/kernel/git/vkoul/slave-dma.git next (yeah
this is different location, i dont see to push to infradead today)

Thanks 

> 
>  drivers/dma/tegra210-adma.c | 12 ++----------
>  1 file changed, 2 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/dma/tegra210-adma.c b/drivers/dma/tegra210-adma.c
> index 2805853e963f..d8646a49ba5b 100644
> --- a/drivers/dma/tegra210-adma.c
> +++ b/drivers/dma/tegra210-adma.c
> @@ -42,12 +42,8 @@
>  #define ADMA_CH_CONFIG_MAX_BUFS				8
>  
>  #define ADMA_CH_FIFO_CTRL				0x2c
> -#define TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0xf) << 24)
> -#define TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0xf) << 16)
>  #define TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0xf) << 8)
>  #define TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0xf)
> -#define TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(val)	(((val) & 0x1f) << 24)
> -#define TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(val)	(((val) & 0x1f) << 16)
>  #define TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(val)		(((val) & 0x1f) << 8)
>  #define TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(val)		((val) & 0x1f)
>  
> @@ -64,14 +60,10 @@
>  
>  #define TEGRA_ADMA_BURST_COMPLETE_TIME			20
>  
> -#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
> -				    TEGRA210_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
> -				    TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
> +#define TEGRA210_FIFO_CTRL_DEFAULT (TEGRA210_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
>  				    TEGRA210_ADMA_CH_FIFO_CTRL_RXSIZE(3))
>  
> -#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_OFLWTHRES(1) | \
> -				    TEGRA186_ADMA_CH_FIFO_CTRL_STRVTHRES(1) | \
> -				    TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3)    | \
> +#define TEGRA186_FIFO_CTRL_DEFAULT (TEGRA186_ADMA_CH_FIFO_CTRL_TXSIZE(3) | \
>  				    TEGRA186_ADMA_CH_FIFO_CTRL_RXSIZE(3))
>  
>  #define ADMA_CH_REG_FIELD_VAL(val, mask, shift)	(((val) & mask) << shift)
> -- 
> 2.17.1

-- 
~Vinod

Re: [RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[Jon Hunter]

On 05/07/2019 14:05, Vinod Koul wrote:
> On 05-07-19, 10:15, Jon Hunter wrote:
>> From: Jonathan Hunter <jonathanh@nvidia.com>
>>
>> The Tegra210 ADMA supports two modes for transferring data to a FIFO
>> which are ...
>>
>> 1. Transfer data to/from the FIFO as soon as a single burst can be
>>    transferred.
>> 2. Transfer data to/from the FIFO based upon FIFO thresholds, where
>>    the FIFO threshold is specified in terms on multiple bursts.
>>
>> Currently, the ADMA driver programs the FIFO threshold values in the
>> FIFO_CTRL register, but never enables the transfer mode that uses
>> these threshold values. Given that these have never been used so far,
>> simplify the ADMA driver by removing the programming of these threshold
>> values.
>>
>> Signed-off-by: Jonathan Hunter <jonathanh@nvidia.com>
>> Acked-by: Thierry Reding <treding@nvidia.com>
>> ---
>>
>> Resending the patch rebased on top next-20190704. I have added Thierry's
>> ACK as well.
> 
> Thanks but this fails as well. I had applied few tegra patches so I
> suspect that is causing issues now. It would have been nice to have them
> in series.
> 
> Would you rebase on
> git.kernel.org/pub/scm/linux/kernel/git/vkoul/slave-dma.git next (yeah
> this is different location, i dont see to push to infradead today)

So this patch should apply cleanly on top of the fixes series I sent for
v5.2 [0] which you merged and is now in mainline. So if I rebase on the
above, I wondering if it is then going to conflict with mainline? Looks
like the above branch is based upon v5.2-rc1 and hence the conflict.

Cheers
Jon

[0] https://patchwork.kernel.org/cover/10946849/

-- 
nvpublic

Re: [RESEND PATCH] dmaengine: tegra210-adma: Don't program FIFO threshold

[Vinod Koul]

On 10-07-19, 10:11, Jon Hunter wrote:
> 
> On 05/07/2019 14:05, Vinod Koul wrote:
> > On 05-07-19, 10:15, Jon Hunter wrote:
> >> From: Jonathan Hunter <jonathanh@nvidia.com>
> >>
> >> The Tegra210 ADMA supports two modes for transferring data to a FIFO
> >> which are ...
> >>
> >> 1. Transfer data to/from the FIFO as soon as a single burst can be
> >>    transferred.
> >> 2. Transfer data to/from the FIFO based upon FIFO thresholds, where
> >>    the FIFO threshold is specified in terms on multiple bursts.
> >>
> >> Currently, the ADMA driver programs the FIFO threshold values in the
> >> FIFO_CTRL register, but never enables the transfer mode that uses
> >> these threshold values. Given that these have never been used so far,
> >> simplify the ADMA driver by removing the programming of these threshold
> >> values.
> >>
> >> Signed-off-by: Jonathan Hunter <jonathanh@nvidia.com>
> >> Acked-by: Thierry Reding <treding@nvidia.com>
> >> ---
> >>
> >> Resending the patch rebased on top next-20190704. I have added Thierry's
> >> ACK as well.
> > 
> > Thanks but this fails as well. I had applied few tegra patches so I
> > suspect that is causing issues now. It would have been nice to have them
> > in series.
> > 
> > Would you rebase on
> > git.kernel.org/pub/scm/linux/kernel/git/vkoul/slave-dma.git next (yeah
> > this is different location, i dont see to push to infradead today)
> 
> So this patch should apply cleanly on top of the fixes series I sent for
> v5.2 [0] which you merged and is now in mainline. So if I rebase on the
> above, I wondering if it is then going to conflict with mainline? Looks
> like the above branch is based upon v5.2-rc1 and hence the conflict.

Can you resend this now that merge window is closed and dependent
patches are merged.

Thanks
-- 
~Vinod