[PATCH 0/5] drivers/net: add NXP FEC-UIO driver

[PATCH 0/5] drivers/net: add NXP FEC-UIO driver

[Apeksha Gupta]

This patch series introduce the fec-uio driver, supported for the inbuilt
NIC found in the NXP i.MX 8M Mini SoC. Basic hardware initialization is
performed in kernel via userspace input/output(UIO) to support FEC
ethernet device detection in user space.

Userspace PMD uses standard UIO interface to access kernel for PHY
initialisation and for mapping the allocated memory of register &
buffer descriptor with DPDK which gives access to non-cacheable memory
for buffer descriptor.

Module fec-uio.ko will get generated.
imx8mm-evk-dpdk.dtb is required to support fec-uio driver.

Apeksha Gupta (5):
  dt-bindings: add binding for fec-uio
  net: fec: fec-uio driver
  ARM64: defconfig: Add config for fec-uio
  MAINTAINERS: add new file
  arm64: dts: imx8mm-evk-dpdk: dts for fec-uio driver

 .../devicetree/bindings/net/fsl,fec-uio.yaml  |  32 ++
 MAINTAINERS                                   |   6 +
 arch/arm64/boot/dts/freescale/Makefile        |   1 +
 .../boot/dts/freescale/imx8mm-evk-dpdk.dts    |  10 +
 arch/arm64/configs/defconfig                  |   1 +
 drivers/net/ethernet/freescale/Kconfig        |  10 +
 drivers/net/ethernet/freescale/Makefile       |   7 +-
 drivers/net/ethernet/freescale/fec_uio.c      | 437 ++++++++++++++++++
 8 files changed, 501 insertions(+), 3 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/net/fsl,fec-uio.yaml
 create mode 100644 arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts
 create mode 100644 drivers/net/ethernet/freescale/fec_uio.c

-- 
2.17.1

[PATCH 1/5] dt-bindings: add binding for fec-uio

[Apeksha Gupta]

binding 'fsl,fec-uio.yaml' represents the basic hardware
initialization which is performed in fec-uio via userspace
input/output to support FEC ethernet device detection in
user space.

Signed-off-by: Sachin Saxena <sachin.saxena@nxp.com>
Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
---
 .../devicetree/bindings/net/fsl,fec-uio.yaml  | 32 +++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/net/fsl,fec-uio.yaml

diff --git a/Documentation/devicetree/bindings/net/fsl,fec-uio.yaml b/Documentation/devicetree/bindings/net/fsl,fec-uio.yaml
new file mode 100644
index 000000000000..dd7477c0d213
--- /dev/null
+++ b/Documentation/devicetree/bindings/net/fsl,fec-uio.yaml
@@ -0,0 +1,32 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/net/fsl,fec-uio.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: NXP Fast Ethernet Controller (FEC) with UIO
+
+maintainers:
+  - Apeksha Gupta <apeksha.gupta@nxp.com>
+  - Sachin Saxena <sachin.saxena@nxp.com>
+
+description:
+  This binding represents the basic hardware initialization which is
+  performed in fec-uio via userspace input/output, to support fec
+  ethernet device detection in user space.
+
+allOf:
+  - $ref: "fsl,fec.yaml#"
+
+properties:
+  compatible:
+    const: fsl,imx8mm-fec-uio
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+additionalProperties: false
-- 
2.17.1

[PATCH 2/5] net: fec: fec-uio driver

[Apeksha Gupta]

i.mx: fec-uio driver

This patch adds the userspace support. In this basic
hardware initialization is performed in kernel via userspace
input/output, while the majority of code is written in the
userspace.

Module fec-uio.ko is generated.
imx8mm-evk-dpdk.dtb is required to support fec-uio driver.

