diff mbox series

[v5,2/7] dt-bindings: PCI: ti,am65: Extend for use with PVU

Message ID 33d08f61fe9bd692da0eceab91209832bf16e804.1725816753.git.jan.kiszka@siemens.com (mailing list archive)
State Superseded
Headers show
Series soc: ti: Add and use PVU on K3-AM65 for DMA isolation | expand

Commit Message

Jan Kiszka Sept. 8, 2024, 5:32 p.m. UTC
From: Jan Kiszka <jan.kiszka@siemens.com>

The PVU on the AM65 SoC is capable of restricting DMA from PCIe devices
to specific regions of host memory. Add the optional property
"memory-regions" to point to such regions of memory when PVU is used.

Since the PVU deals with system physical addresses, utilizing the PVU
with PCIe devices also requires setting up the VMAP registers to map the
Requester ID of the PCIe device to the CBA Virtual ID, which in turn is
mapped to the system physical address. Hence, describe the VMAP
registers which are optional unless the PVU shall be used for PCIe.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
CC: Lorenzo Pieralisi <lpieralisi@kernel.org>
CC: "Krzysztof Wilczyński" <kw@linux.com>
CC: Bjorn Helgaas <bhelgaas@google.com>
CC: linux-pci@vger.kernel.org
---
 .../bindings/pci/ti,am65-pci-host.yaml        | 29 +++++++++++++++++--
 1 file changed, 26 insertions(+), 3 deletions(-)

Comments

Krzysztof Kozlowski Sept. 9, 2024, 6:22 a.m. UTC | #1
On Sun, Sep 08, 2024 at 07:32:28PM +0200, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> The PVU on the AM65 SoC is capable of restricting DMA from PCIe devices
> to specific regions of host memory. Add the optional property
> "memory-regions" to point to such regions of memory when PVU is used.
> 
> Since the PVU deals with system physical addresses, utilizing the PVU
> with PCIe devices also requires setting up the VMAP registers to map the
> Requester ID of the PCIe device to the CBA Virtual ID, which in turn is
> mapped to the system physical address. Hence, describe the VMAP
> registers which are optional unless the PVU shall be used for PCIe.
> 
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
> ---
> CC: Lorenzo Pieralisi <lpieralisi@kernel.org>
> CC: "Krzysztof Wilczyński" <kw@linux.com>
> CC: Bjorn Helgaas <bhelgaas@google.com>
> CC: linux-pci@vger.kernel.org
> ---
>  .../bindings/pci/ti,am65-pci-host.yaml        | 29 +++++++++++++++++--
>  1 file changed, 26 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
> index 0a9d10532cc8..0c297d12173c 100644
> --- a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
> +++ b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
> @@ -20,14 +20,18 @@ properties:
>        - ti,keystone-pcie
>  
>    reg:
> -    maxItems: 4
> +    minItems: 4
> +    maxItems: 6
>  
>    reg-names:
> +    minItems: 4
>      items:
>        - const: app
>        - const: dbics
>        - const: config
>        - const: atu
> +      - const: vmap_lp
> +      - const: vmap_hp
>  
>    interrupts:
>      maxItems: 1
> @@ -83,13 +87,30 @@ if:
>      compatible:
>        enum:
>          - ti,am654-pcie-rc
> +
>  then:
> +  properties:
> +    memory-region:

I think I said it two times already. You must define properties in
top-level. That's how we expect, that's how dtschema works (even if it
works fine otherwise, it's not always that case), that's how almost all
bindings are written.

> +      maxItems: 1
> +      description: |
> +        phandle to a restricted DMA pool to be used for all devices behind
> +        this controller. The regions should be defined according to
> +        reserved-memory/shared-dma-pool.yaml.

