[PATCH V6 0/4] Add device tree support for NVIDIA Tegra CMDQV

[PATCH V6 0/4] Add device tree support for NVIDIA Tegra CMDQV

[Ashish Mhetre]

This series adds device tree support for the CMDQ-Virtualization (CMDQV)
hardware on NVIDIA Tegra264 SoCs.

CMDQV is a hardware block that works alongside the ARM SMMUv3 to assist in
virtualizing the command queue. It was previously only supported through
ACPI on Tegra241. This series extends the existing driver to support device
tree based initialization, which is required for Tegra264 platforms.

The series is structured as follows:

Patch 1: Decouple tegra241-cmdqv driver from ACPI by replacing all the
         ACPI functions with standard platform functions.

Patch 2: Extends the tegra241-cmdqv driver to support device tree probing
         alongside the existing ACPI support. The SMMU driver now parses
         the nvidia,cmdqv phandle to associate each SMMU with its
         corresponding CMDQV instance.

Patch 3: Adds device tree binding documentation for nvidia,tegra264-cmdqv
         and extends the arm,smmu-v3 binding with an optional nvidia,cmdqv
         property.

Patch 4: Adds CMDQV device nodes to the Tegra264 device tree and enables
         them on the tegra264-p3834 platform.

The implementation follows the existing ACPI probe path to minimize code
divergence and maintain consistency with Tegra241 support.

Changes in V6:
- Added Nicolin's patch for decoupling tegra241-cmdqv from ACPI and
  rebased other patches on top of it.
- Removed put_device() from DT path as tegra241-cmdqv driver takes care
  of dropping reference in remove callback.

Changes in V5:
- Guard ACPI functions with CONFIG_ACPI to fix build without ACPI.

Changes in V4:
- Removed platform driver registration from tegra241-cmdqv.c to avoid
  duplicate module init/cleanup symbols
- Fixed the ordering of properties in CMDQV device tree nodes
- Added "Reviewed-by: Rob Herring" for binding documentation patch(2/3)
  carry-forwarding from V3.

Changes in V3:
- Remove the Kconfig dependency for CONFIG_TEGRA241_CMDQV
- Drop the reference on the platform device after getting it from
  of_find_device_by_node()
- Remove the unnecessary "nvidia,tegra264-smmu" compatible string from
  arm-smmu-v3 match table
- Order CMDQV nodes in device tree files according to its address

Changes in V2:
- Updated dependency for CONFIG_TEGRA241_CMDQV on OF || ACPI
- Changed maintainer to Nicolin Chen
- Removed interrupt-names property

Ashish Mhetre (3):
  iommu/arm-smmu-v3: Add device-tree support for CMDQV driver
  dt-bindings: iommu: Add NVIDIA Tegra CMDQV support
  arm64: dts: nvidia: Add nodes for CMDQV

Nicolin Chen (1):
  iommu/tegra241-cmdqv: Decouple driver from ACPI

 .../bindings/iommu/arm,smmu-v3.yaml           | 30 ++++++-
 .../bindings/iommu/nvidia,tegra264-cmdqv.yaml | 42 ++++++++++
 .../arm64/boot/dts/nvidia/tegra264-p3834.dtsi |  8 ++
 arch/arm64/boot/dts/nvidia/tegra264.dtsi      | 50 +++++++++--
 drivers/iommu/arm/Kconfig                     |  1 -
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   | 34 +++++++-
 .../iommu/arm/arm-smmu-v3/tegra241-cmdqv.c    | 84 +++----------------
 7 files changed, 168 insertions(+), 81 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/iommu/nvidia,tegra264-cmdqv.yaml

-- 
2.25.1

[PATCH V6 1/4] iommu/tegra241-cmdqv: Decouple driver from ACPI

[Ashish Mhetre]

From: Nicolin Chen <nicolinc@nvidia.com>

A platform device is created by acpi_create_platform_device() per CMDQV's
adev. That means there is no point in going through _CRS of ACPI.

Replace all the ACPI functions with standard platform functions. And drop
all ACPI dependencies. This will make the driver compatible with DT also.

