[PATCH] at91: sam9x5 dt: add usart dma definitions to dt

[PATCH] at91: sam9x5 dt: add usart dma definitions to dt

[Jiri Prchal]

This patch adds usart dma definitions to both dtsi for sam9x5 chips. Without
usage of dma it's unable to catch all bytes on usart receiver.

Signed-off-by: Jiri Prchal <jiri.prchal@aksignal.cz>

v2: Add only definitions, enable them in board file.
---
 arch/arm/boot/dts/at91sam9x5.dtsi        | 12 ++++++++++++
 arch/arm/boot/dts/at91sam9x5_usart3.dtsi |  3 +++
 2 files changed, 15 insertions(+)

diff --git a/arch/arm/boot/dts/at91sam9x5.dtsi b/arch/arm/boot/dts/at91sam9x5.dtsi
index 726274f..bbb3ba6 100644
--- a/arch/arm/boot/dts/at91sam9x5.dtsi
+++ b/arch/arm/boot/dts/at91sam9x5.dtsi
@@ -860,6 +860,9 @@
 				interrupts = <1 IRQ_TYPE_LEVEL_HIGH 7>;
 				pinctrl-names = "default";
 				pinctrl-0 = <&pinctrl_dbgu>;
+				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(8)>,
+				       <&dma1 1 (AT91_DMA_CFG_PER_ID(9) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
+				dma-names = "tx", "rx";
 				clocks = <&mck>;
 				clock-names = "usart";
 				status = "disabled";
@@ -871,6 +874,9 @@
 				interrupts = <5 IRQ_TYPE_LEVEL_HIGH 5>;
 				pinctrl-names = "default";
 				pinctrl-0 = <&pinctrl_usart0>;
+				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(3)>,
+				       <&dma0 1 (AT91_DMA_CFG_PER_ID(4) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
+				dma-names = "tx", "rx";
 				clocks = <&usart0_clk>;
 				clock-names = "usart";
 				status = "disabled";
@@ -882,6 +888,9 @@
 				interrupts = <6 IRQ_TYPE_LEVEL_HIGH 5>;
 				pinctrl-names = "default";
 				pinctrl-0 = <&pinctrl_usart1>;
+				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(5)>,
+				       <&dma0 1 (AT91_DMA_CFG_PER_ID(6) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
+				dma-names = "tx", "rx";
 				clocks = <&usart1_clk>;
 				clock-names = "usart";
 				status = "disabled";
@@ -893,6 +902,9 @@
 				interrupts = <7 IRQ_TYPE_LEVEL_HIGH 5>;
 				pinctrl-names = "default";
 				pinctrl-0 = <&pinctrl_usart2>;
+				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(12)>,
+				       <&dma1 1 (AT91_DMA_CFG_PER_ID(13) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
+				dma-names = "tx", "rx";
 				clocks = <&usart2_clk>;
 				clock-names = "usart";
 				status = "disabled";
diff --git a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
index 140217a5..43bb5b5 100644
--- a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
+++ b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
@@ -57,6 +57,9 @@
 				interrupts = <8 IRQ_TYPE_LEVEL_HIGH 5>;
 				pinctrl-names = "default";
 				pinctrl-0 = <&pinctrl_usart3>;
+				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(14)>,
+				       <&dma1 1 (AT91_DMA_CFG_PER_ID(15) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
+				dma-names = "tx", "rx";
 				clocks = <&usart3_clk>;
 				clock-names = "usart";
 				status = "disabled";
-- 
1.9.1

[PATCH] at91: sam9x5 dt: add usart dma definitions to dt

[Ludovic Desroches]

On Mon, Oct 13, 2014 at 11:02:16AM +0200, Jiri Prchal wrote:
> This patch adds usart dma definitions to both dtsi for sam9x5 chips. Without
> usage of dma it's unable to catch all bytes on usart receiver.
> 
> Signed-off-by: Jiri Prchal <jiri.prchal@aksignal.cz>
Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>

