[PATCH 0/3] Add larbid init routine for mediatek's gen1 smi larb driver

[PATCH 0/3] Add larbid init routine for mediatek's gen1 smi larb driver

[honghui.zhang]

From: Honghui Zhang <honghui.zhang@mediatek.com>

Mediatek's gen1 smi need the hardware larbid to identify the offset for
the register which controls whether enable iommu for this larb.
In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
the larbid was used without properly initialized. This patchset fixed
that.

Honghui Zhang (3):
  memory: mtk-smi: add larbid init routine
  dt-bindings: mediatek: add descriptions for larbid
  arm: dts: mediatek: add larbid property for larb

 .../bindings/memory-controllers/mediatek,smi-larb.txt     | 15 +++++++++++++++
 arch/arm/boot/dts/mt2701.dtsi                             |  3 +++
 drivers/memory/mtk-smi.c                                  |  9 +++++++++
 3 files changed, 27 insertions(+)

-- 
2.6.4

[PATCH 1/3] memory: mtk-smi: add larbid init routine

[honghui.zhang]

From: Honghui Zhang <honghui.zhang@mediatek.com>

In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
the larb->larbid was added but not initialized.
Mediatek's gen1 smi need this hardware larbid information to get the
register offset which controls whether enable iommu for this larb.
This patch add the initialize routine for larbid.

Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
---
 drivers/memory/mtk-smi.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
index 2b798bb4..fe36b3e 100644
--- a/drivers/memory/mtk-smi.c
+++ b/drivers/memory/mtk-smi.c
@@ -240,6 +240,7 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
 	struct device *dev = &pdev->dev;
 	struct device_node *smi_node;
 	struct platform_device *smi_pdev;
+	int err;
 
 	if (!dev->pm_domain)
 		return -EPROBE_DEFER;
@@ -263,6 +264,14 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
 		return PTR_ERR(larb->smi.clk_smi);
 	larb->smi.dev = dev;
 
+	err = of_property_read_u32(dev->of_node, "mediatek,larbid",
+				   &larb->larbid);
+	if (err && of_device_is_compatible(dev->of_node,
+					   "mediatek,mt2701-smi-larb")) {
+		dev_err(dev, "missing larbid property\n");
+		return err;
+	}
+
 	smi_node = of_parse_phandle(dev->of_node, "mediatek,smi", 0);
 	if (!smi_node)
 		return -EINVAL;
-- 
2.6.4

Re: [PATCH 1/3] memory: mtk-smi: add larbid init routine

[Matthias Brugger]

On 07/27/2017 04:01 AM, honghui.zhang@mediatek.com wrote:
> From: Honghui Zhang <honghui.zhang@mediatek.com>
> 
> In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
> the larb->larbid was added but not initialized.
> Mediatek's gen1 smi need this hardware larbid information to get the
> register offset which controls whether enable iommu for this larb.
> This patch add the initialize routine for larbid.
> 
> Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
> ---
>   drivers/memory/mtk-smi.c | 9 +++++++++
>   1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
> index 2b798bb4..fe36b3e 100644
> --- a/drivers/memory/mtk-smi.c
> +++ b/drivers/memory/mtk-smi.c
> @@ -240,6 +240,7 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
>   	struct device *dev = &pdev->dev;
>   	struct device_node *smi_node;
>   	struct platform_device *smi_pdev;
> +	int err;
>   
>   	if (!dev->pm_domain)
>   		return -EPROBE_DEFER;
> @@ -263,6 +264,14 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
>   		return PTR_ERR(larb->smi.clk_smi);
>   	larb->smi.dev = dev;
>   
> +	err = of_property_read_u32(dev->of_node, "mediatek,larbid",
> +				   &larb->larbid);
> +	if (err && of_device_is_compatible(dev->of_node,
> +					   "mediatek,mt2701-smi-larb")) {

Instead of checking hard-coded bindings you should be able to determine the 
generation using the smi_pdev.
Apart I think it would be better to check for mediatek,larbid only on 
MTK_SIM_GEN1, or do I miss something?

Regards,
Matthias

> +		dev_err(dev, "missing larbid property\n");
> +		return err;
> +	}
> +
>   	smi_node = of_parse_phandle(dev->of_node, "mediatek,smi", 0);
>   	if (!smi_node)
>   		return -EINVAL;
> 

Re: [PATCH 1/3] memory: mtk-smi: add larbid init routine

[Honghui Zhang]