Signed-off-by: Sachin Saxena <sachin.saxena@nxp.com>
Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
---
 drivers/net/ethernet/freescale/Kconfig   |  10 +
 drivers/net/ethernet/freescale/Makefile  |   7 +-
 drivers/net/ethernet/freescale/fec_uio.c | 437 +++++++++++++++++++++++
 3 files changed, 451 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/freescale/fec_uio.c

diff --git a/drivers/net/ethernet/freescale/Kconfig b/drivers/net/ethernet/freescale/Kconfig
index e04e1c5cb013..c41289e9ebfa 100644
--- a/drivers/net/ethernet/freescale/Kconfig
+++ b/drivers/net/ethernet/freescale/Kconfig
@@ -33,6 +33,16 @@ config FEC
 	  Say Y here if you want to use the built-in 10/100 Fast ethernet
 	  controller on some Motorola ColdFire and Freescale i.MX processors.
 
+config FEC_UIO
+	tristate "FEC_UIO ethernet controller (i.MX 8M Mini CPU)"
+	default n
+	select UIO
+	help
+	  Say Y here if you want to use the built-in 10/100/1000-Mbit/s Fast
+	  ethernet controller on Freescale i.MX 8M Mini processor. Basic
+	  hardware initialization is performed in kernel via UIO, fec-uio
+	  driver adds the userspace support.
+
 config FEC_MPC52xx
 	tristate "FEC MPC52xx driver"
 	depends on PPC_MPC52xx && PPC_BESTCOMM
diff --git a/drivers/net/ethernet/freescale/Makefile b/drivers/net/ethernet/freescale/Makefile
index 61d417694e0e..a32c807a626b 100644
--- a/drivers/net/ethernet/freescale/Makefile
+++ b/drivers/net/ethernet/freescale/Makefile
@@ -2,11 +2,12 @@
 #
 # Makefile for the Freescale network device drivers.
 #
-
-common-objs := fec_phy.o
+common-obj := fec_phy.o
 
 obj-$(CONFIG_FEC) += fec.o
-fec-objs :=fec_main.o fec_ptp.o $(common-objs)
+fec-objs :=fec_main.o fec_ptp.o $(common-obj)
+obj-$(CONFIG_FEC_UIO) += enetfec_uio.o
+enetfec_uio-objs := fec_uio.o $(common-obj)
 