> 
> v2: Add only definitions, enable them in board file.
> ---
>  arch/arm/boot/dts/at91sam9x5.dtsi        | 12 ++++++++++++
>  arch/arm/boot/dts/at91sam9x5_usart3.dtsi |  3 +++
>  2 files changed, 15 insertions(+)
> 
> diff --git a/arch/arm/boot/dts/at91sam9x5.dtsi b/arch/arm/boot/dts/at91sam9x5.dtsi
> index 726274f..bbb3ba6 100644
> --- a/arch/arm/boot/dts/at91sam9x5.dtsi
> +++ b/arch/arm/boot/dts/at91sam9x5.dtsi
> @@ -860,6 +860,9 @@
>  				interrupts = <1 IRQ_TYPE_LEVEL_HIGH 7>;
>  				pinctrl-names = "default";
>  				pinctrl-0 = <&pinctrl_dbgu>;
> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(8)>,
> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(9) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
> +				dma-names = "tx", "rx";
>  				clocks = <&mck>;
>  				clock-names = "usart";
>  				status = "disabled";
> @@ -871,6 +874,9 @@
>  				interrupts = <5 IRQ_TYPE_LEVEL_HIGH 5>;
>  				pinctrl-names = "default";
>  				pinctrl-0 = <&pinctrl_usart0>;
> +				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(3)>,
> +				       <&dma0 1 (AT91_DMA_CFG_PER_ID(4) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
> +				dma-names = "tx", "rx";
>  				clocks = <&usart0_clk>;
>  				clock-names = "usart";
>  				status = "disabled";
> @@ -882,6 +888,9 @@
>  				interrupts = <6 IRQ_TYPE_LEVEL_HIGH 5>;
>  				pinctrl-names = "default";
>  				pinctrl-0 = <&pinctrl_usart1>;
> +				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(5)>,
> +				       <&dma0 1 (AT91_DMA_CFG_PER_ID(6) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
> +				dma-names = "tx", "rx";
>  				clocks = <&usart1_clk>;
>  				clock-names = "usart";
>  				status = "disabled";
> @@ -893,6 +902,9 @@
>  				interrupts = <7 IRQ_TYPE_LEVEL_HIGH 5>;
>  				pinctrl-names = "default";
>  				pinctrl-0 = <&pinctrl_usart2>;
> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(12)>,
> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(13) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
> +				dma-names = "tx", "rx";
>  				clocks = <&usart2_clk>;
>  				clock-names = "usart";
>  				status = "disabled";
> diff --git a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
> index 140217a5..43bb5b5 100644
> --- a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
> +++ b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
> @@ -57,6 +57,9 @@
>  				interrupts = <8 IRQ_TYPE_LEVEL_HIGH 5>;
>  				pinctrl-names = "default";
>  				pinctrl-0 = <&pinctrl_usart3>;
> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(14)>,
> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(15) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
> +				dma-names = "tx", "rx";
>  				clocks = <&usart3_clk>;
>  				clock-names = "usart";
>  				status = "disabled";
> -- 
> 1.9.1
> 
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH] at91: sam9x5 dt: add usart dma definitions to dt

[Nicolas Ferre]

On 14/10/2014 14:24, Ludovic Desroches :
> On Mon, Oct 13, 2014 at 11:02:16AM +0200, Jiri Prchal wrote:
>> This patch adds usart dma definitions to both dtsi for sam9x5 chips. Without
>> usage of dma it's unable to catch all bytes on usart receiver.
>>
>> Signed-off-by: Jiri Prchal <jiri.prchal@aksignal.cz>
> Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>

Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>


Stacked on top of at91-3.19-dt. Thanks!

>> v2: Add only definitions, enable them in board file.
>> ---
>>  arch/arm/boot/dts/at91sam9x5.dtsi        | 12 ++++++++++++
>>  arch/arm/boot/dts/at91sam9x5_usart3.dtsi |  3 +++
>>  2 files changed, 15 insertions(+)
>>
>> diff --git a/arch/arm/boot/dts/at91sam9x5.dtsi b/arch/arm/boot/dts/at91sam9x5.dtsi
>> index 726274f..bbb3ba6 100644
>> --- a/arch/arm/boot/dts/at91sam9x5.dtsi
>> +++ b/arch/arm/boot/dts/at91sam9x5.dtsi
>> @@ -860,6 +860,9 @@
>>  				interrupts = <1 IRQ_TYPE_LEVEL_HIGH 7>;
>>  				pinctrl-names = "default";
>>  				pinctrl-0 = <&pinctrl_dbgu>;
>> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(8)>,
>> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(9) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
>> +				dma-names = "tx", "rx";
>>  				clocks = <&mck>;
>>  				clock-names = "usart";
>>  				status = "disabled";
>> @@ -871,6 +874,9 @@
>>  				interrupts = <5 IRQ_TYPE_LEVEL_HIGH 5>;
>>  				pinctrl-names = "default";
>>  				pinctrl-0 = <&pinctrl_usart0>;
>> +				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(3)>,
>> +				       <&dma0 1 (AT91_DMA_CFG_PER_ID(4) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
>> +				dma-names = "tx", "rx";
>>  				clocks = <&usart0_clk>;
>>  				clock-names = "usart";
>>  				status = "disabled";
>> @@ -882,6 +888,9 @@
>>  				interrupts = <6 IRQ_TYPE_LEVEL_HIGH 5>;
>>  				pinctrl-names = "default";
>>  				pinctrl-0 = <&pinctrl_usart1>;
>> +				dmas = <&dma0 1 AT91_DMA_CFG_PER_ID(5)>,
>> +				       <&dma0 1 (AT91_DMA_CFG_PER_ID(6) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
>> +				dma-names = "tx", "rx";
>>  				clocks = <&usart1_clk>;
>>  				clock-names = "usart";
>>  				status = "disabled";
>> @@ -893,6 +902,9 @@
>>  				interrupts = <7 IRQ_TYPE_LEVEL_HIGH 5>;
>>  				pinctrl-names = "default";
>>  				pinctrl-0 = <&pinctrl_usart2>;
>> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(12)>,
>> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(13) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
>> +				dma-names = "tx", "rx";
>>  				clocks = <&usart2_clk>;
>>  				clock-names = "usart";
>>  				status = "disabled";
>> diff --git a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
>> index 140217a5..43bb5b5 100644
>> --- a/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
>> +++ b/arch/arm/boot/dts/at91sam9x5_usart3.dtsi
>> @@ -57,6 +57,9 @@
>>  				interrupts = <8 IRQ_TYPE_LEVEL_HIGH 5>;
>>  				pinctrl-names = "default";
>>  				pinctrl-0 = <&pinctrl_usart3>;
>> +				dmas = <&dma1 1 AT91_DMA_CFG_PER_ID(14)>,
>> +				       <&dma1 1 (AT91_DMA_CFG_PER_ID(15) | AT91_DMA_CFG_FIFOCFG_ASAP)>;
>> +				dma-names = "tx", "rx";
>>  				clocks = <&usart3_clk>;
>>  				clock-names = "usart";
>>  				status = "disabled";
>> -- 
>> 1.9.1
>>
>>
>> _______________________________________________
>> linux-arm-kernel mailing list
>> linux-arm-kernel at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
> 