Suggested-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 drivers/iommu/arm/Kconfig                     |  1 -
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   |  2 +-
 .../iommu/arm/arm-smmu-v3/tegra241-cmdqv.c    | 84 +++----------------
 3 files changed, 13 insertions(+), 74 deletions(-)

diff --git a/drivers/iommu/arm/Kconfig b/drivers/iommu/arm/Kconfig
index ef42bbe07dbe..5fac08b89dee 100644
--- a/drivers/iommu/arm/Kconfig
+++ b/drivers/iommu/arm/Kconfig
@@ -121,7 +121,6 @@ config ARM_SMMU_V3_KUNIT_TEST
 
 config TEGRA241_CMDQV
 	bool "NVIDIA Tegra241 CMDQ-V extension support for ARM SMMUv3"
-	depends on ACPI
 	help
 	  Support for NVIDIA CMDQ-Virtualization extension for ARM SMMUv3. The
 	  CMDQ-V extension is similar to v3.3 ECMDQ for multi command queues
diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index d16d35c78c06..42de3da54858 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -4542,7 +4542,7 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
 	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
 	if (adev) {
 		/* Tegra241 CMDQV driver is responsible for put_device() */
-		smmu->impl_dev = &adev->dev;
+		smmu->impl_dev = acpi_get_first_physical_node(adev);
 		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
 		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
 			 dev_name(smmu->impl_dev));
diff --git a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
index 378104cd395e..1fc03b72beb8 100644
--- a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
+++ b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
@@ -3,17 +3,15 @@
 
 #define dev_fmt(fmt) "tegra241_cmdqv: " fmt
 
-#include <linux/acpi.h>
 #include <linux/debugfs.h>
 #include <linux/dma-mapping.h>
 #include <linux/interrupt.h>
 #include <linux/iommu.h>
 #include <linux/iommufd.h>
 #include <linux/iopoll.h>
+#include <linux/platform_device.h>
 #include <uapi/linux/iommufd.h>
 
-#include <acpi/acpixf.h>
-
 #include "arm-smmu-v3.h"
 
 /* CMDQV register page base and size defines */
@@ -854,69 +852,6 @@ static struct arm_smmu_impl_ops tegra241_cmdqv_impl_ops = {
 
 /* Probe Functions */
 
-static int tegra241_cmdqv_acpi_is_memory(struct acpi_resource *res, void *data)
-{
-	struct resource_win win;
-
-	return !acpi_dev_resource_address_space(res, &win);
-}
-
-static int tegra241_cmdqv_acpi_get_irqs(struct acpi_resource *ares, void *data)
-{
-	struct resource r;
-	int *irq = data;
-
-	if (*irq <= 0 && acpi_dev_resource_interrupt(ares, 0, &r))
-		*irq = r.start;
-	return 1; /* No need to add resource to the list */
-}
-
-static struct resource *
-tegra241_cmdqv_find_acpi_resource(struct device *dev, int *irq)
-{
-	struct acpi_device *adev = to_acpi_device(dev);
-	struct list_head resource_list;
-	struct resource_entry *rentry;
-	struct resource *res = NULL;
-	int ret;
-
-	INIT_LIST_HEAD(&resource_list);
-	ret = acpi_dev_get_resources(adev, &resource_list,
-				     tegra241_cmdqv_acpi_is_memory, NULL);
-	if (ret < 0) {
-		dev_err(dev, "failed to get memory resource: %d\n", ret);
-		return NULL;
-	}
-
-	rentry = list_first_entry_or_null(&resource_list,
-					  struct resource_entry, node);
-	if (!rentry) {
-		dev_err(dev, "failed to get memory resource entry\n");
-		goto free_list;
-	}
-
-	/* Caller must free the res */
-	res = kzalloc(sizeof(*res), GFP_KERNEL);
-	if (!res)
-		goto free_list;
-
-	*res = *rentry->res;
-
-	acpi_dev_free_resource_list(&resource_list);
-
-	INIT_LIST_HEAD(&resource_list);
-
-	if (irq)
-		ret = acpi_dev_get_resources(adev, &resource_list,
-					     tegra241_cmdqv_acpi_get_irqs, irq);
-	if (ret < 0 || !irq || *irq <= 0)
-		dev_warn(dev, "no interrupt. errors will not be reported\n");
-
-free_list:
-	acpi_dev_free_resource_list(&resource_list);
-	return res;
-}
-
 static int tegra241_cmdqv_init_structures(struct arm_smmu_device *smmu)
 {
 	struct tegra241_cmdqv *cmdqv =
@@ -1042,18 +977,23 @@ __tegra241_cmdqv_probe(struct arm_smmu_device *smmu, struct resource *res,
 
 struct arm_smmu_device *tegra241_cmdqv_probe(struct arm_smmu_device *smmu)
 {
+	struct platform_device *pdev = to_platform_device(smmu->impl_dev);
 	struct arm_smmu_device *new_smmu;
-	struct resource *res = NULL;
+	struct resource *res;
 	int irq;
 