 obj-$(CONFIG_FEC_MPC52xx) += fec_mpc52xx.o
 ifeq ($(CONFIG_FEC_MPC52xx_MDIO),y)
diff --git a/drivers/net/ethernet/freescale/fec_uio.c b/drivers/net/ethernet/freescale/fec_uio.c
new file mode 100644
index 000000000000..5aa90e1f5c55
--- /dev/null
+++ b/drivers/net/ethernet/freescale/fec_uio.c
@@ -0,0 +1,437 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2021 NXP
+ */
+
+#include <linux/kernel.h>
+#include <linux/of_platform.h>
+#include <linux/uio_driver.h>
+#include <linux/pm_runtime.h>
+#include <linux/errno.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/regulator/consumer.h>
+#include <linux/clk.h>
+#include <linux/of_gpio.h>
+#include <linux/of_net.h>
+
+#include "fec.h"
+#include "fec_phy.h"
+
+struct fec_dev *fec_dev;
+static const char fec_uio_version[] = "FEC UIO driver v1.0";
+dma_addr_t bd_dma;
+int bd_size;
+struct bufdesc *cbd_base;
+
+#define NAME_LENGTH		10
+#define DRIVER_NAME		"fec-uio"
+#define FEC_PRIV_SIZE		200
+#define FEC_MAX_Q		3
+#define ENABLE_ENET		BIT(8)
+#define ETHER_EN		0x2
+
+static const char uio_device_name[] = "imx-fec-uio";
+struct fec_uio_info {
+	atomic_t ref; /* exclusive, only one open() at a time */
+	struct uio_info uio_info;
+	char name[NAME_LENGTH];
+};
+
+struct fec_dev {
+	u32 index;
+	struct device *dev;
+	struct resource *res;
+	struct fec_uio_info info;
+};
+
+struct fec_uio_devinfo {
+	u32 quirks;
+};
+
+static const struct fec_uio_devinfo fec_imx8mm_info = {
+	.quirks = FEC_QUIRK_ENET_MAC,
+};
+
+static struct platform_device_id fec_enet_uio_devtype[] = {
+	{
+		.name = DRIVER_NAME,
+		.driver_data = (kernel_ulong_t)&fec_imx8mm_info,
+	}, {
+		.name = "imx8mm-fec",
+		.driver_data = (kernel_ulong_t)&fec_imx8mm_info,
+	}, {
+		/* sentinel */
+	}
+};
+MODULE_DEVICE_TABLE(platform, fec_enet_uio_devtype);
+
+static const struct of_device_id fec_enet_uio_ids[] = {
+	{ .compatible = "fsl,imx8mm-fec-uio", .data = &fec_enet_uio_devtype },
+	{ /* sentinel */ },
+};
+MODULE_DEVICE_TABLE(of, fec_enet_uio_ids);
+
+static unsigned char macaddr[ETH_ALEN];
+module_param_array(macaddr, byte, NULL, 0);
+MODULE_PARM_DESC(macaddr, "FEC Ethernet MAC address");
+
+static int fec_uio_open(struct uio_info *info, struct inode *inode)
+{
+	return 0;
+}
+
+static int fec_uio_release(struct uio_info *info, struct inode *inode)
+{
+	return 0;
+}
+
+static int fec_uio_mmap(struct uio_info *info, struct vm_area_struct *vma)
+{
+	u32 ret;
+	u32 pfn;
+
+	pfn = (info->mem[vma->vm_pgoff].addr) >> PAGE_SHIFT;
+
+	if (vma->vm_pgoff)
+		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+	else
+		vma->vm_page_prot = pgprot_device(vma->vm_page_prot);
+
+	ret = remap_pfn_range(vma, vma->vm_start, pfn,
+			      vma->vm_end - vma->vm_start, vma->vm_page_prot);
+	if (ret) {
+		/* Error Handle */
+		pr_info("remap_pfn_range failed");
+	}
+	return ret;
+}
+
+static int __init fec_uio_init(struct fec_dev *fec_dev)
+{
+	struct fec_uio_info *fec_uio_info;
+	int ret;
+
+	fec_uio_info = &fec_dev->info;
+	atomic_set(&fec_uio_info->ref, 0);
+	fec_uio_info->uio_info.version = fec_uio_version;
+	fec_uio_info->uio_info.name = fec_dev->info.name;
+
+	fec_uio_info->uio_info.mem[0].name = "FEC_REG_SPACE";
+	fec_uio_info->uio_info.mem[0].addr = fec_dev->res->start;
+	fec_uio_info->uio_info.mem[0].size = 0x1000;
+	fec_uio_info->uio_info.mem[0].internal_addr = 0;
+	fec_uio_info->uio_info.mem[0].memtype = UIO_MEM_PHYS;
+
+	fec_uio_info->uio_info.mem[1].name = "FEC_BD_SPACE";
+	fec_uio_info->uio_info.mem[1].addr = bd_dma;
+	fec_uio_info->uio_info.mem[1].size = bd_size;
+	fec_uio_info->uio_info.mem[1].memtype = UIO_MEM_PHYS;
+
+	fec_uio_info->uio_info.open = fec_uio_open;
+	fec_uio_info->uio_info.release = fec_uio_release;
+	/* Custom mmap function. */
+	fec_uio_info->uio_info.mmap = fec_uio_mmap;