-- 
Nicolas Ferre

at91: dt: dma usage

[Alexandre Belloni]

Hi,

On 22/10/2014 at 10:53:43 +0200, Ji?? Prchal wrote :
> in fact that in at91sam chips has a lot of peripherals but only 2*8 dma channels
> and as my patch in last week "at91: sam9x5 dt: add usart dma definitions to dt"
> adds only dma definitions, not make to use them, leave it on board definition,
> shouldn't do the same all other peripherals definitions?
> I think that I2C, SPI are not as necessary as USART, but thous are always "on".

Not much discussion happened on that and I have the feeling this is
something we have to sort out. Until the dma driver are able to allocate
channels dynamically, it is true that this may be something we shouldn't
enable in the dtsi and let people activate dma in the board dts.

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

at91: dt: dma usage

[Arnd Bergmann]

On Wednesday 17 December 2014 21:16:22 Alexandre Belloni wrote:
> On 22/10/2014 at 10:53:43 +0200, Ji?? Prchal wrote :
> > in fact that in at91sam chips has a lot of peripherals but only 2*8 dma channels
> > and as my patch in last week "at91: sam9x5 dt: add usart dma definitions to dt"
> > adds only dma definitions, not make to use them, leave it on board definition,
> > shouldn't do the same all other peripherals definitions?
> > I think that I2C, SPI are not as necessary as USART, but thous are always "on".
> 
> Not much discussion happened on that and I have the feeling this is
> something we have to sort out. Until the dma driver are able to allocate
> channels dynamically, it is true that this may be something we shouldn't
> enable in the dtsi and let people activate dma in the board dts.

Could you use the drivers/dma/virt-dma.c infrastructure in the driver
to do this? Are you already planning to use it?

	Arnd

at91: dt: dma usage

[Ludovic Desroches]

On Wed, Dec 17, 2014 at 10:49:55PM +0100, Arnd Bergmann wrote:
> On Wednesday 17 December 2014 21:16:22 Alexandre Belloni wrote:
> > On 22/10/2014 at 10:53:43 +0200, Ji?? Prchal wrote :
> > > in fact that in at91sam chips has a lot of peripherals but only 2*8 dma channels
> > > and as my patch in last week "at91: sam9x5 dt: add usart dma definitions to dt"
> > > adds only dma definitions, not make to use them, leave it on board definition,
> > > shouldn't do the same all other peripherals definitions?
> > > I think that I2C, SPI are not as necessary as USART, but thous are always "on".
> > 
> > Not much discussion happened on that and I have the feeling this is
> > something we have to sort out. Until the dma driver are able to allocate
> > channels dynamically, it is true that this may be something we shouldn't
> > enable in the dtsi and let people activate dma in the board dts.

People can override dma-names to disable dma use for a device.

> 
> Could you use the drivers/dma/virt-dma.c infrastructure in the driver
> to do this? Are you already planning to use it?

Yes we could use virtual channels. We planned to use it for xdmac but
there is more channels so it's not a high priority. Concerning hdmac, it
will be useful due to the low number of channels but, at the moment,
it's not planned to switch to virtual channels.


Ludovic