[PATCH v2 3/7] dt-bindings: PCI: qcom: Add IPQ9574 PCIe controller

[PATCH v2 3/7] dt-bindings: PCI: qcom: Add IPQ9574 PCIe controller

[Alexandru Gagniuc]

IPQ9574 has PCIe controllers which are almost identical to IPQ6018.
The only difference is that the "iface" clock is not required.
Document this difference along with the compatible string.

Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
---
 .../devicetree/bindings/pci/qcom,pcie.yaml    | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)

diff --git a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
index cf9a6910b542..1915bea580d3 100644
--- a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
+++ b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
@@ -26,6 +26,7 @@ properties:
           - qcom,pcie-ipq8064-v2
           - qcom,pcie-ipq8074
           - qcom,pcie-ipq8074-gen3
+          - qcom,pcie-ipq9574
           - qcom,pcie-msm8996
           - qcom,pcie-qcs404
           - qcom,pcie-sdm845
@@ -397,6 +398,37 @@ allOf:
             - const: axi_m_sticky # AXI Master Sticky reset
             - const: axi_s_sticky # AXI Slave Sticky reset
 
+  - if:
+      properties:
+        compatible:
+          contains:
+            enum:
+              - qcom,pcie-ipq9574
+    then:
+      properties:
+        clocks:
+          minItems: 4
+          maxItems: 4
+        clock-names:
+          items:
+            - const: axi_m # AXI Master clock
+            - const: axi_s # AXI Slave clock
+            - const: axi_bridge # AXI bridge clock
+            - const: rchng
+        resets:
+          minItems: 8
+          maxItems: 8
+        reset-names:
+          items:
+            - const: pipe # PIPE reset
+            - const: sleep # Sleep reset
+            - const: sticky # Core Sticky reset
+            - const: axi_m # AXI Master reset
+            - const: axi_s # AXI Slave reset
+            - const: ahb # AHB Reset
+            - const: axi_m_sticky # AXI Master Sticky reset
+            - const: axi_s_sticky # AXI Slave Sticky reset
+
   - if:
       properties:
         compatible:
@@ -507,6 +539,7 @@ allOf:
                 - qcom,pcie-ipq8064v2
                 - qcom,pcie-ipq8074
                 - qcom,pcie-ipq8074-gen3
+                - qcom,pcie-ipq9574
                 - qcom,pcie-qcs404
     then:
       required:
@@ -566,6 +599,7 @@ allOf:
               - qcom,pcie-ipq8064-v2
               - qcom,pcie-ipq8074
               - qcom,pcie-ipq8074-gen3
+              - qcom,pcie-ipq9574
               - qcom,pcie-qcs404
     then:
       properties:
-- 
2.40.1

[PATCH v2 4/7] PCI: qcom: Add support for IPQ9574

[Alexandru Gagniuc]

Add support for the PCIe on IPQ9574. The main difference from ipq6018
is that the "iface" clock is not necessarry. Add a special case in
qcom_pcie_get_resources_2_9_0() to handle this.

Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
---
 drivers/pci/controller/dwc/pcie-qcom.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
index 14772edcf0d3..10560d6d6336 100644
--- a/drivers/pci/controller/dwc/pcie-qcom.c
+++ b/drivers/pci/controller/dwc/pcie-qcom.c
@@ -1101,15 +1101,19 @@ static int qcom_pcie_get_resources_2_9_0(struct qcom_pcie *pcie)
 	struct qcom_pcie_resources_2_9_0 *res = &pcie->res.v2_9_0;
 	struct dw_pcie *pci = pcie->pci;
 	struct device *dev = pci->dev;
-	int ret;
+	int ret, num_clks = ARRAY_SIZE(res->clks) - 1;
 
-	res->clks[0].id = "iface";
+	res->clks[0].id = "rchng";
 	res->clks[1].id = "axi_m";
 	res->clks[2].id = "axi_s";
 	res->clks[3].id = "axi_bridge";
-	res->clks[4].id = "rchng";
 
-	ret = devm_clk_bulk_get(dev, ARRAY_SIZE(res->clks), res->clks);
+	if (!of_device_is_compatible(dev->of_node, "qcom,pcie-ipq9574")) {
+		res->clks[4].id = "iface";
+		num_clks++;
+	}
+
+	ret = devm_clk_bulk_get(dev, num_clks, res->clks);
 	if (ret < 0)
 		return ret;
 
