[v14,8/8] soc: amd: Add support for AMD Pensando SoC Controller

  The Pensando SoC controller is a SPI connected companion device
that is present in all Pensando SoC board designs.  The essential
board management registers are accessed on chip select 0 with
board mgmt IO support accessed using additional chip selects.

Signed-off-by: Brad Larson <blarson@amd.com>
---

v14 changes:
- Save 8 bytes of code size by swapping spi_device and reset_controller_dev
  in penctrl_device
- Code simplification and clarity from review inputs
- Set penctrl_spi_driver.driver.name to match compatible pensando-elba-ctrl
- Remove unused include in amd-pensando-ctrl.h
- Rebase to linux-next 6.4.0-rc1 class_create() API change
 
v13 changes:
- Update include list in pensando-ctrl.c
- Change variable spi_dev to spi throughout
- Removed unneeded variable initialization, simplification of
  error checks, remove extra castings, and use dev_err_probe()
- Sort the includes in amd-pensando-ctrl.h
- Updates to cleanup if there is an error in penctrl_spi_probe()

v12 changes:
- Fix gcc-12.1.0 warning:
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/oe-kbuild-all/202303120925.SxLjwOd2-lkp@intel.com/

v11 changes:
- Fix the compatible to be specific 'amd,pensando-elba-ctrl'

v10 changes:
- Different driver implementation specific to this Pensando controller device.
- Moved to soc/amd directory under new name based on guidance.  This driver is
  of no use to any design other than all Pensando SoC based cards.
- Removed use of builtin_driver, can be built as a module.

v9 changes:
- Previously patch 14/17
- After the change to the device tree node and squashing
  reset-cells into the parent simplified this to not use
  any MFD API and move it to drivers/spi/pensando-sr.c.
- Change the naming to remove elba since this driver is common
  for all Pensando SoC designs .
- Default yes SPI_PENSANDO_SR for ARCH_PENSANDO

---
 drivers/soc/Kconfig                    |   1 +
 drivers/soc/Makefile                   |   1 +
 drivers/soc/amd/Kconfig                |  16 ++
 drivers/soc/amd/Makefile               |   2 +
 drivers/soc/amd/pensando-ctrl.c        | 368 +++++++++++++++++++++++++
 include/uapi/linux/amd-pensando-ctrl.h |  29 ++
 6 files changed, 417 insertions(+)
 create mode 100644 drivers/soc/amd/Kconfig
 create mode 100644 drivers/soc/amd/Makefile
 create mode 100644 drivers/soc/amd/pensando-ctrl.c
 create mode 100644 include/uapi/linux/amd-pensando-ctrl.h
  

On Mon, May 15, 2023 at 9:18 PM Brad Larson <blarson@amd.com> wrote:
>
> The Pensando SoC controller is a SPI connected companion device
> that is present in all Pensando SoC board designs.  The essential
> board management registers are accessed on chip select 0 with
> board mgmt IO support accessed using additional chip selects.

...

> +#include <linux/cdev.h>
> +#include <linux/device.h>
> +#include <linux/err.h>
> +#include <linux/init.h>
> +#include <linux/mod_devicetable.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>

> +#include <linux/of.h>

Unneeded inclusion.

> +#include <linux/reset-controller.h>
> +#include <linux/spi/spi.h>

...


> +       u8 tx_buf[PENCTRL_MAX_MSG_LEN];
> +       u8 rx_buf[PENCTRL_MAX_MSG_LEN];

Does it need to be DMA-capable?

...

> +       spi->chip_select = current_cs;
> +       spi->cs_gpiod = spi->controller->cs_gpiods[current_cs];

Nowadays these require API calls instead of direct assignments.

...

> +static int penctrl_release(struct inode *inode, struct file *filp)
> +{
> +       filp->private_data = NULL;
> +       return 0;
> +}

Is it possible to unload the module without releasing the device node?

...

> +       u8 txbuf[3];
> +       u8 rxbuf[1];

Same question about DMA.

...

> +       ret = spi_sync(spi, &m);

> +       if (ret == 0)
> +               *val = rxbuf[0];
> +
> +       return ret;

Can also be written in more usual way:

  if (ret)
    return ret;
  ...
  return 0;

...

> +       u8 txbuf[4];

DMA?

...

> +       spi->chip_select = 0;
> +       spi->cs_gpiod = spi->controller->cs_gpiods[0];

Setter APIs.

...

> +       spi->chip_select = 0;
> +       spi->cs_gpiod = spi->controller->cs_gpiods[0];

Ditto.


> +       ret = device_property_read_u32(spi->dev.parent, "num-cs", &num_cs);
> +       if (ret)
> +               return dev_err_probe(&spi->dev, ret,
> +                                    "number of chip-selects not defined\n");

Hmm... Shouldn't SPI core take care of this in a generic way? Yes, I
understand that you need the number for the allocation, but I would
expect something like spi_fw_get_num_cs() to exist (seems not?).