+	fec_uio_info->uio_info.priv = fec_dev;
+
+	ret = uio_register_device(fec_dev->dev, &fec_uio_info->uio_info);
+	/* return if driver requests probe retry */
+	if (ret == -EPROBE_DEFER)
+		return ret;
+	if (ret) {
+		dev_err(fec_dev->dev, "fec_uio: UIO registration failed\n");
+		return ret;
+	}
+	return 0;
+}
+
+/* Calculate the Rx,Tx ring size to get the total buffer discriptor(BD)
+ * size and allocate memory for BD ring.
+ */
+static int fec_enet_uio_init(struct net_device *ndev)
+{
+	unsigned int total_tx_ring_size = 0, total_rx_ring_size = 0;
+	struct fec_enet_private *fep = netdev_priv(ndev);
+	unsigned int dsize = sizeof(struct bufdesc);
+	unsigned short tx_ring_size, rx_ring_size;
+	int ret, i;
+
+	/* Check mask of the streaming and coherent API */
+	ret = dma_set_mask_and_coherent(&fep->pdev->dev, DMA_BIT_MASK(32));
+	if (ret < 0) {
+		dev_warn(&fep->pdev->dev, "No suitable DMA available\n");
+		return ret;
+	}
+
+	tx_ring_size = TX_RING_SIZE;
+	rx_ring_size = RX_RING_SIZE;
+
+	for (i = 0; i <	FEC_ENET_MAX_TX_QS; i++)
+		total_tx_ring_size += tx_ring_size;
+	for (i = 0; i <	FEC_ENET_MAX_RX_QS; i++)
+		total_rx_ring_size += rx_ring_size;
+
+	bd_size = (total_tx_ring_size + total_rx_ring_size) * dsize;
+
+	/* Allocate memory for buffer descriptors. */
+	cbd_base = dma_alloc_coherent(&fep->pdev->dev, bd_size, &bd_dma,
+				      GFP_KERNEL);
+	if (!cbd_base) {
+		ret = -ENOMEM;
+		goto free_mem;
+	}
+
+	return 0;
+free_mem:
+	dma_free_coherent(&fep->pdev->dev, bd_size, cbd_base, bd_dma);
+	return ret;
+}
+
+static int
+fec_enet_uio_probe(struct platform_device *pdev)
+{
+	struct fec_uio_devinfo *dev_info;
+	const struct of_device_id *of_id;
+	struct fec_enet_private *fep;
+	struct net_device *ndev;
+	u32 ecntl = ETHER_EN;
+	static int dev_id;
+	bool reset_again;
+	int ret = 0;
+
+	/* Init network device */
+	ndev = alloc_etherdev_mq(sizeof(struct fec_enet_private) +
+				FEC_PRIV_SIZE, FEC_MAX_Q);
+	if (!ndev)
+		return -ENOMEM;
+
+	SET_NETDEV_DEV(ndev, &pdev->dev);
+
+	/* setup board info structure */
+	fep = netdev_priv(ndev);
+
+	of_id = of_match_device(fec_enet_uio_ids, &pdev->dev);
+	if (of_id)
+		pdev->id_entry = of_id->data;
+
+	dev_info = (struct fec_uio_devinfo *)pdev->id_entry->driver_data;
+	if (dev_info)
+		fep->quirks = dev_info->quirks;
+
+	/* Select default pin state */
+	pinctrl_pm_select_default_state(&pdev->dev);
+
+	/* allocate memory for uio structure */
+	fec_dev = kzalloc(sizeof(*fec_dev), GFP_KERNEL);
+	if (!fec_dev)
+		return -ENOMEM;
+
+	snprintf(fec_dev->info.name, sizeof(fec_dev->info.name) - 1,
+		 "%s", uio_device_name);
+
+	fec_dev->dev = &pdev->dev;
+
+	fec_dev->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	fep->hwp = ioremap(fec_dev->res->start, 0x1000);
+	if (IS_ERR(fep->hwp)) {
+		ret = PTR_ERR(fep->hwp);
+		goto failed_ioremap;
+	}
+	fep->pdev = pdev;
+	fep->dev_id = dev_id++;
+
+	platform_set_drvdata(pdev, ndev);
+
+	fep->clk_ipg = devm_clk_get(&pdev->dev, "ipg");
+	if (IS_ERR(fep->clk_ipg)) {
+		ret = PTR_ERR(fep->clk_ipg);
+		goto failed_clk;
+	}
+
+	fep->clk_ahb = devm_clk_get(&pdev->dev, "ahb");
+	if (IS_ERR(fep->clk_ahb)) {
+		ret = PTR_ERR(fep->clk_ahb);
+		goto failed_clk;
+	}
+
+	/* enet_out is optional, depends on board */
+	fep->clk_enet_out = devm_clk_get(&pdev->dev, "enet_out");
+	if (IS_ERR(fep->clk_enet_out))
+		fep->clk_enet_out = NULL;
+
+	/* clk_ref is optional, depends on board */
+	fep->clk_ref = devm_clk_get(&pdev->dev, "enet_clk_ref");
+	if (IS_ERR(fep->clk_ref))
+		fep->clk_ref = NULL;
+
+	ret = clk_prepare_enable(fep->clk_enet_out);
+	if (ret)
+		return ret;
+
+	ret = clk_prepare_enable(fep->clk_ref);
+	if (ret)
+		goto failed_clk_ref;
+
+	fec_enet_phy_reset_after_clk_enable(ndev);
+
+	ret = clk_prepare_enable(fep->clk_ipg);