@@ -1664,6 +1668,7 @@ static const struct of_device_id qcom_pcie_match[] = {
 	{ .compatible = "qcom,pcie-ipq8064-v2", .data = &cfg_2_1_0 },
 	{ .compatible = "qcom,pcie-ipq8074", .data = &cfg_2_3_3 },
 	{ .compatible = "qcom,pcie-ipq8074-gen3", .data = &cfg_2_9_0 },
+	{ .compatible = "qcom,pcie-ipq9574", .data = &cfg_2_9_0 },
 	{ .compatible = "qcom,pcie-msm8996", .data = &cfg_2_3_2 },
 	{ .compatible = "qcom,pcie-qcs404", .data = &cfg_2_4_0 },
 	{ .compatible = "qcom,pcie-sa8540p", .data = &cfg_sc8280xp },
-- 
2.40.1

Re: [PATCH v2 3/7] dt-bindings: PCI: qcom: Add IPQ9574 PCIe controller

[Krzysztof Kozlowski]

On 09/04/2024 21:08, Alexandru Gagniuc wrote:
> IPQ9574 has PCIe controllers which are almost identical to IPQ6018.
> The only difference is that the "iface" clock is not required.
> Document this difference along with the compatible string.
> 
> Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
> ---
>  .../devicetree/bindings/pci/qcom,pcie.yaml    | 34 +++++++++++++++++++
>  1 file changed, 34 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
> index cf9a6910b542..1915bea580d3 100644
> --- a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
> +++ b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
> @@ -26,6 +26,7 @@ properties:
>            - qcom,pcie-ipq8064-v2
>            - qcom,pcie-ipq8074
>            - qcom,pcie-ipq8074-gen3
> +          - qcom,pcie-ipq9574
>            - qcom,pcie-msm8996
>            - qcom,pcie-qcs404
>            - qcom,pcie-sdm845
> @@ -397,6 +398,37 @@ allOf:
>              - const: axi_m_sticky # AXI Master Sticky reset
>              - const: axi_s_sticky # AXI Slave Sticky reset
>  

Where do you constrain the reg?

Best regards,
Krzysztof

Re: [PATCH v2 4/7] PCI: qcom: Add support for IPQ9574

[Konrad Dybcio]

On 4/9/24 21:08, Alexandru Gagniuc wrote:
> Add support for the PCIe on IPQ9574. The main difference from ipq6018
> is that the "iface" clock is not necessarry. Add a special case in
> qcom_pcie_get_resources_2_9_0() to handle this.
> 
> Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
> Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
> ---
>   drivers/pci/controller/dwc/pcie-qcom.c | 13 +++++++++----
>   1 file changed, 9 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> index 14772edcf0d3..10560d6d6336 100644
> --- a/drivers/pci/controller/dwc/pcie-qcom.c
> +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> @@ -1101,15 +1101,19 @@ static int qcom_pcie_get_resources_2_9_0(struct qcom_pcie *pcie)
>   	struct qcom_pcie_resources_2_9_0 *res = &pcie->res.v2_9_0;
>   	struct dw_pcie *pci = pcie->pci;
>   	struct device *dev = pci->dev;
> -	int ret;
> +	int ret, num_clks = ARRAY_SIZE(res->clks) - 1;
>   
> -	res->clks[0].id = "iface";
> +	res->clks[0].id = "rchng";
>   	res->clks[1].id = "axi_m";
>   	res->clks[2].id = "axi_s";
>   	res->clks[3].id = "axi_bridge";
> -	res->clks[4].id = "rchng";
>   
> -	ret = devm_clk_bulk_get(dev, ARRAY_SIZE(res->clks), res->clks);
> +	if (!of_device_is_compatible(dev->of_node, "qcom,pcie-ipq9574")) {
> +		res->clks[4].id = "iface";
> +		num_clks++;

Or use devm_clk_bulk_get_optional and rely on the bindings to sanity-check.

Mani, thoughts?

Konrad

Re: [PATCH v2 3/7] dt-bindings: PCI: qcom: Add IPQ9574 PCIe controller

[mr.nuke.me]

On 4/9/24 15:08, Krzysztof Kozlowski wrote:
> On 09/04/2024 21:08, Alexandru Gagniuc wrote:
>> IPQ9574 has PCIe controllers which are almost identical to IPQ6018.
>> The only difference is that the "iface" clock is not required.
>> Document this difference along with the compatible string.
>>
>> Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
>> ---
>>   .../devicetree/bindings/pci/qcom,pcie.yaml    | 34 +++++++++++++++++++
>>   1 file changed, 34 insertions(+)
>>
>> diff --git a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>> index cf9a6910b542..1915bea580d3 100644
>> --- a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>> +++ b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>> @@ -26,6 +26,7 @@ properties:
>>             - qcom,pcie-ipq8064-v2
>>             - qcom,pcie-ipq8074
>>             - qcom,pcie-ipq8074-gen3
>> +          - qcom,pcie-ipq9574
>>             - qcom,pcie-msm8996
>>             - qcom,pcie-qcs404
>>             - qcom,pcie-sdm845
>> @@ -397,6 +398,37 @@ allOf:
>>               - const: axi_m_sticky # AXI Master Sticky reset
>>               - const: axi_s_sticky # AXI Slave Sticky reset
>>   
> 
> Where do you constrain the reg?