Best regards,
Krzysztof
Jan Kiszka Sept. 9, 2024, 6:48 a.m. UTC | #2
On 09.09.24 08:22, Krzysztof Kozlowski wrote:
> On Sun, Sep 08, 2024 at 07:32:28PM +0200, Jan Kiszka wrote:
>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>
>> The PVU on the AM65 SoC is capable of restricting DMA from PCIe devices
>> to specific regions of host memory. Add the optional property
>> "memory-regions" to point to such regions of memory when PVU is used.
>>
>> Since the PVU deals with system physical addresses, utilizing the PVU
>> with PCIe devices also requires setting up the VMAP registers to map the
>> Requester ID of the PCIe device to the CBA Virtual ID, which in turn is
>> mapped to the system physical address. Hence, describe the VMAP
>> registers which are optional unless the PVU shall be used for PCIe.
>>
>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>> ---
>> CC: Lorenzo Pieralisi <lpieralisi@kernel.org>
>> CC: "Krzysztof Wilczyński" <kw@linux.com>
>> CC: Bjorn Helgaas <bhelgaas@google.com>
>> CC: linux-pci@vger.kernel.org
>> ---
>>  .../bindings/pci/ti,am65-pci-host.yaml        | 29 +++++++++++++++++--
>>  1 file changed, 26 insertions(+), 3 deletions(-)
>>
>> diff --git a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>> index 0a9d10532cc8..0c297d12173c 100644
>> --- a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>> +++ b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>> @@ -20,14 +20,18 @@ properties:
>>        - ti,keystone-pcie
>>  
>>    reg:
>> -    maxItems: 4
>> +    minItems: 4
>> +    maxItems: 6
>>  
>>    reg-names:
>> +    minItems: 4
>>      items:
>>        - const: app
>>        - const: dbics
>>        - const: config
>>        - const: atu
>> +      - const: vmap_lp
>> +      - const: vmap_hp
>>  
>>    interrupts:
>>      maxItems: 1
>> @@ -83,13 +87,30 @@ if:
>>      compatible:
>>        enum:
>>          - ti,am654-pcie-rc
>> +
>>  then:
>> +  properties:
>> +    memory-region:
> 
> I think I said it two times already. You must define properties in
> top-level. That's how we expect, that's how dtschema works (even if it
> works fine otherwise, it's not always that case), that's how almost all
> bindings are written.

Look, if you have such rules, also enhance the checker, or people like
me will continue to work intuitively. Add reasoning along that as well,
would help further to reduce your review effort. The current situation
with rather fuzzy results from the checker and strange mechanisms inside
(see my maxItems finding) is not very helpful IMHO.

I this concrete case, I would add this item top-level, just to set
maxItems to 0 for ti,keystone-pcie? Not a pattern I'm finding anywhere.
Or do we have to allow memory-regions for all compatibles now?

Sorry for all these iterations, but you should see from my questions and
actions where the problems in the concepts are.

Jan
Krzysztof Kozlowski Sept. 9, 2024, 7:49 a.m. UTC | #3
On 09/09/2024 08:48, Jan Kiszka wrote:
> On 09.09.24 08:22, Krzysztof Kozlowski wrote:
>> On Sun, Sep 08, 2024 at 07:32:28PM +0200, Jan Kiszka wrote:
>>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>>
>>> The PVU on the AM65 SoC is capable of restricting DMA from PCIe devices
>>> to specific regions of host memory. Add the optional property
>>> "memory-regions" to point to such regions of memory when PVU is used.
>>>
>>> Since the PVU deals with system physical addresses, utilizing the PVU
>>> with PCIe devices also requires setting up the VMAP registers to map the
>>> Requester ID of the PCIe device to the CBA Virtual ID, which in turn is
>>> mapped to the system physical address. Hence, describe the VMAP
>>> registers which are optional unless the PVU shall be used for PCIe.
>>>
>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>>> ---
>>> CC: Lorenzo Pieralisi <lpieralisi@kernel.org>
>>> CC: "Krzysztof Wilczyński" <kw@linux.com>
>>> CC: Bjorn Helgaas <bhelgaas@google.com>
>>> CC: linux-pci@vger.kernel.org
>>> ---
>>>  .../bindings/pci/ti,am65-pci-host.yaml        | 29 +++++++++++++++++--
>>>  1 file changed, 26 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>> index 0a9d10532cc8..0c297d12173c 100644
>>> --- a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>> +++ b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>> @@ -20,14 +20,18 @@ properties:
>>>        - ti,keystone-pcie
>>>  
>>>    reg:
>>> -    maxItems: 4
>>> +    minItems: 4
>>> +    maxItems: 6
>>>  
>>>    reg-names:
>>> +    minItems: 4
>>>      items:
>>>        - const: app
>>>        - const: dbics
>>>        - const: config
>>>        - const: atu
>>> +      - const: vmap_lp
>>> +      - const: vmap_hp
>>>  
>>>    interrupts:
>>>      maxItems: 1
>>> @@ -83,13 +87,30 @@ if:
>>>      compatible:
>>>        enum:
>>>          - ti,am654-pcie-rc
>>> +
>>>  then:
>>> +  properties:
>>> +    memory-region:
>>
>> I think I said it two times already. You must define properties in
>> top-level. That's how we expect, that's how dtschema works (even if it
>> works fine otherwise, it's not always that case), that's how almost all
>> bindings are written.
> 
> Look, if you have such rules, also enhance the checker, or people like
> me will continue to work intuitively. Add reasoning along that as well,