-	if (!smmu->dev->of_node)
-		res = tegra241_cmdqv_find_acpi_resource(smmu->impl_dev, &irq);
-	if (!res)
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res) {
+		dev_err(&pdev->dev, "no memory resource found for CMDQV\n");
 		goto out_fallback;
+	}
 
-	new_smmu = __tegra241_cmdqv_probe(smmu, res, irq);
-	kfree(res);
+	irq = platform_get_irq_optional(pdev, 0);
+	if (irq <= 0)
+		dev_warn(&pdev->dev,
+			 "no interrupt. errors will not be reported\n");
 
+	new_smmu = __tegra241_cmdqv_probe(smmu, res, irq);
 	if (new_smmu)
 		return new_smmu;
 
-- 
2.25.1

[PATCH V6 2/4] iommu/arm-smmu-v3: Add device-tree support for CMDQV driver

[Ashish Mhetre]

Add device tree support to the CMDQV driver to enable usage on Tegra264
SoCs. The implementation parses the nvidia,cmdqv phandle from the SMMU
device tree node to associate each SMMU with its corresponding CMDQV
instance based on compatible string.

Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
---
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 32 +++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index 42de3da54858..fbd16b210e88 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -4530,6 +4530,35 @@ static int arm_smmu_device_hw_probe(struct arm_smmu_device *smmu)
 	return 0;
 }
 
+#ifdef CONFIG_TEGRA241_CMDQV
+static void tegra_cmdqv_dt_probe(struct device_node *smmu_node,
+				 struct arm_smmu_device *smmu)
+{
+	struct platform_device *pdev;
+	struct device_node *np;
+
+	np = of_parse_phandle(smmu_node, "nvidia,cmdqv", 0);
+	if (!np)
+		return;
+
+	/* Tegra241 CMDQV driver is responsible for put_device() */
+	pdev = of_find_device_by_node(np);
+	of_node_put(np);
+	if (!pdev)
+		return;
+
+	smmu->impl_dev = &pdev->dev;
+	smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
+	dev_info(smmu->dev, "found companion CMDQV device: %s\n",
+		 dev_name(smmu->impl_dev));
+}
+#else
+static void tegra_cmdqv_dt_probe(struct device_node *smmu_node,
+				 struct arm_smmu_device *smmu)
+{
+}
+#endif
+
 #ifdef CONFIG_ACPI
 #ifdef CONFIG_TEGRA241_CMDQV
 static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
@@ -4634,6 +4663,9 @@ static int arm_smmu_device_dt_probe(struct platform_device *pdev,
 	if (of_dma_is_coherent(dev->of_node))
 		smmu->features |= ARM_SMMU_FEAT_COHERENCY;
 
+	if (of_device_is_compatible(dev->of_node, "nvidia,tegra264-smmu"))
+		tegra_cmdqv_dt_probe(dev->of_node, smmu);
+
 	return ret;
 }
 
-- 
2.25.1

[PATCH V6 3/4] dt-bindings: iommu: Add NVIDIA Tegra CMDQV support

[Ashish Mhetre]

The Command Queue Virtualization (CMDQV) hardware is part of the
SMMUv3 implementation on NVIDIA Tegra SoCs. It assists in
virtualizing the command queue for the SMMU.

Add a new device tree binding document for nvidia,tegra264-cmdqv.

Also update the arm,smmu-v3 binding to include an optional nvidia,cmdqv
property. This property is a phandle to the CMDQV device node, allowing
the SMMU driver to associate with its corresponding CMDQV instance.
Restrict this property usage to Nvidia Tegra264 only.

Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
---
 .../bindings/iommu/arm,smmu-v3.yaml           | 30 ++++++++++++-
 .../bindings/iommu/nvidia,tegra264-cmdqv.yaml | 42 +++++++++++++++++++
 2 files changed, 70 insertions(+), 2 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/iommu/nvidia,tegra264-cmdqv.yaml

diff --git a/Documentation/devicetree/bindings/iommu/arm,smmu-v3.yaml b/Documentation/devicetree/bindings/iommu/arm,smmu-v3.yaml
index 75fcf4cb52d9..1c03482e4c61 100644
--- a/Documentation/devicetree/bindings/iommu/arm,smmu-v3.yaml
+++ b/Documentation/devicetree/bindings/iommu/arm,smmu-v3.yaml
@@ -20,7 +20,12 @@ properties:
   $nodename:
     pattern: "^iommu@[0-9a-f]*"
   compatible:
-    const: arm,smmu-v3
+    oneOf:
+      - const: arm,smmu-v3
+      - items:
+          - enum:
+              - nvidia,tegra264-smmu
+          - const: arm,smmu-v3
 
   reg:
     maxItems: 1
@@ -58,6 +63,15 @@ properties:
 
