[PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Martin Kaiser]

According to the i.MX25 reference manuals, each SSI has four FIFOs. All
of those FIFOs can store up to 15 entries.

The fsl_ssi driver's internal default for the FIFO depth in 8. Set our
non-default FIFO depth explicitly in the Device Tree.

Signed-off-by: Martin Kaiser <martin@kaiser.cx>
---
 arch/arm/boot/dts/imx25.dtsi | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
index 9725705..cf70df2 100644
--- a/arch/arm/boot/dts/imx25.dtsi
+++ b/arch/arm/boot/dts/imx25.dtsi
@@ -269,6 +269,7 @@
 				dmas = <&sdma 24 1 0>,
 				       <&sdma 25 1 0>;
 				dma-names = "rx", "tx";
+				fsl,fifo-depth = <15>;
 				status = "disabled";
 			};
 
@@ -329,6 +330,7 @@
 				dmas = <&sdma 28 1 0>,
 				       <&sdma 29 1 0>;
 				dma-names = "rx", "tx";
+				fsl,fifo-depth = <15>;
 				status = "disabled";
 			};
 
-- 
2.1.4

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Fabio Estevam]

On Tue, Mar 6, 2018 at 6:58 PM, Martin Kaiser <martin@kaiser.cx> wrote:
> According to the i.MX25 reference manuals, each SSI has four FIFOs. All
> of those FIFOs can store up to 15 entries.
>
> The fsl_ssi driver's internal default for the FIFO depth in 8. Set our
> non-default FIFO depth explicitly in the Device Tree.
>
> Signed-off-by: Martin Kaiser <martin@kaiser.cx>

Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Lothar Waßmann]

Hi,

On Tue,  6 Mar 2018 22:58:19 +0100 Martin Kaiser wrote:
> According to the i.MX25 reference manuals, each SSI has four FIFOs. All
> of those FIFOs can store up to 15 entries.
> 
> The fsl_ssi driver's internal default for the FIFO depth in 8. Set our
> non-default FIFO depth explicitly in the Device Tree.
> 
> Signed-off-by: Martin Kaiser <martin@kaiser.cx>
> ---
>  arch/arm/boot/dts/imx25.dtsi | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> index 9725705..cf70df2 100644
> --- a/arch/arm/boot/dts/imx25.dtsi
> +++ b/arch/arm/boot/dts/imx25.dtsi
> @@ -269,6 +269,7 @@
>  				dmas = <&sdma 24 1 0>,
>  				       <&sdma 25 1 0>;
>  				dma-names = "rx", "tx";
> +				fsl,fifo-depth = <15>;
>  				status = "disabled";
>  			};
>  
> @@ -329,6 +330,7 @@
>  				dmas = <&sdma 28 1 0>,
>  				       <&sdma 29 1 0>;
>  				dma-names = "rx", "tx";
> +				fsl,fifo-depth = <15>;
>  				status = "disabled";
>  			};
>  
You are changing the global .dtsi file. Did you test this change with
all devices that are affected by it?


Lothar Waßmann

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Martin Kaiser]

Hi Lothar,

Thus wrote Lothar Waßmann (LW@KARO-electronics.de):

> > diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> > index 9725705..cf70df2 100644
> > --- a/arch/arm/boot/dts/imx25.dtsi
> > +++ b/arch/arm/boot/dts/imx25.dtsi
> > @@ -269,6 +269,7 @@
> >  				dmas = <&sdma 24 1 0>,
> >  				       <&sdma 25 1 0>;
> >  				dma-names = "rx", "tx";
> > +				fsl,fifo-depth = <15>;
> >  				status = "disabled";
> >  			};

> > @@ -329,6 +330,7 @@
> >  				dmas = <&sdma 28 1 0>,
> >  				       <&sdma 29 1 0>;
> >  				dma-names = "rx", "tx";
> > +				fsl,fifo-depth = <15>;
> >  				status = "disabled";
> >  			};

> You are changing the global .dtsi file. Did you test this change with
> all devices that are affected by it?

I changed the hardware description of the imx25 SSI to match the
reference manual.

I did test this change on an imx25 board with audio playback. This uses
the SSI description I modified. I verified that the driver is actually
taking the modified setting into account and that this causes no
problems.

As of today, this setting is used by the fsl_ssi driver to set the fifo
water level for dma requests.

Of course, I don't have access to the enitre range of supported imx25
boards and I don't think this is required for submitting patches.

Do you have any indication why this patch should not be merged?