That would be ideal, but I also asked to do this twice. It does not
matter if dtschema  or me tells you this, if you do not implement it.

> would help further to reduce your review effort. The current situation
> with rather fuzzy results from the checker and strange mechanisms inside
> (see my maxItems finding) is not very helpful IMHO.
> 
> I this concrete case, I would add this item top-level, just to set
> maxItems to 0 for ti,keystone-pcie? Not a pattern I'm finding anywhere.
> Or do we have to allow memory-regions for all compatibles now?

Is it really not suitable for all the compatibles? Maybe these are quite
different devices in such case?

But if it is not really suitable, then you can disallow it for other
variants with :false. This is also explicitly documented in example-schema:
https://elixir.bootlin.com/linux/v5.19/source/Documentation/devicetree/bindings/example-schema.yaml#L212

> 
> Sorry for all these iterations, but you should see from my questions and
> actions where the problems in the concepts are.

Best regards,
Krzysztof
Jan Kiszka Sept. 9, 2024, 12:23 p.m. UTC | #4
On 09.09.24 09:49, Krzysztof Kozlowski wrote:
> On 09/09/2024 08:48, Jan Kiszka wrote:
>> On 09.09.24 08:22, Krzysztof Kozlowski wrote:
>>> On Sun, Sep 08, 2024 at 07:32:28PM +0200, Jan Kiszka wrote:
>>>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>>>
>>>> The PVU on the AM65 SoC is capable of restricting DMA from PCIe devices
>>>> to specific regions of host memory. Add the optional property
>>>> "memory-regions" to point to such regions of memory when PVU is used.
>>>>
>>>> Since the PVU deals with system physical addresses, utilizing the PVU
>>>> with PCIe devices also requires setting up the VMAP registers to map the
>>>> Requester ID of the PCIe device to the CBA Virtual ID, which in turn is
>>>> mapped to the system physical address. Hence, describe the VMAP
>>>> registers which are optional unless the PVU shall be used for PCIe.
>>>>
>>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>>>> ---
>>>> CC: Lorenzo Pieralisi <lpieralisi@kernel.org>
>>>> CC: "Krzysztof Wilczyński" <kw@linux.com>
>>>> CC: Bjorn Helgaas <bhelgaas@google.com>
>>>> CC: linux-pci@vger.kernel.org
>>>> ---
>>>>  .../bindings/pci/ti,am65-pci-host.yaml        | 29 +++++++++++++++++--
>>>>  1 file changed, 26 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>>> index 0a9d10532cc8..0c297d12173c 100644
>>>> --- a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>>> +++ b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
>>>> @@ -20,14 +20,18 @@ properties:
>>>>        - ti,keystone-pcie
>>>>  
>>>>    reg:
>>>> -    maxItems: 4
>>>> +    minItems: 4
>>>> +    maxItems: 6
>>>>  
>>>>    reg-names:
>>>> +    minItems: 4
>>>>      items:
>>>>        - const: app
>>>>        - const: dbics
>>>>        - const: config
>>>>        - const: atu
>>>> +      - const: vmap_lp
>>>> +      - const: vmap_hp
>>>>  
>>>>    interrupts:
>>>>      maxItems: 1
>>>> @@ -83,13 +87,30 @@ if:
>>>>      compatible:
>>>>        enum:
>>>>          - ti,am654-pcie-rc
>>>> +
>>>>  then:
>>>> +  properties:
>>>> +    memory-region:
>>>
>>> I think I said it two times already. You must define properties in