   msi-parent: true
 
+  nvidia,cmdqv:
+    description: |
+      A phandle to its pairing CMDQV extension for an implementation on NVIDIA
+      Tegra SoC.
+
+      If this property is absent, CMDQ-Virtualization won't be used and SMMU
+      will only use its own CMDQ.
+    $ref: /schemas/types.yaml#/definitions/phandle
+
   hisilicon,broken-prefetch-cmd:
     type: boolean
     description: Avoid sending CMD_PREFETCH_* commands to the SMMU.
@@ -69,6 +83,17 @@ properties:
       register access with page 0 offsets. Set for Cavium ThunderX2 silicon that
       doesn't support SMMU page1 register space.
 
+allOf:
+  - if:
+      not:
+        properties:
+          compatible:
+            contains:
+              const: nvidia,tegra264-smmu
+    then:
+      properties:
+        nvidia,cmdqv: false
+
 required:
   - compatible
   - reg
@@ -82,7 +107,7 @@ examples:
     #include <dt-bindings/interrupt-controller/irq.h>
 
     iommu@2b400000 {
-            compatible = "arm,smmu-v3";
+            compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
             reg = <0x2b400000 0x20000>;
             interrupts = <GIC_SPI 74 IRQ_TYPE_EDGE_RISING>,
                          <GIC_SPI 75 IRQ_TYPE_EDGE_RISING>,
@@ -92,4 +117,5 @@ examples:
             dma-coherent;
             #iommu-cells = <1>;
             msi-parent = <&its 0xff0000>;
+            nvidia,cmdqv = <&cmdqv>;
     };
diff --git a/Documentation/devicetree/bindings/iommu/nvidia,tegra264-cmdqv.yaml b/Documentation/devicetree/bindings/iommu/nvidia,tegra264-cmdqv.yaml
new file mode 100644
index 000000000000..3f5006a59805
--- /dev/null
+++ b/Documentation/devicetree/bindings/iommu/nvidia,tegra264-cmdqv.yaml
@@ -0,0 +1,42 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/iommu/nvidia,tegra264-cmdqv.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: NVIDIA Tegra264 CMDQV
+
+description:
+  The CMDQ-Virtualization hardware block is part of the SMMUv3 implementation
+  on Tegra264 SoCs. It assists in virtualizing the command queue for the SMMU.
+
+maintainers:
+  - Nicolin Chen <nicolinc@nvidia.com>
+
+properties:
+  compatible:
+    const: nvidia,tegra264-cmdqv
+
+  reg:
+    maxItems: 1
+
+  interrupts:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+  - interrupts
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/interrupt-controller/arm-gic.h>
+    #include <dt-bindings/interrupt-controller/irq.h>
+
+    cmdqv@5200000 {
+            compatible = "nvidia,tegra264-cmdqv";
+            reg = <0x5200000 0x830000>;
+            interrupts = <GIC_SPI 19 IRQ_TYPE_LEVEL_HIGH>;
+    };
-- 
2.25.1

[PATCH V6 4/4] arm64: dts: nvidia: Add nodes for CMDQV

[Ashish Mhetre]

The Command Queue Virtualization (CMDQV) hardware is part of the
SMMUv3 implementation on NVIDIA Tegra SoCs. It assists in
virtualizing the command queue for the SMMU.

Update SMMU compatible strings to use nvidia,tegra264-smmu to enable
CMDQV support. Add device tree nodes for the CMDQV hardware and enable
them on the tegra264-p3834 platform where SMMUs are enabled. Each SMMU
instance is paired with its corresponding CMDQV instance via the
nvidia,cmdqv property.

Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
---
 .../arm64/boot/dts/nvidia/tegra264-p3834.dtsi |  8 +++
 arch/arm64/boot/dts/nvidia/tegra264.dtsi      | 50 +++++++++++++++++--
 2 files changed, 53 insertions(+), 5 deletions(-)

diff --git a/arch/arm64/boot/dts/nvidia/tegra264-p3834.dtsi b/arch/arm64/boot/dts/nvidia/tegra264-p3834.dtsi
index 06795c82427a..7e2c3e66c2ab 100644
--- a/arch/arm64/boot/dts/nvidia/tegra264-p3834.dtsi
+++ b/arch/arm64/boot/dts/nvidia/tegra264-p3834.dtsi
@@ -23,8 +23,16 @@ iommu@5000000 {
 			status = "okay";
 		};
 