Best regards,
Martin

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Lothar Waßmann]

Hi,

On Thu, 8 Mar 2018 16:38:32 +0100 Martin Kaiser wrote:
> Hi Lothar,
> 
> Thus wrote Lothar Waßmann (LW@KARO-electronics.de):
> 
> > > diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> > > index 9725705..cf70df2 100644
> > > --- a/arch/arm/boot/dts/imx25.dtsi
> > > +++ b/arch/arm/boot/dts/imx25.dtsi
> > > @@ -269,6 +269,7 @@
> > >  				dmas = <&sdma 24 1 0>,
> > >  				       <&sdma 25 1 0>;
> > >  				dma-names = "rx", "tx";
> > > +				fsl,fifo-depth = <15>;
> > >  				status = "disabled";
> > >  			};
> 
> > > @@ -329,6 +330,7 @@
> > >  				dmas = <&sdma 28 1 0>,
> > >  				       <&sdma 29 1 0>;
> > >  				dma-names = "rx", "tx";
> > > +				fsl,fifo-depth = <15>;
> > >  				status = "disabled";
> > >  			};
> 
> > You are changing the global .dtsi file. Did you test this change with
> > all devices that are affected by it?
> 
> I changed the hardware description of the imx25 SSI to match the
> reference manual.
> 
> I did test this change on an imx25 board with audio playback. This uses
> the SSI description I modified. I verified that the driver is actually
> taking the modified setting into account and that this causes no
> problems.
> 
> As of today, this setting is used by the fsl_ssi driver to set the fifo
> water level for dma requests.
> 
> Of course, I don't have access to the enitre range of supported imx25
> boards and I don't think this is required for submitting patches.
> 
> Do you have any indication why this patch should not be merged?
> 
Usually patches should not willy-nilly change the behaviour of existing
configurations unless they fix any real life bugs.


Lothar Waßmann

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Martin Kaiser]

Hi Lothar,

Thus wrote Lothar Waßmann (LW@KARO-electronics.de):

> On Thu, 8 Mar 2018 16:38:32 +0100 Martin Kaiser wrote:
> > Hi Lothar,

> > Thus wrote Lothar Waßmann (LW@KARO-electronics.de):

> > > > diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> > > > index 9725705..cf70df2 100644
> > > > --- a/arch/arm/boot/dts/imx25.dtsi
> > > > +++ b/arch/arm/boot/dts/imx25.dtsi
> > > > @@ -269,6 +269,7 @@
> > > >  				dmas = <&sdma 24 1 0>,
> > > >  				       <&sdma 25 1 0>;
> > > >  				dma-names = "rx", "tx";
> > > > +				fsl,fifo-depth = <15>;
> > > >  				status = "disabled";
> > > >  			};

> > > > @@ -329,6 +330,7 @@
> > > >  				dmas = <&sdma 28 1 0>,
> > > >  				       <&sdma 29 1 0>;
> > > >  				dma-names = "rx", "tx";
> > > > +				fsl,fifo-depth = <15>;
> > > >  				status = "disabled";
> > > >  			};

> > > You are changing the global .dtsi file. Did you test this change with
> > > all devices that are affected by it?

> > I changed the hardware description of the imx25 SSI to match the
> > reference manual.

> > I did test this change on an imx25 board with audio playback. This uses
> > the SSI description I modified. I verified that the driver is actually
> > taking the modified setting into account and that this causes no
> > problems.

> > As of today, this setting is used by the fsl_ssi driver to set the fifo
> > water level for dma requests.

> > Of course, I don't have access to the enitre range of supported imx25
> > boards and I don't think this is required for submitting patches.

> > Do you have any indication why this patch should not be merged?

> Usually patches should not willy-nilly change the behaviour of existing
> configurations unless they fix any real life bugs.

We both made our points, let the maintainer decide what to do with the
patch.

Thanks & Best regards,

   Martin

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Lucas Stach]

Hi Lothar,

Am Freitag, den 09.03.2018, 09:37 +0100 schrieb Lothar Waßmann:
> Hi,
> 
> On Thu, 8 Mar 2018 16:38:32 +0100 Martin Kaiser wrote:
> > Hi Lothar,
> > 
> > > > Thus wrote Lothar Waßmann (LW@KARO-electronics.de):
> > 
> > > > diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> > > > index 9725705..cf70df2 100644
> > > > --- a/arch/arm/boot/dts/imx25.dtsi
> > > > +++ b/arch/arm/boot/dts/imx25.dtsi
> > > > @@ -269,6 +269,7 @@
> > > > > > > >  				dmas = <&sdma 24 1 0>,
> > > > > > > >  				       <&sdma 25 1 0>;
> > > > > > > >  				dma-names = "rx", "tx";
> > > > > > > > +				fsl,fifo-depth = <15>;
> > > > > > > >  				status = "disabled";
> > > > > > > >  			};
> > > > @@ -329,6 +330,7 @@
> > > > > > > >  				dmas = <&sdma 28 1 0>,
> > > > > > > >  				       <&sdma 29 1 0>;
> > > > > > > >  				dma-names = "rx", "tx";
> > > > > > > > +				fsl,fifo-depth = <15>;
> > > > > > > >  				status = "disabled";
> > > >  			};
> > > You are changing the global .dtsi file. Did you test this change with
> > > all devices that are affected by it?
> > 
> > I changed the hardware description of the imx25 SSI to match the
> > reference manual.
> > 
> > I did test this change on an imx25 board with audio playback. This uses
> > the SSI description I modified. I verified that the driver is actually
> > taking the modified setting into account and that this causes no
> > problems.
> > 
> > As of today, this setting is used by the fsl_ssi driver to set the fifo
> > water level for dma requests.
> > 
> > Of course, I don't have access to the enitre range of supported imx25
> > boards and I don't think this is required for submitting patches.
> > 
> > Do you have any indication why this patch should not be merged?
> > 
> 
> Usually patches should not willy-nilly change the behaviour of existing
> configurations unless they fix any real life bugs.