+	if (ret)
+		goto failed_clk_ipg;
+
+	ret = clk_prepare_enable(fep->clk_ahb);
+	if (ret)
+		goto failed_clk_ahb;
+
+	fep->reg_phy = devm_regulator_get_optional(&pdev->dev, "phy");
+	if (!IS_ERR(fep->reg_phy)) {
+		ret = regulator_enable(fep->reg_phy);
+		if (ret) {
+			dev_err(&pdev->dev,
+				"Failed to enable phy regulator: %d\n", ret);
+			goto failed_regulator;
+		}
+	} else {
+		if (PTR_ERR(fep->reg_phy) == -EPROBE_DEFER) {
+			ret = -EPROBE_DEFER;
+			goto failed_regulator;
+		}
+		fep->reg_phy = NULL;
+	}
+
+	pm_runtime_enable(&pdev->dev);
+	ret = fec_reset_phy(pdev);
+	if (ret)
+		goto failed_reset;
+
+	ret = fec_enet_uio_init(ndev);
+	if (ret)
+		goto failed_init;
+
+	/* Register UIO */
+	ret = fec_uio_init(fec_dev);
+	if (ret) {
+		/* return if driver requests probe retry */
+		if (ret == -EPROBE_DEFER) {
+			dev_info(&pdev->dev,
+				 "Driver request probe retry: %s\n", __func__);
+			goto out_unmap;
+		} else {
+			dev_err(&pdev->dev, "UIO init Failed\n");
+			goto abort;
+		}
+	}
+	dev_info(fec_dev->dev, "UIO device full name %s initialized\n",
+		 fec_dev->info.name);
+
+	if (fep->quirks & FEC_QUIRK_ENET_MAC) {
+		/* enable ENET endian swap */
+		ecntl |= ENABLE_ENET;
+		/* enable ENET store and forward mode */
+		writel(ENABLE_ENET, fep->hwp + FEC_X_WMRK);
+	}
+
+	/* And last, enable the transmit and receive processing */
+	writel(ecntl, fep->hwp + FEC_ECNTRL);
+
+	ret = fec_enet_mii_init(pdev);
+	if (ret)
+		goto failed_mii_init;
+
+	if (ndev->phydev && ndev->phydev->drv)
+		reset_again = false;
+	else
+		reset_again = true;
+
+	return 0;
+
+failed_mii_init:
+failed_init:
+failed_reset:
+	pm_runtime_disable(&pdev->dev);
+	if (fep->reg_phy)
+		regulator_disable(fep->reg_phy);
+failed_clk_ref:
+	clk_disable_unprepare(fep->clk_enet_out);
+failed_regulator:
+	clk_disable_unprepare(fep->clk_ahb);
+failed_clk_ahb:
+	clk_disable_unprepare(fep->clk_ipg);
+failed_clk_ipg:
+	clk_disable_unprepare(fep->clk_enet_out);
+	clk_disable_unprepare(fep->clk_ref);
+failed_clk:
+	dev_id--;
+failed_ioremap:
+	free_netdev(ndev);
+
+	return ret;
+out_unmap:
+	dev_id--;
+	kfree(fec_dev);
+	iounmap(fep->hwp);
+	dma_free_coherent(&fep->pdev->dev, bd_size, cbd_base, bd_dma);
+	free_netdev(ndev);
+	clk_disable_unprepare(fep->clk_ahb);
+	clk_disable_unprepare(fep->clk_ipg);
+	clk_disable_unprepare(fep->clk_enet_out);
+	clk_disable_unprepare(fep->clk_ref);
+	pm_runtime_disable(&pdev->dev);
+
+	return -EPROBE_DEFER;
+abort:
+	return ret;
+}
+
+static int
+fec_enet_uio_remove(struct platform_device *pdev)
+{
+	struct net_device *ndev = platform_get_drvdata(pdev);
+	struct fec_enet_private *fep = netdev_priv(ndev);
+
+	kfree(fec_dev);
+	iounmap(fep->hwp);
+	dma_free_coherent(&fep->pdev->dev, bd_size, cbd_base, bd_dma);
+
+	uio_unregister_device(&fec_dev->info.uio_info);
+
+	fec_enet_mii_remove(fep);
+	if (fep->reg_phy)
+		regulator_disable(fep->reg_phy);
+
+	free_netdev(ndev);
+
+	clk_disable_unprepare(fep->clk_ahb);
+	clk_disable_unprepare(fep->clk_ipg);
+	pm_runtime_disable(&pdev->dev);
+
+	return 0;
+}
+
+static struct platform_driver fec_enet_uio_driver = {
+	.driver = {
+		.name = DRIVER_NAME,
+		.of_match_table = fec_enet_uio_ids,
+		.suppress_bind_attrs = true,
+	},
+	.id_table = fec_enet_uio_devtype,
+	.prevent_deferred_probe = false,
+	.probe = fec_enet_uio_probe,
+	.remove = fec_enet_uio_remove,
+};
+
+static int __init fec_enet_uio_load(void)
+{
+	return platform_driver_register(&fec_enet_uio_driver);
+}
+
+static void __exit fec_enet_uio_unload(void)
+{
+	platform_driver_unregister(&fec_enet_uio_driver);
+}
+
+module_init(fec_enet_uio_load);
+module_exit(fec_enet_uio_unload);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("NXP");
+MODULE_DESCRIPTION("i.MX FEC UIO Driver");
-- 
2.17.1