I didn't realize that was also required -- the make checks should have 
picked this up too? I might be invoking the tests incorrectly.

I should add the ipq9574 in the same list as ipq8074-gen3 and ipq6018, 
correct?

Alex

Re: [PATCH v2 3/7] dt-bindings: PCI: qcom: Add IPQ9574 PCIe controller

[Krzysztof Kozlowski]

On 11/04/2024 20:00, mr.nuke.me@gmail.com wrote:
> 
> 
> On 4/9/24 15:08, Krzysztof Kozlowski wrote:
>> On 09/04/2024 21:08, Alexandru Gagniuc wrote:
>>> IPQ9574 has PCIe controllers which are almost identical to IPQ6018.
>>> The only difference is that the "iface" clock is not required.
>>> Document this difference along with the compatible string.
>>>
>>> Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
>>> ---
>>>   .../devicetree/bindings/pci/qcom,pcie.yaml    | 34 +++++++++++++++++++
>>>   1 file changed, 34 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>>> index cf9a6910b542..1915bea580d3 100644
>>> --- a/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>>> +++ b/Documentation/devicetree/bindings/pci/qcom,pcie.yaml
>>> @@ -26,6 +26,7 @@ properties:
>>>             - qcom,pcie-ipq8064-v2
>>>             - qcom,pcie-ipq8074
>>>             - qcom,pcie-ipq8074-gen3
>>> +          - qcom,pcie-ipq9574
>>>             - qcom,pcie-msm8996
>>>             - qcom,pcie-qcs404
>>>             - qcom,pcie-sdm845
>>> @@ -397,6 +398,37 @@ allOf:
>>>               - const: axi_m_sticky # AXI Master Sticky reset
>>>               - const: axi_s_sticky # AXI Slave Sticky reset
>>>   
>>
>> Where do you constrain the reg?
> 
> I didn't realize that was also required -- the make checks should have 
> picked this up too? I might be invoking the tests incorrectly.
> 
> I should add the ipq9574 in the same list as ipq8074-gen3 and ipq6018, 
> correct?

If you add new variant, look at existing compatibles where they appear.
If there is a if: constraining compatibles, then it's a hint you should
do the same for your device. So yes, you must constrain all properties
which are made flexible in top-level properties.

Best regards,
Krzysztof

Re: [PATCH v2 4/7] PCI: qcom: Add support for IPQ9574

[Manivannan Sadhasivam]

On Wed, Apr 10, 2024 at 01:50:26PM +0200, Konrad Dybcio wrote:
> 
> 
> On 4/9/24 21:08, Alexandru Gagniuc wrote:
> > Add support for the PCIe on IPQ9574. The main difference from ipq6018
> > is that the "iface" clock is not necessarry. Add a special case in
> > qcom_pcie_get_resources_2_9_0() to handle this.
> > 
> > Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
> > Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
> > ---
> >   drivers/pci/controller/dwc/pcie-qcom.c | 13 +++++++++----
> >   1 file changed, 9 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> > index 14772edcf0d3..10560d6d6336 100644
> > --- a/drivers/pci/controller/dwc/pcie-qcom.c
> > +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> > @@ -1101,15 +1101,19 @@ static int qcom_pcie_get_resources_2_9_0(struct qcom_pcie *pcie)
> >   	struct qcom_pcie_resources_2_9_0 *res = &pcie->res.v2_9_0;
> >   	struct dw_pcie *pci = pcie->pci;
> >   	struct device *dev = pci->dev;
> > -	int ret;
> > +	int ret, num_clks = ARRAY_SIZE(res->clks) - 1;
> > -	res->clks[0].id = "iface";
> > +	res->clks[0].id = "rchng";
> >   	res->clks[1].id = "axi_m";
> >   	res->clks[2].id = "axi_s";
> >   	res->clks[3].id = "axi_bridge";
> > -	res->clks[4].id = "rchng";
> > -	ret = devm_clk_bulk_get(dev, ARRAY_SIZE(res->clks), res->clks);
> > +	if (!of_device_is_compatible(dev->of_node, "qcom,pcie-ipq9574")) {
> > +		res->clks[4].id = "iface";
> > +		num_clks++;
> 
> Or use devm_clk_bulk_get_optional and rely on the bindings to sanity-check.
> 
> Mani, thoughts?
> 

I'd prefer to use devm_clk_bulk_get_all() and just rely on DT schema to do the
validation. There was a patch hanging in my branch for some time and I sent it
now: https://lore.kernel.org/linux-pci/20240417-pci-qcom-clk-bulk-v1-1-52ca19b3d6b2@linaro.org/

- Mani

-- 
மணிவண்ணன் சதாசிவம்