...

> +       penctrl->rcdev.of_node = spi->dev.of_node;

Use device_set_node(). It helps to modify the data types beneath.

--
With Best Regards,
Andy Shevchenko
  

> -----Original Message-----
> From: Brad Larson <blarson@amd.com>
> Sent: Monday, May 15, 2023 11:46 PM
> To: linux-arm-kernel@lists.infradead.org
> Cc: linux-kernel@vger.kernel.org; linux-mmc@vger.kernel.org; linux-
> spi@vger.kernel.org; adrian.hunter@intel.com; alcooperx@gmail.com;
> andy.shevchenko@gmail.com; arnd@arndb.de; Larson, Bradley
> <Bradley.Larson@amd.com>; brendan.higgins@linux.dev;
> briannorris@chromium.org; catalin.marinas@arm.com;
> conor+dt@kernel.org; davidgow@google.com; gsomlo@gmail.com;
> gerg@linux-m68k.org; hal.feng@starfivetech.com; hasegawa-
> hitomi@fujitsu.com; j.neuschaefer@gmx.net; joel@jms.id.au;
> kernel@esmil.dk; krzk@kernel.org; krzysztof.kozlowski+dt@linaro.org;
> lee@kernel.org; lee.jones@linaro.org; broonie@kernel.org;
> p.zabel@pengutronix.de; rdunlap@infradead.org; robh+dt@kernel.org;
> samuel@sholland.org; fancer.lancer@gmail.com;
> skhan@linuxfoundation.org; Suthikulpanit, Suravee
> <Suravee.Suthikulpanit@amd.com>; Lendacky, Thomas
> <Thomas.Lendacky@amd.com>; tonyhuang.sunplus@gmail.com;
> ulf.hansson@linaro.org; vaishnav.a@ti.com; walker.chen@starfivetech.com;
> will@kernel.org; zhuyinbo@loongson.cn; devicetree@vger.kernel.org
> Subject: [PATCH v14 8/8] soc: amd: Add support for AMD Pensando SoC
> Controller
> 
> The Pensando SoC controller is a SPI connected companion device that is
> present in all Pensando SoC board designs.  The essential board management
> registers are accessed on chip select 0 with board mgmt IO support accessed
> using additional chip selects.
> 
> Signed-off-by: Brad Larson <blarson@amd.com>
> ---
> 
> v14 changes:
> - Save 8 bytes of code size by swapping spi_device and reset_controller_dev
>   in penctrl_device
> - Code simplification and clarity from review inputs
> - Set penctrl_spi_driver.driver.name to match compatible pensando-elba-ctrl
> - Remove unused include in amd-pensando-ctrl.h
> - Rebase to linux-next 6.4.0-rc1 class_create() API change
> 
> v13 changes:
> - Update include list in pensando-ctrl.c
> - Change variable spi_dev to spi throughout
> - Removed unneeded variable initialization, simplification of
>   error checks, remove extra castings, and use dev_err_probe()
> - Sort the includes in amd-pensando-ctrl.h
> - Updates to cleanup if there is an error in penctrl_spi_probe()
> 
> v12 changes:
> - Fix gcc-12.1.0 warning:
> Reported-by: kernel test robot <lkp@intel.com>
> Link: https://lore.kernel.org/oe-kbuild-all/202303120925.SxLjwOd2-
> lkp@intel.com/
> 
> v11 changes:
> - Fix the compatible to be specific 'amd,pensando-elba-ctrl'
> 
> v10 changes:
> - Different driver implementation specific to this Pensando controller device.
> - Moved to soc/amd directory under new name based on guidance.  This
> driver is
>   of no use to any design other than all Pensando SoC based cards.
> - Removed use of builtin_driver, can be built as a module.
> 
> v9 changes:
> - Previously patch 14/17
> - After the change to the device tree node and squashing
>   reset-cells into the parent simplified this to not use
>   any MFD API and move it to drivers/spi/pensando-sr.c.
> - Change the naming to remove elba since this driver is common
>   for all Pensando SoC designs .
> - Default yes SPI_PENSANDO_SR for ARCH_PENSANDO
> 
> ---
>  drivers/soc/Kconfig                    |   1 +
>  drivers/soc/Makefile                   |   1 +
>  drivers/soc/amd/Kconfig                |  16 ++
>  drivers/soc/amd/Makefile               |   2 +
>  drivers/soc/amd/pensando-ctrl.c        | 368 +++++++++++++++++++++++++
>  include/uapi/linux/amd-pensando-ctrl.h |  29 ++
>  6 files changed, 417 insertions(+)
>  create mode 100644 drivers/soc/amd/Kconfig  create mode 100644
> drivers/soc/amd/Makefile  create mode 100644 drivers/soc/amd/pensando-
> ctrl.c  create mode 100644 include/uapi/linux/amd-pensando-ctrl.h
> 
> diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig index
> 4e176280113a..9e023f74e47c 100644
> --- a/drivers/soc/Kconfig
> +++ b/drivers/soc/Kconfig
> @@ -2,6 +2,7 @@
>  menu "SOC (System On Chip) specific Drivers"
> 
>  source "drivers/soc/actions/Kconfig"
> +source "drivers/soc/amd/Kconfig"
>  source "drivers/soc/amlogic/Kconfig"
>  source "drivers/soc/apple/Kconfig"
>  source "drivers/soc/aspeed/Kconfig"
> diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile index
> 3b0f9fb3b5c8..8914530f2721 100644
> --- a/drivers/soc/Makefile
> +++ b/drivers/soc/Makefile
> @@ -4,6 +4,7 @@
>  #
> 
>  obj-$(CONFIG_ARCH_ACTIONS)	+= actions/
> +obj-y				+= amd/
>  obj-y				+= apple/
>  obj-y				+= aspeed/
>  obj-$(CONFIG_ARCH_AT91)		+= atmel/