+		cmdqv@5200000 {
+			status = "okay";
+		};
+
 		iommu@6000000 {
 			status = "okay";
 		};
+
+		cmdqv@6200000 {
+			status = "okay";
+		};
 	};
 };
diff --git a/arch/arm64/boot/dts/nvidia/tegra264.dtsi b/arch/arm64/boot/dts/nvidia/tegra264.dtsi
index f137565da804..9eb7058e3149 100644
--- a/arch/arm64/boot/dts/nvidia/tegra264.dtsi
+++ b/arch/arm64/boot/dts/nvidia/tegra264.dtsi
@@ -3361,7 +3361,7 @@ bus@8100000000 {
 			 <0x02 0x00000000 0xd0 0x00000000 0x08 0x80000000>; /* ECAM, prefetchable memory, I/O */
 
 		smmu1: iommu@5000000 {
-			compatible = "arm,smmu-v3";
+			compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
 			reg = <0x00 0x5000000 0x0 0x200000>;
 			interrupts = <GIC_SPI 12 IRQ_TYPE_EDGE_RISING>,
 				     <GIC_SPI 13 IRQ_TYPE_EDGE_RISING>;
@@ -3370,10 +3370,18 @@ smmu1: iommu@5000000 {
 
 			#iommu-cells = <1>;
 			dma-coherent;
+			nvidia,cmdqv = <&cmdqv1>;
+		};
+
+		cmdqv1: cmdqv@5200000 {
+			compatible = "nvidia,tegra264-cmdqv";
+			reg = <0x00 0x5200000 0x0 0x830000>;
+			interrupts = <GIC_SPI 19 IRQ_TYPE_LEVEL_HIGH>;
+			status = "disabled";
 		};
 
 		smmu2: iommu@6000000 {
-			compatible = "arm,smmu-v3";
+			compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
 			reg = <0x00 0x6000000 0x0 0x200000>;
 			interrupts = <GIC_SPI 1 IRQ_TYPE_EDGE_RISING>,
 				     <GIC_SPI 2 IRQ_TYPE_EDGE_RISING>;
@@ -3382,6 +3390,14 @@ smmu2: iommu@6000000 {
 
 			#iommu-cells = <1>;
 			dma-coherent;
+			nvidia,cmdqv = <&cmdqv2>;
+		};
+
+		cmdqv2: cmdqv@6200000 {
+			compatible = "nvidia,tegra264-cmdqv";
+			reg = <0x00 0x6200000 0x0 0x830000>;
+			interrupts = <GIC_SPI 8 IRQ_TYPE_LEVEL_HIGH>;
+			status = "disabled";
 		};
 
 		mc: memory-controller@8020000 {
@@ -3437,7 +3453,7 @@ emc: external-memory-controller@8800000 {
 		};
 
 		smmu0: iommu@a000000 {
-			compatible = "arm,smmu-v3";
+			compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
 			reg = <0x00 0xa000000 0x0 0x200000>;
 			interrupts = <GIC_SPI 21 IRQ_TYPE_EDGE_RISING>,
 				     <GIC_SPI 22 IRQ_TYPE_EDGE_RISING>;
@@ -3446,10 +3462,18 @@ smmu0: iommu@a000000 {
 
 			#iommu-cells = <1>;
 			dma-coherent;
+			nvidia,cmdqv = <&cmdqv0>;
+		};
+
+		cmdqv0: cmdqv@a200000 {
+			compatible = "nvidia,tegra264-cmdqv";
+			reg = <0x00 0xa200000 0x0 0x830000>;
+			interrupts = <GIC_SPI 28 IRQ_TYPE_LEVEL_HIGH>;
+			status = "disabled";
 		};
 
 		smmu4: iommu@b000000 {
-			compatible = "arm,smmu-v3";
+			compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
 			reg = <0x00 0xb000000 0x0 0x200000>;
 			interrupts = <GIC_SPI 30 IRQ_TYPE_EDGE_RISING>,
 				     <GIC_SPI 31 IRQ_TYPE_EDGE_RISING>;
@@ -3458,6 +3482,14 @@ smmu4: iommu@b000000 {
 
 			#iommu-cells = <1>;
 			dma-coherent;
+			nvidia,cmdqv = <&cmdqv4>;
+		};
+
+		cmdqv4: cmdqv@b200000 {
+			compatible = "nvidia,tegra264-cmdqv";
+			reg = <0x00 0xb200000 0x0 0x830000>;
+			interrupts = <GIC_SPI 37 IRQ_TYPE_LEVEL_HIGH>;
+			status = "disabled";
 		};
 
 		i2c14: i2c@c410000 {
@@ -3690,7 +3722,7 @@ bus@8800000000 {
 		ranges = <0x00 0x00000000 0x88 0x00000000 0x01 0x00000000>;
 
 		smmu3: iommu@6000000 {
-			compatible = "arm,smmu-v3";
+			compatible = "nvidia,tegra264-smmu", "arm,smmu-v3";
 			reg = <0x00 0x6000000 0x0 0x200000>;
 			interrupts = <GIC_SPI 225 IRQ_TYPE_EDGE_RISING>,
 				     <GIC_SPI 226 IRQ_TYPE_EDGE_RISING>;
@@ -3699,6 +3731,14 @@ smmu3: iommu@6000000 {
 