With this logic we wouldn't be able to get most driver changes applied.
If it is matching the reference manual and has been tested on the
affected hardware it should be good to go.

If you know about any specific corner cases that might break with this
change, please speak up now. Don't reject patches based on the
overarching "it might break something" argument.

Regards,
Lucas

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Lothar Waßmann]

Hi,

On Fri, 09 Mar 2018 10:27:12 +0100 Lucas Stach wrote:
> Hi Lothar,
> 
> Am Freitag, den 09.03.2018, 09:37 +0100 schrieb Lothar Waßmann:
> > Hi,
> > 
> > On Thu, 8 Mar 2018 16:38:32 +0100 Martin Kaiser wrote:
> > > Hi Lothar,
> > > 
> > > > > Thus wrote Lothar Waßmann (LW@KARO-electronics.de):
> > > 
> > > > > diff --git a/arch/arm/boot/dts/imx25.dtsi b/arch/arm/boot/dts/imx25.dtsi
> > > > > index 9725705..cf70df2 100644
> > > > > --- a/arch/arm/boot/dts/imx25.dtsi
> > > > > +++ b/arch/arm/boot/dts/imx25.dtsi
> > > > > @@ -269,6 +269,7 @@
> > > > > > > > >  				dmas = <&sdma 24 1 0>,
> > > > > > > > >  				       <&sdma 25 1 0>;
> > > > > > > > >  				dma-names = "rx", "tx";
> > > > > > > > > +				fsl,fifo-depth = <15>;
> > > > > > > > >  				status = "disabled";
> > > > > > > > >  			};
> > > > > @@ -329,6 +330,7 @@
> > > > > > > > >  				dmas = <&sdma 28 1 0>,
> > > > > > > > >  				       <&sdma 29 1 0>;
> > > > > > > > >  				dma-names = "rx", "tx";
> > > > > > > > > +				fsl,fifo-depth = <15>;
> > > > > > > > >  				status = "disabled";
> > > > >  			};
> > > > You are changing the global .dtsi file. Did you test this change with
> > > > all devices that are affected by it?
> > > 
> > > I changed the hardware description of the imx25 SSI to match the
> > > reference manual.
> > > 
> > > I did test this change on an imx25 board with audio playback. This uses
> > > the SSI description I modified. I verified that the driver is actually
> > > taking the modified setting into account and that this causes no
> > > problems.
> > > 
> > > As of today, this setting is used by the fsl_ssi driver to set the fifo
> > > water level for dma requests.
> > > 
> > > Of course, I don't have access to the enitre range of supported imx25
> > > boards and I don't think this is required for submitting patches.
> > > 
> > > Do you have any indication why this patch should not be merged?
> > > 
> > 
> > Usually patches should not willy-nilly change the behaviour of existing
> > configurations unless they fix any real life bugs.
> 
> With this logic we wouldn't be able to get most driver changes applied.
> If it is matching the reference manual and has been tested on the
> affected hardware it should be good to go.
> 
> If you know about any specific corner cases that might break with this
> change, please speak up now. Don't reject patches based on the
> overarching "it might break something" argument.
> 
I didn't reject anything, I just wanted to make sure, that the
implications of this patch were sufficiently considered.


Lothar Waßmann

Re: [PATCH] ARM: dts: i.MX25: define SSI FIFO depth

[Shawn Guo]

On Tue, Mar 06, 2018 at 10:58:19PM +0100, Martin Kaiser wrote:
> According to the i.MX25 reference manuals, each SSI has four FIFOs. All
> of those FIFOs can store up to 15 entries.
> 
> The fsl_ssi driver's internal default for the FIFO depth in 8. Set our
> non-default FIFO depth explicitly in the Device Tree.
> 
> Signed-off-by: Martin Kaiser <martin@kaiser.cx>

Applied, thanks.