> diff --git a/drivers/soc/amd/Kconfig b/drivers/soc/amd/Kconfig new file
> mode 100644 index 000000000000..011d5339d14e
> --- /dev/null
> +++ b/drivers/soc/amd/Kconfig
> @@ -0,0 +1,16 @@
> +# SPDX-License-Identifier: GPL-2.0-only menu "AMD Pensando SoC drivers"
> +
> +config AMD_PENSANDO_CTRL
> +	tristate "AMD Pensando SoC Controller"
> +	depends on SPI_MASTER=y
> +	depends on (ARCH_PENSANDO && OF) || COMPILE_TEST
> +	default ARCH_PENSANDO
> +	select REGMAP_SPI
> +	select MFD_SYSCON
> +	help
> +	  Enables AMD Pensando SoC controller device support.  This is a SPI
> +	  attached companion device in all Pensando SoC board designs which
> +	  provides essential board control/status registers and management
> IO
> +	  support.
> +endmenu
> diff --git a/drivers/soc/amd/Makefile b/drivers/soc/amd/Makefile new file
> mode 100644 index 000000000000..a2de0424f68d
> --- /dev/null
> +++ b/drivers/soc/amd/Makefile
> @@ -0,0 +1,2 @@
> +# SPDX-License-Identifier: GPL-2.0-only
> +obj-$(CONFIG_AMD_PENSANDO_CTRL)	+= pensando-ctrl.o
> diff --git a/drivers/soc/amd/pensando-ctrl.c b/drivers/soc/amd/pensando-
> ctrl.c new file mode 100644 index 000000000000..a7ddd181dfe8
> --- /dev/null
> +++ b/drivers/soc/amd/pensando-ctrl.c
> @@ -0,0 +1,368 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * AMD Pensando SoC Controller
> + *
> + * Userspace interface and reset driver support for SPI connected
> +Pensando SoC
> + * controller device.  This device is present in all Pensando SoC
> +designs and
> + * contains board control/status registers and management IO support.
> + *
> + * Copyright 2023 Advanced Micro Devices, Inc.
> + */
> +
> +#include <linux/cdev.h>
> +#include <linux/device.h>
> +#include <linux/err.h>
> +#include <linux/init.h>
> +#include <linux/mod_devicetable.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>
> +#include <linux/of.h>
> +#include <linux/reset-controller.h>
> +#include <linux/spi/spi.h>
> +
> +#include <linux/amd-pensando-ctrl.h>
> +
> +struct penctrl_device {
> +	struct reset_controller_dev rcdev;
> +	struct spi_device *spi;
> +};
> +
> +static DEFINE_MUTEX(spi_lock);
> +static dev_t penctrl_devt;
> +static struct penctrl_device *penctrl;
> +static struct class *penctrl_class;
> +
> +static long
> +penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) {
> +	void __user *in_arg = (void __user *)arg;
> +	struct penctrl_device *penctrl;
> +	u8 tx_buf[PENCTRL_MAX_MSG_LEN];
> +	u8 rx_buf[PENCTRL_MAX_MSG_LEN];
> +	struct spi_transfer t[2] = {};
> +	struct penctrl_spi_xfer *msg;
> +	struct spi_device *spi;
> +	unsigned int num_msgs;
> +	struct spi_message m;
> +	u32 size;
> +	int ret;
> +
> +	/* Check for a valid command */
> +	if (_IOC_TYPE(cmd) != PENCTRL_IOC_MAGIC)
> +		return -ENOTTY;
> +
> +	if (_IOC_NR(cmd) > PENCTRL_IOC_MAXNR)
> +		return -ENOTTY;
> +
> +	if (((_IOC_DIR(cmd) & _IOC_READ)) && !access_ok(in_arg,
> _IOC_SIZE(cmd)))
> +		return -EFAULT;
> +
> +	if (((_IOC_DIR(cmd) & _IOC_WRITE)) && !access_ok(in_arg,
> _IOC_SIZE(cmd)))
> +		return -EFAULT;
> +
> +	/* Get a reference to the SPI device */
> +	penctrl = filp->private_data;
> +	if (!penctrl)
> +		return -ESHUTDOWN;
> +
> +	spi = spi_dev_get(penctrl->spi);
> +	if (!spi)
> +		return -ESHUTDOWN;
> +
> +	/* Verify and prepare SPI message */
> +	size = _IOC_SIZE(cmd);
> +	num_msgs = size / sizeof(struct penctrl_spi_xfer);
> +	if (num_msgs > 2 || size == 0 || size % sizeof(struct
> penctrl_spi_xfer)) {
> +		ret = -EINVAL;
> +		goto out_unlock;
> +	}
> +	msg = memdup_user((struct penctrl_spi_xfer *)arg, size);
> +	if (IS_ERR(msg)) {
> +		ret = PTR_ERR(msg);
> +		goto out_unlock;
> +	}
> +	if (msg->len > PENCTRL_MAX_MSG_LEN) {
> +		ret = -EINVAL;
> +		goto out_unlock;
> +	}
> +
> +	t[0].tx_buf = tx_buf;
> +	t[0].len = msg->len;
> +	if (copy_from_user(tx_buf, (void __user *)msg->tx_buf, msg->len))
> {
> +		ret = -EFAULT;
> +		goto out_unlock;
> +	}
> +	if (num_msgs > 1) {
> +		msg++;
> +		if (msg->len > PENCTRL_MAX_MSG_LEN) {
> +			ret = -EINVAL;
> +			goto out_unlock;
> +		}
> +		t[1].rx_buf = rx_buf;
> +		t[1].len = msg->len;
> +	}
> +	spi_message_init_with_transfers(&m, t, num_msgs);
> +
> +	/* Perform the transfer */
> +	mutex_lock(&spi_lock);
> +	ret = spi_sync(spi, &m);
> +	mutex_unlock(&spi_lock);
> +
> +	if (ret || (num_msgs == 1))
> +		goto out_unlock;
> +
> +	if (copy_to_user((void __user *)msg->rx_buf, rx_buf, msg->len))
> +		ret = -EFAULT;
> +
> +out_unlock:
> +	spi_dev_put(spi);
> +	return ret;
> +}
> +
> +static int penctrl_open(struct inode *inode, struct file *filp) {
> +	struct spi_device *spi;
> +	u8 current_cs;
> +
> +	filp->private_data = penctrl;
> +	current_cs = iminor(inode);
> +	spi = penctrl->spi;
> +	spi->chip_select = current_cs;