>>> top-level. That's how we expect, that's how dtschema works (even if it
>>> works fine otherwise, it's not always that case), that's how almost all
>>> bindings are written.
>>
>> Look, if you have such rules, also enhance the checker, or people like
>> me will continue to work intuitively. Add reasoning along that as well,
> 
> That would be ideal, but I also asked to do this twice. It does not
> matter if dtschema  or me tells you this, if you do not implement it.
> 
>> would help further to reduce your review effort. The current situation
>> with rather fuzzy results from the checker and strange mechanisms inside
>> (see my maxItems finding) is not very helpful IMHO.
>>
>> I this concrete case, I would add this item top-level, just to set
>> maxItems to 0 for ti,keystone-pcie? Not a pattern I'm finding anywhere.
>> Or do we have to allow memory-regions for all compatibles now?
> 
> Is it really not suitable for all the compatibles? Maybe these are quite
> different devices in such case?
> 
> But if it is not really suitable, then you can disallow it for other
> variants with :false. This is also explicitly documented in example-schema:
> https://elixir.bootlin.com/linux/v5.19/source/Documentation/devicetree/bindings/example-schema.yaml#L212
> 

I will address this via documentation: The property is not hardware
related but permits swiotlb. It can be applied to devices as well that
have no hardware enforcement, unlike the am654.

Jan
diff mbox series

Patch

diff --git a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
index 0a9d10532cc8..0c297d12173c 100644
--- a/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
+++ b/Documentation/devicetree/bindings/pci/ti,am65-pci-host.yaml
@@ -20,14 +20,18 @@  properties:
       - ti,keystone-pcie
 
   reg:
-    maxItems: 4
+    minItems: 4
+    maxItems: 6
 
   reg-names:
+    minItems: 4
     items:
       - const: app
       - const: dbics
       - const: config
       - const: atu
+      - const: vmap_lp
+      - const: vmap_hp
 
   interrupts:
     maxItems: 1
@@ -83,13 +87,30 @@  if:
     compatible:
       enum:
         - ti,am654-pcie-rc
+
 then:
+  properties:
+    memory-region:
+      maxItems: 1
+      description: |
+        phandle to a restricted DMA pool to be used for all devices behind
+        this controller. The regions should be defined according to
+        reserved-memory/shared-dma-pool.yaml.
+
   required:
     - dma-coherent
     - power-domains
     - msi-map
     - num-viewport
 
+else:
+  properties:
+    reg:
+      maxItems: 4
+
+    reg-names:
+      maxItems: 4
+
 unevaluatedProperties: false
 
 examples:
@@ -104,8 +125,10 @@  examples:
         reg =  <0x5500000 0x1000>,
                <0x5501000 0x1000>,
                <0x10000000 0x2000>,
-               <0x5506000 0x1000>;
-        reg-names = "app", "dbics", "config", "atu";
+               <0x5506000 0x1000>,
+               <0x2900000 0x1000>,
+               <0x2908000 0x1000>;
+        reg-names = "app", "dbics", "config", "atu", "vmap_lp", "vmap_hp";
         power-domains = <&k3_pds 120 TI_SCI_PD_EXCLUSIVE>;
         #address-cells = <3>;
         #size-cells = <2>;