On Thu, 2017-07-27 at 17:42 +0200, Matthias Brugger wrote:
> 
> On 07/27/2017 04:01 AM, honghui.zhang@mediatek.com wrote:
> > From: Honghui Zhang <honghui.zhang@mediatek.com>
> > 
> > In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
> > the larb->larbid was added but not initialized.
> > Mediatek's gen1 smi need this hardware larbid information to get the
> > register offset which controls whether enable iommu for this larb.
> > This patch add the initialize routine for larbid.
> > 
> > Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
> > ---
> >   drivers/memory/mtk-smi.c | 9 +++++++++
> >   1 file changed, 9 insertions(+)
> > 
> > diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
> > index 2b798bb4..fe36b3e 100644
> > --- a/drivers/memory/mtk-smi.c
> > +++ b/drivers/memory/mtk-smi.c
> > @@ -240,6 +240,7 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
> >   	struct device *dev = &pdev->dev;
> >   	struct device_node *smi_node;
> >   	struct platform_device *smi_pdev;
> > +	int err;
> >   
> >   	if (!dev->pm_domain)
> >   		return -EPROBE_DEFER;
> > @@ -263,6 +264,14 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
> >   		return PTR_ERR(larb->smi.clk_smi);
> >   	larb->smi.dev = dev;
> >   
> > +	err = of_property_read_u32(dev->of_node, "mediatek,larbid",
> > +				   &larb->larbid);
> > +	if (err && of_device_is_compatible(dev->of_node,
> > +					   "mediatek,mt2701-smi-larb")) {
> 
> Instead of checking hard-coded bindings you should be able to determine the 
> generation using the smi_pdev.
> Apart I think it would be better to check for mediatek,larbid only on 
> MTK_SIM_GEN1, or do I miss something?
> 

Hi, Matthias,
	There's a new SoC of mediatek which use MTK_SMI_GEN2 hardware, but it
has 2 iommus and 2 smi_common, and we are intend to send it's driver
soon. It need the mediatek,larbid to identify which iommu should be
binding to.

This patch use the hard-code bindings will make the future work easier.

thanks.
> Regards,
> Matthias
> 
> > +		dev_err(dev, "missing larbid property\n");
> > +		return err;
> > +	}
> > +
> >   	smi_node = of_parse_phandle(dev->of_node, "mediatek,smi", 0);
> >   	if (!smi_node)
> >   		return -EINVAL;
> > 
> 
> _______________________________________________
> Linux-mediatek mailing list
> Linux-mediatek@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-mediatek

Re: [PATCH 1/3] memory: mtk-smi: add larbid init routine

[Robin Murphy]

On 28/07/17 02:15, Honghui Zhang wrote:
> On Thu, 2017-07-27 at 17:42 +0200, Matthias Brugger wrote:
>>
>> On 07/27/2017 04:01 AM, honghui.zhang@mediatek.com wrote:
>>> From: Honghui Zhang <honghui.zhang@mediatek.com>
>>>
>>> In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
>>> the larb->larbid was added but not initialized.
>>> Mediatek's gen1 smi need this hardware larbid information to get the
>>> register offset which controls whether enable iommu for this larb.
>>> This patch add the initialize routine for larbid.
>>>
>>> Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
>>> ---
>>>   drivers/memory/mtk-smi.c | 9 +++++++++
>>>   1 file changed, 9 insertions(+)
>>>
>>> diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
>>> index 2b798bb4..fe36b3e 100644
>>> --- a/drivers/memory/mtk-smi.c
>>> +++ b/drivers/memory/mtk-smi.c
>>> @@ -240,6 +240,7 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
>>>   	struct device *dev = &pdev->dev;
>>>   	struct device_node *smi_node;
>>>   	struct platform_device *smi_pdev;
>>> +	int err;
>>>   
>>>   	if (!dev->pm_domain)
>>>   		return -EPROBE_DEFER;
>>> @@ -263,6 +264,14 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
>>>   		return PTR_ERR(larb->smi.clk_smi);
>>>   	larb->smi.dev = dev;
>>>   
>>> +	err = of_property_read_u32(dev->of_node, "mediatek,larbid",
>>> +				   &larb->larbid);
>>> +	if (err && of_device_is_compatible(dev->of_node,
>>> +					   "mediatek,mt2701-smi-larb")) {
>>
>> Instead of checking hard-coded bindings you should be able to determine the 
>> generation using the smi_pdev.
>> Apart I think it would be better to check for mediatek,larbid only on 
>> MTK_SIM_GEN1, or do I miss something?
>>
> 
> Hi, Matthias,
> 	There's a new SoC of mediatek which use MTK_SMI_GEN2 hardware, but it
> has 2 iommus and 2 smi_common, and we are intend to send it's driver
> soon. It need the mediatek,larbid to identify which iommu should be
> binding to.
> 
> This patch use the hard-code bindings will make the future work easier.

In that case, might it make sense to add some indicator like "bool
need_larb_id" to struct mtk_smi_larb_gen?

Robin.