Re: [PATCH 2/5] net: fec: fec-uio driver

[Andrew Lunn]

On Wed, Nov 10, 2021 at 11:18:35AM +0530, Apeksha Gupta wrote:
> i.mx: fec-uio driver
> 
> This patch adds the userspace support. In this basic
> hardware initialization is performed in kernel via userspace
> input/output, while the majority of code is written in the
> userspace.

Where do i find this usespace code. Please include a URL to a git
repo.

> +static unsigned char macaddr[ETH_ALEN];
> +module_param_array(macaddr, byte, NULL, 0);
> +MODULE_PARM_DESC(macaddr, "FEC Ethernet MAC address");

No module parameters please. Use the standard device tree bindings.

> +static int fec_enet_uio_init(struct net_device *ndev)
> +{
> +	unsigned int total_tx_ring_size = 0, total_rx_ring_size = 0;
> +	struct fec_enet_private *fep = netdev_priv(ndev);
> +	unsigned int dsize = sizeof(struct bufdesc);
> +	unsigned short tx_ring_size, rx_ring_size;
> +	int ret, i;
> +
> +	/* Check mask of the streaming and coherent API */
> +	ret = dma_set_mask_and_coherent(&fep->pdev->dev, DMA_BIT_MASK(32));
> +	if (ret < 0) {
> +		dev_warn(&fep->pdev->dev, "No suitable DMA available\n");
> +		return ret;
> +	}
> +
> +	tx_ring_size = TX_RING_SIZE;
> +	rx_ring_size = RX_RING_SIZE;
> +
> +	for (i = 0; i <	FEC_ENET_MAX_TX_QS; i++)
> +		total_tx_ring_size += tx_ring_size;
> +	for (i = 0; i <	FEC_ENET_MAX_RX_QS; i++)
> +		total_rx_ring_size += rx_ring_size;
> +
> +	bd_size = (total_tx_ring_size + total_rx_ring_size) * dsize;