 			#iommu-cells = <1>;
 			dma-coherent;
+			nvidia,cmdqv = <&cmdqv3>;
+		};
+
+		cmdqv3: cmdqv@6200000 {
+			compatible = "nvidia,tegra264-cmdqv";
+			reg = <0x00 0x6200000 0x0 0x830000>;
+			interrupts = <GIC_SPI 232 IRQ_TYPE_LEVEL_HIGH>;
+			status = "disabled";
 		};
 
 		hda@90b0000 {
-- 
2.25.1

Re: [PATCH V6 1/4] iommu/tegra241-cmdqv: Decouple driver from ACPI

[Robin Murphy]

On 2025-12-12 6:08 am, Ashish Mhetre wrote:
> From: Nicolin Chen <nicolinc@nvidia.com>
> 
> A platform device is created by acpi_create_platform_device() per CMDQV's
> adev. That means there is no point in going through _CRS of ACPI.
> 
> Replace all the ACPI functions with standard platform functions. And drop
> all ACPI dependencies. This will make the driver compatible with DT also.
> 
> Suggested-by: Robin Murphy <robin.murphy@arm.com>
> Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
> ---
>   drivers/iommu/arm/Kconfig                     |  1 -
>   drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   |  2 +-
>   .../iommu/arm/arm-smmu-v3/tegra241-cmdqv.c    | 84 +++----------------
>   3 files changed, 13 insertions(+), 74 deletions(-)
> 
> diff --git a/drivers/iommu/arm/Kconfig b/drivers/iommu/arm/Kconfig
> index ef42bbe07dbe..5fac08b89dee 100644
> --- a/drivers/iommu/arm/Kconfig
> +++ b/drivers/iommu/arm/Kconfig
> @@ -121,7 +121,6 @@ config ARM_SMMU_V3_KUNIT_TEST
>   
>   config TEGRA241_CMDQV
>   	bool "NVIDIA Tegra241 CMDQ-V extension support for ARM SMMUv3"
> -	depends on ACPI
>   	help
>   	  Support for NVIDIA CMDQ-Virtualization extension for ARM SMMUv3. The
>   	  CMDQ-V extension is similar to v3.3 ECMDQ for multi command queues
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> index d16d35c78c06..42de3da54858 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> @@ -4542,7 +4542,7 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
>   	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
>   	if (adev) {
>   		/* Tegra241 CMDQV driver is responsible for put_device() */

Don't we need to bring the put_device(adev) out to this level, since 
impl_dev is now something else that AFAICS we are *not* taking a new 
reference on (and thus should not be putting either)?

Otherwise this looks pretty nice!

Thanks,
Robin.