> 
> thanks.
>> Regards,
>> Matthias
>>
>>> +		dev_err(dev, "missing larbid property\n");
>>> +		return err;
>>> +	}
>>> +
>>>   	smi_node = of_parse_phandle(dev->of_node, "mediatek,smi", 0);
>>>   	if (!smi_node)
>>>   		return -EINVAL;
>>>
>>
>> _______________________________________________
>> Linux-mediatek mailing list
>> Linux-mediatek@lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-mediatek
> 
> 

Re: [PATCH 1/3] memory: mtk-smi: add larbid init routine

[Honghui Zhang]

On Fri, 2017-07-28 at 10:59 +0100, Robin Murphy wrote:
> On 28/07/17 02:15, Honghui Zhang wrote:
> > On Thu, 2017-07-27 at 17:42 +0200, Matthias Brugger wrote:
> >>
> >> On 07/27/2017 04:01 AM, honghui.zhang@mediatek.com wrote:
> >>> From: Honghui Zhang <honghui.zhang@mediatek.com>
> >>>
> >>> In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
> >>> the larb->larbid was added but not initialized.
> >>> Mediatek's gen1 smi need this hardware larbid information to get the
> >>> register offset which controls whether enable iommu for this larb.
> >>> This patch add the initialize routine for larbid.
> >>>
> >>> Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
> >>> ---
> >>>   drivers/memory/mtk-smi.c | 9 +++++++++
> >>>   1 file changed, 9 insertions(+)
> >>>
> >>> diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
> >>> index 2b798bb4..fe36b3e 100644
> >>> --- a/drivers/memory/mtk-smi.c
> >>> +++ b/drivers/memory/mtk-smi.c
> >>> @@ -240,6 +240,7 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
> >>>   	struct device *dev = &pdev->dev;
> >>>   	struct device_node *smi_node;
> >>>   	struct platform_device *smi_pdev;
> >>> +	int err;
> >>>   
> >>>   	if (!dev->pm_domain)
> >>>   		return -EPROBE_DEFER;
> >>> @@ -263,6 +264,14 @@ static int mtk_smi_larb_probe(struct platform_device *pdev)
> >>>   		return PTR_ERR(larb->smi.clk_smi);
> >>>   	larb->smi.dev = dev;
> >>>   
> >>> +	err = of_property_read_u32(dev->of_node, "mediatek,larbid",
> >>> +				   &larb->larbid);
> >>> +	if (err && of_device_is_compatible(dev->of_node,
> >>> +					   "mediatek,mt2701-smi-larb")) {
> >>
> >> Instead of checking hard-coded bindings you should be able to determine the 
> >> generation using the smi_pdev.
> >> Apart I think it would be better to check for mediatek,larbid only on 
> >> MTK_SIM_GEN1, or do I miss something?
> >>
> > 
> > Hi, Matthias,
> > 	There's a new SoC of mediatek which use MTK_SMI_GEN2 hardware, but it
> > has 2 iommus and 2 smi_common, and we are intend to send it's driver
> > soon. It need the mediatek,larbid to identify which iommu should be
> > binding to.
> > 
> > This patch use the hard-code bindings will make the future work easier.
> 
> In that case, might it make sense to add some indicator like "bool
> need_larb_id" to struct mtk_smi_larb_gen?
> 
> Robin.
> 

Hi, Robin, thanks for your advise, I will send another version later.

Thanks.

> > 
> > thanks.
> >> Regards,
> >> Matthias
> >>
> >>> +		dev_err(dev, "missing larbid property\n");
> >>> +		return err;
> >>> +	}
> >>> +
> >>>   	smi_node = of_parse_phandle(dev->of_node, "mediatek,smi", 0);
> >>>   	if (!smi_node)
> >>>   		return -EINVAL;
> >>>
> >>
> >> _______________________________________________
> >> Linux-mediatek mailing list
> >> Linux-mediatek@lists.infradead.org
> >> http://lists.infradead.org/mailman/listinfo/linux-mediatek
> > 
> > 
> 

[PATCH 2/3] dt-bindings: mediatek: add descriptions for larbid

[honghui.zhang]

From: Honghui Zhang <honghui.zhang@mediatek.com>

This patch add larbid descritptions for mediatek's gen1 smi larb hardware.

Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
---
 .../bindings/memory-controllers/mediatek,smi-larb.txt     | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/Documentation/devicetree/bindings/memory-controllers/mediatek,smi-larb.txt b/Documentation/devicetree/bindings/memory-controllers/mediatek,smi-larb.txt
index 21277a5..d1b115c 100644
--- a/Documentation/devicetree/bindings/memory-controllers/mediatek,smi-larb.txt
+++ b/Documentation/devicetree/bindings/memory-controllers/mediatek,smi-larb.txt
@@ -15,6 +15,9 @@ Required properties:
 	    the register.
   - "smi" : It's the clock for transfer data and command.
 