These are the buffer descriptors, not buffers themselves. I assume the
user space driver is allocating the buffer? And your userspace then
set the descriptor to point to user allocated memory? Or some other
memory in the address space, and overwrite whatever you want on the
next DMA? Or DMAing kernel memory out as frames?

> +static int
> +fec_enet_uio_probe(struct platform_device *pdev)
> +{
> +	struct fec_uio_devinfo *dev_info;
> +	const struct of_device_id *of_id;
> +	struct fec_enet_private *fep;
> +	struct net_device *ndev;
> +	u32 ecntl = ETHER_EN;
> +	static int dev_id;
> +	bool reset_again;
> +	int ret = 0;
> +
> +	/* Init network device */
> +	ndev = alloc_etherdev_mq(sizeof(struct fec_enet_private) +
> +				FEC_PRIV_SIZE, FEC_MAX_Q);

Why do you need this. This is not a netdev driver, since it does not
connect to the network stack.

> +static int
> +fec_enet_uio_remove(struct platform_device *pdev)
> +{
> +	struct net_device *ndev = platform_get_drvdata(pdev);
> +	struct fec_enet_private *fep = netdev_priv(ndev);
> +
> +	kfree(fec_dev);
> +	iounmap(fep->hwp);
> +	dma_free_coherent(&fep->pdev->dev, bd_size, cbd_base, bd_dma);

Don't you have to assume that the userspace driver has crashed and
burned, leaving the hardware in an undefined state. It could still be
receiving, into buffers we have no idea about. Don't you need to stop
the hardware, and then wait for all DMA activity to stop, and only
then can you free the buffer descriptors?

     Andrew

[PATCH 3/5] ARM64: defconfig: Add config for fec-uio

[Apeksha Gupta]

This enables fec-uio driver on imx8mm.

Signed-off-by: Sachin Saxena <sachin.saxena@nxp.com>
Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
---
 arch/arm64/configs/defconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index 545197bc0501..2303eda7f511 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -320,6 +320,7 @@ CONFIG_BNX2X=m
 CONFIG_MACB=y
 CONFIG_THUNDER_NIC_PF=y
 CONFIG_FEC=y
+CONFIG_FEC_UIO=m
 CONFIG_FSL_FMAN=y
 CONFIG_FSL_DPAA_ETH=y
 CONFIG_FSL_DPAA2_ETH=y
-- 
2.17.1

[PATCH 4/5] MAINTAINERS: add new file

[Apeksha Gupta]

For fec-uio driver, maintainers and fec_uio.c file added.

Signed-off-by: Sachin Saxena <sachin.saxena@nxp.com>
Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
---
 MAINTAINERS | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index e93fdf73e383..2bfa9a3a91c4 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -7460,6 +7460,12 @@ F:	drivers/net/ethernet/freescale/fec_phy.c
 F:	drivers/net/ethernet/freescale/fec_phy.h
 F:	drivers/net/ethernet/freescale/fec_ptp.c
 
+FREESCALE IMX / FEC-UIO DRIVER
+M:	Apeksha Gupta <apeksha.gupta@nxp.com>
+M:	Sachin Saxena <sachin.saxena@nxp.com>
+F:	Documentation/devicetree/bindings/net/fsl,fec-uio.yaml
+F:	drivers/net/ethernet/freescale/fec_uio.c
+
 FREESCALE IMX / MXC FRAMEBUFFER DRIVER
 M:	Sascha Hauer <s.hauer@pengutronix.de>
 R:	Pengutronix Kernel Team <kernel@pengutronix.de>
-- 
2.17.1

[PATCH 5/5] arm64: dts: imx8mm-evk-dpdk: dts for fec-uio driver

[Apeksha Gupta]

New DTS file 'imx8mm-evk-dpdk.dts' is added to support fec
ethernet device detection in user space DPDK application via
fec-uio driver.