> -		smmu->impl_dev = &adev->dev;
> +		smmu->impl_dev = acpi_get_first_physical_node(adev);
>   		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
>   		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
>   			 dev_name(smmu->impl_dev));
> diff --git a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
> index 378104cd395e..1fc03b72beb8 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
> @@ -3,17 +3,15 @@
>   
>   #define dev_fmt(fmt) "tegra241_cmdqv: " fmt
>   
> -#include <linux/acpi.h>
>   #include <linux/debugfs.h>
>   #include <linux/dma-mapping.h>
>   #include <linux/interrupt.h>
>   #include <linux/iommu.h>
>   #include <linux/iommufd.h>
>   #include <linux/iopoll.h>
> +#include <linux/platform_device.h>
>   #include <uapi/linux/iommufd.h>
>   
> -#include <acpi/acpixf.h>
> -
>   #include "arm-smmu-v3.h"
>   
>   /* CMDQV register page base and size defines */
> @@ -854,69 +852,6 @@ static struct arm_smmu_impl_ops tegra241_cmdqv_impl_ops = {
>   
>   /* Probe Functions */
>   
> -static int tegra241_cmdqv_acpi_is_memory(struct acpi_resource *res, void *data)
> -{
> -	struct resource_win win;
> -
> -	return !acpi_dev_resource_address_space(res, &win);
> -}
> -
> -static int tegra241_cmdqv_acpi_get_irqs(struct acpi_resource *ares, void *data)
> -{
> -	struct resource r;
> -	int *irq = data;
> -
> -	if (*irq <= 0 && acpi_dev_resource_interrupt(ares, 0, &r))
> -		*irq = r.start;
> -	return 1; /* No need to add resource to the list */
> -}
> -
> -static struct resource *
> -tegra241_cmdqv_find_acpi_resource(struct device *dev, int *irq)
> -{
> -	struct acpi_device *adev = to_acpi_device(dev);
> -	struct list_head resource_list;
> -	struct resource_entry *rentry;
> -	struct resource *res = NULL;
> -	int ret;
> -
> -	INIT_LIST_HEAD(&resource_list);
> -	ret = acpi_dev_get_resources(adev, &resource_list,
> -				     tegra241_cmdqv_acpi_is_memory, NULL);
> -	if (ret < 0) {
> -		dev_err(dev, "failed to get memory resource: %d\n", ret);
> -		return NULL;
> -	}
> -
> -	rentry = list_first_entry_or_null(&resource_list,
> -					  struct resource_entry, node);
> -	if (!rentry) {
> -		dev_err(dev, "failed to get memory resource entry\n");
> -		goto free_list;
> -	}
> -
> -	/* Caller must free the res */
> -	res = kzalloc(sizeof(*res), GFP_KERNEL);
> -	if (!res)
> -		goto free_list;
> -
> -	*res = *rentry->res;
> -
> -	acpi_dev_free_resource_list(&resource_list);
> -
> -	INIT_LIST_HEAD(&resource_list);
> -
> -	if (irq)
> -		ret = acpi_dev_get_resources(adev, &resource_list,
> -					     tegra241_cmdqv_acpi_get_irqs, irq);
> -	if (ret < 0 || !irq || *irq <= 0)
> -		dev_warn(dev, "no interrupt. errors will not be reported\n");
> -
> -free_list:
> -	acpi_dev_free_resource_list(&resource_list);
> -	return res;
> -}
> -
>   static int tegra241_cmdqv_init_structures(struct arm_smmu_device *smmu)
>   {
>   	struct tegra241_cmdqv *cmdqv =
> @@ -1042,18 +977,23 @@ __tegra241_cmdqv_probe(struct arm_smmu_device *smmu, struct resource *res,
>   
>   struct arm_smmu_device *tegra241_cmdqv_probe(struct arm_smmu_device *smmu)
>   {
> +	struct platform_device *pdev = to_platform_device(smmu->impl_dev);
>   	struct arm_smmu_device *new_smmu;
> -	struct resource *res = NULL;
> +	struct resource *res;
>   	int irq;
>   
> -	if (!smmu->dev->of_node)
> -		res = tegra241_cmdqv_find_acpi_resource(smmu->impl_dev, &irq);
> -	if (!res)
> +	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> +	if (!res) {
> +		dev_err(&pdev->dev, "no memory resource found for CMDQV\n");
>   		goto out_fallback;
> +	}
>   
> -	new_smmu = __tegra241_cmdqv_probe(smmu, res, irq);
> -	kfree(res);
> +	irq = platform_get_irq_optional(pdev, 0);
> +	if (irq <= 0)
> +		dev_warn(&pdev->dev,
> +			 "no interrupt. errors will not be reported\n");
>   
> +	new_smmu = __tegra241_cmdqv_probe(smmu, res, irq);
>   	if (new_smmu)
>   		return new_smmu;
>   

Re: [PATCH V6 1/4] iommu/tegra241-cmdqv: Decouple driver from ACPI

[Nicolin Chen]

On Fri, Dec 12, 2025 at 12:01:41PM +0000, Robin Murphy wrote:
> > @@ -4542,7 +4542,7 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
> >   	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
> >   	if (adev) {
> >   		/* Tegra241 CMDQV driver is responsible for put_device() */
> 
> Don't we need to bring the put_device(adev) out to this level, since
> impl_dev is now something else that AFAICS we are *not* taking a new
> reference on (and thus should not be putting either)?

Ah, right! We are using the platform device now.