+Required property for mt2701:
+- mediatek,larbid :the hardware id of this larb.
+
 Example:
 	larb1: larb@16010000 {
 		compatible = "mediatek,mt8173-smi-larb";
@@ -25,3 +28,15 @@ Example:
 			 <&vdecsys CLK_VDEC_LARB_CKEN>;
 		clock-names = "apb", "smi";
 	};
+
+Example for mt2701:
+	larb0: larb@14010000 {
+		compatible = "mediatek,mt2701-smi-larb";
+		reg = <0 0x14010000 0 0x1000>;
+		mediatek,smi = <&smi_common>;
+		mediatek,larbid = <0>;
+		clocks = <&mmsys CLK_MM_SMI_LARB0>,
+			 <&mmsys CLK_MM_SMI_LARB0>;
+		clock-names = "apb", "smi";
+		power-domains = <&scpsys MT2701_POWER_DOMAIN_DISP>;
+	};
-- 
2.6.4

[PATCH 3/3] arm: dts: mediatek: add larbid property for larb

[honghui.zhang]

From: Honghui Zhang <honghui.zhang@mediatek.com>

Add mediatek's hardware id information for smi larb.

Signed-off-by: Honghui Zhang <honghui.zhang@mediatek.com>
---
 arch/arm/boot/dts/mt2701.dtsi | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/arm/boot/dts/mt2701.dtsi b/arch/arm/boot/dts/mt2701.dtsi
index f1efdc6..2cddbec 100644
--- a/arch/arm/boot/dts/mt2701.dtsi
+++ b/arch/arm/boot/dts/mt2701.dtsi
@@ -533,6 +533,7 @@
 		compatible = "mediatek,mt2701-smi-larb";
 		reg = <0 0x14010000 0 0x1000>;
 		mediatek,smi = <&smi_common>;
+		mediatek,larbid = <0>;
 		clocks = <&mmsys CLK_MM_SMI_LARB0>,
 			 <&mmsys CLK_MM_SMI_LARB0>;
 		clock-names = "apb", "smi";
@@ -549,6 +550,7 @@
 		compatible = "mediatek,mt2701-smi-larb";
 		reg = <0 0x15001000 0 0x1000>;
 		mediatek,smi = <&smi_common>;
+		mediatek,larbid = <2>;
 		clocks = <&imgsys CLK_IMG_SMI_COMM>,
 			 <&imgsys CLK_IMG_SMI_COMM>;
 		clock-names = "apb", "smi";
@@ -579,6 +581,7 @@
 		compatible = "mediatek,mt2701-smi-larb";
 		reg = <0 0x16010000 0 0x1000>;
 		mediatek,smi = <&smi_common>;
+		mediatek,larbid = <1>;
 		clocks = <&vdecsys CLK_VDEC_CKGEN>,
 			 <&vdecsys CLK_VDEC_LARB>;
 		clock-names = "apb", "smi";
-- 
2.6.4

Re: [PATCH 0/3] Add larbid init routine for mediatek's gen1 smi larb driver

[Joerg Roedel]

On Thu, Jul 27, 2017 at 10:01:09AM +0800, honghui.zhang@mediatek.com wrote:
> From: Honghui Zhang <honghui.zhang@mediatek.com>
> 
> Mediatek's gen1 smi need the hardware larbid to identify the offset for
> the register which controls whether enable iommu for this larb.
> In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
> the larbid was used without properly initialized. This patchset fixed
> that.
> 
> Honghui Zhang (3):
>   memory: mtk-smi: add larbid init routine
>   dt-bindings: mediatek: add descriptions for larbid
>   arm: dts: mediatek: add larbid property for larb

What are these patches based on? They don't apply to v4.13-rc2.



	Joerg

Re: [PATCH 0/3] Add larbid init routine for mediatek's gen1 smi larb driver

[Honghui Zhang]

On Thu, 2017-07-27 at 14:58 +0200, Joerg Roedel wrote:
> On Thu, Jul 27, 2017 at 10:01:09AM +0800, honghui.zhang@mediatek.com wrote:
> > From: Honghui Zhang <honghui.zhang@mediatek.com>
> > 
> > Mediatek's gen1 smi need the hardware larbid to identify the offset for
> > the register which controls whether enable iommu for this larb.
> > In the commit 3c8f4ad85c4b ("memory/mediatek: add support for mt2701"),
> > the larbid was used without properly initialized. This patchset fixed
> > that.
> > 
> > Honghui Zhang (3):
> >   memory: mtk-smi: add larbid init routine
> >   dt-bindings: mediatek: add descriptions for larbid
> >   arm: dts: mediatek: add larbid property for larb
> 
> What are these patches based on? They don't apply to v4.13-rc2.
> 
> 

Hi, Joerg,
	this patchset was base on the patch v2 version
memory: mtk-smi: Use of_device_get_match_data helper[1],
and it's base on 4.13-rc1

[1] https://patchwork.kernel.org/patch/9864719

thanks
> 
> 	Joerg
>