Signed-off-by: Sachin Saxena <sachin.saxena@nxp.com>
Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
---
 arch/arm64/boot/dts/freescale/Makefile            |  1 +
 arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts | 10 ++++++++++
 2 files changed, 11 insertions(+)
 create mode 100644 arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts

diff --git a/arch/arm64/boot/dts/freescale/Makefile b/arch/arm64/boot/dts/freescale/Makefile
index db9e36ebe932..d8aabfc770fd 100644
--- a/arch/arm64/boot/dts/freescale/Makefile
+++ b/arch/arm64/boot/dts/freescale/Makefile
@@ -33,6 +33,7 @@ dtb-$(CONFIG_ARCH_LAYERSCAPE) += fsl-lx2162a-qds.dtb
 
 dtb-$(CONFIG_ARCH_MXC) += imx8mm-beacon-kit.dtb
 dtb-$(CONFIG_ARCH_MXC) += imx8mm-evk.dtb
+dtb-$(CONFIG_ARCH_MXC) += imx8mm-evk-dpdk.dtb
 dtb-$(CONFIG_ARCH_MXC) += imx8mm-ddr4-evk.dtb
 dtb-$(CONFIG_ARCH_MXC) += imx8mm-icore-mx8mm-ctouch2.dtb
 dtb-$(CONFIG_ARCH_MXC) += imx8mm-icore-mx8mm-edimm2.2.dtb
diff --git a/arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts b/arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts
new file mode 100644
index 000000000000..3b5503933a35
--- /dev/null
+++ b/arch/arm64/boot/dts/freescale/imx8mm-evk-dpdk.dts
@@ -0,0 +1,10 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2021 NXP
+ */
+
+#include "imx8mm-evk.dts"
+
+&fec1 {
+	compatible = "fsl,imx8mm-fec-uio";
+};
-- 
2.17.1

Re: [PATCH 5/5] arm64: dts: imx8mm-evk-dpdk: dts for fec-uio driver

[Andrew Lunn]

On Wed, Nov 10, 2021 at 11:18:38AM +0530, Apeksha Gupta wrote:
> New DTS file 'imx8mm-evk-dpdk.dts' is added to support fec
> ethernet device detection in user space DPDK application via
> fec-uio driver.

Isn't DPDK in its own parallel universe? It is not in mainline?  A
quick grep of drivers/net does not find anything. I think you are
submitting this to the wrong universe. You should be using XDP with
mainline.

	   Andrew

Re: [PATCH 0/5] drivers/net: add NXP FEC-UIO driver

[Michael Walle]

[linux-devel@linux.nxdi.nxp.com doesn't exist in the public, which is quite
annoying when you reply to an email and the MTA doesn't accept it due to
invalid domains]

> This patch series introduce the fec-uio driver, supported for the inbuilt
> NIC found in the NXP i.MX 8M Mini SoC. Basic hardware initialization is
> performed in kernel via userspace input/output(UIO) to support FEC
> ethernet device detection in user space.

Could you elaborate for what this driver is needed? Doesn't the imx8mm
already have a network driver? What is the difference between them?

As a user, I couldn't find what this is all about, neither in this
commit message nor in the Kconfig help text.

> Userspace PMD uses standard UIO interface to access kernel for PHY
> initialisation and for mapping the allocated memory of register &
> buffer descriptor with DPDK which gives access to non-cacheable memory
> for buffer descriptor.
> 
> Module fec-uio.ko will get generated.
> imx8mm-evk-dpdk.dtb is required to support fec-uio driver.

-michael