> > -		smmu->impl_dev = &adev->dev;
> > +		smmu->impl_dev = acpi_get_first_physical_node(adev);
> >   		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
> >   		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
> >   			 dev_name(smmu->impl_dev));

I think we should squash this:

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index 0c98be3135c63..88625e3c27a65 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -5280,10 +5280,11 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
 	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
 	if (adev) {
 		/* Tegra241 CMDQV driver is responsible for put_device() */
-		smmu->impl_dev = acpi_get_first_physical_node(adev);
+		smmu->impl_dev = get_device(acpi_get_first_physical_node(adev));
 		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
 		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
 			 dev_name(smmu->impl_dev));
+		acpi_dev_put(adev);
 	}
 	kfree(uid);
 }

Thanks
Nicolin

Re: [PATCH V6 1/4] iommu/tegra241-cmdqv: Decouple driver from ACPI

[Robin Murphy]

On 12/12/2025 7:46 pm, Nicolin Chen wrote:
> On Fri, Dec 12, 2025 at 12:01:41PM +0000, Robin Murphy wrote:
>>> @@ -4542,7 +4542,7 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
>>>    	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
>>>    	if (adev) {
>>>    		/* Tegra241 CMDQV driver is responsible for put_device() */
>>
>> Don't we need to bring the put_device(adev) out to this level, since
>> impl_dev is now something else that AFAICS we are *not* taking a new
>> reference on (and thus should not be putting either)?
> 
> Ah, right! We are using the platform device now.
> 
>>> -		smmu->impl_dev = &adev->dev;
>>> +		smmu->impl_dev = acpi_get_first_physical_node(adev);
>>>    		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
>>>    		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
>>>    			 dev_name(smmu->impl_dev));
> 
> I think we should squash this:

Ack, it occurred to me shortly after sending that in fact it's good to 
still hold our own reference on impl_dev itself, as we're not relying on 
a bound driver to hold one for us.

Cheers,
Robin.

> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> index 0c98be3135c63..88625e3c27a65 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> @@ -5280,10 +5280,11 @@ static void acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
>   	adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
>   	if (adev) {
>   		/* Tegra241 CMDQV driver is responsible for put_device() */
> -		smmu->impl_dev = acpi_get_first_physical_node(adev);
> +		smmu->impl_dev = get_device(acpi_get_first_physical_node(adev));
>   		smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
>   		dev_info(smmu->dev, "found companion CMDQV device: %s\n",
>   			 dev_name(smmu->impl_dev));
> +		acpi_dev_put(adev);
>   	}
>   	kfree(uid);
>   }
> 
> Thanks
> Nicolin

Re: [PATCH V6 1/4] iommu/tegra241-cmdqv: Decouple driver from ACPI

[Ashish Mhetre]

On 12/13/2025 2:06 AM, Robin Murphy wrote:
> External email: Use caution opening links or attachments
>
>
> On 12/12/2025 7:46 pm, Nicolin Chen wrote:
>> On Fri, Dec 12, 2025 at 12:01:41PM +0000, Robin Murphy wrote:
>>>> @@ -4542,7 +4542,7 @@ static void 
>>>> acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
>>>>     adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
>>>>     if (adev) {
>>>>             /* Tegra241 CMDQV driver is responsible for 
>>>> put_device() */
>>>
>>> Don't we need to bring the put_device(adev) out to this level, since
>>> impl_dev is now something else that AFAICS we are *not* taking a new
>>> reference on (and thus should not be putting either)?
>>
>> Ah, right! We are using the platform device now.
>>
>>>> -           smmu->impl_dev = &adev->dev;
>>>> +           smmu->impl_dev = acpi_get_first_physical_node(adev);
>>>>             smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
>>>>             dev_info(smmu->dev, "found companion CMDQV device: %s\n",
>>>>                      dev_name(smmu->impl_dev));
>>
>> I think we should squash this:
>
> Ack, it occurred to me shortly after sending that in fact it's good to
> still hold our own reference on impl_dev itself, as we're not relying on
> a bound driver to hold one for us.
>
> Cheers,
> Robin.
>

Ack, thanks Nic and Robin. I will include this and send
new version.

>> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c 
>> b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>> index 0c98be3135c63..88625e3c27a65 100644
>> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>> @@ -5280,10 +5280,11 @@ static void 
>> acpi_smmu_dsdt_probe_tegra241_cmdqv(struct acpi_iort_node *node,
>>       adev = acpi_dev_get_first_match_dev("NVDA200C", uid, -1);
>>       if (adev) {
>>               /* Tegra241 CMDQV driver is responsible for 
>> put_device() */
>> -             smmu->impl_dev = acpi_get_first_physical_node(adev);
>> +             smmu->impl_dev = 
>> get_device(acpi_get_first_physical_node(adev));
>>               smmu->options |= ARM_SMMU_OPT_TEGRA241_CMDQV;
>>               dev_info(smmu->dev, "found companion CMDQV device: %s\n",
>>                        dev_name(smmu->impl_dev));
>> +             acpi_dev_put(adev);
>>       }
>>       kfree(uid);
>>   }
>>
>> Thanks
>> Nicolin
>