New set/get APIs for accessing spi->chip_select were introduced by 
https://github.com/torvalds/linux/commit/303feb3cc06ac0665d0ee9c1414941200e60e8a3
please use these APIs instead of accessing spi->chip_select dricetly. 

> +	spi->cs_gpiod = spi->controller->cs_gpiods[current_cs];
> +	spi_setup(spi);
> +	return stream_open(inode, filp);
> +}
> +
> +static int penctrl_release(struct inode *inode, struct file *filp) {
> +	filp->private_data = NULL;
> +	return 0;
> +}
> +
> +static const struct file_operations penctrl_fops = {
> +	.owner		= THIS_MODULE,
> +	.unlocked_ioctl = penctrl_ioctl,
> +	.open		= penctrl_open,
> +	.release	= penctrl_release,
> +	.llseek		= no_llseek,
> +};
> +
> +static int penctrl_regs_read(struct penctrl_device *penctrl, u32 reg,
> +u32 *val) {
> +	struct spi_device *spi = penctrl->spi;
> +	struct spi_transfer t[2] = {};
> +	struct spi_message m;
> +	u8 txbuf[3];
> +	u8 rxbuf[1];
> +	int ret;
> +
> +	txbuf[0] = PENCTRL_SPI_CMD_REGRD;
> +	txbuf[1] = reg;
> +	txbuf[2] = 0;
> +	t[0].tx_buf = txbuf;
> +	t[0].len = sizeof(txbuf);
> +
> +	rxbuf[0] = 0;
> +	t[1].rx_buf = rxbuf;
> +	t[1].len = sizeof(rxbuf);
> +
> +	spi_message_init_with_transfers(&m, t, ARRAY_SIZE(t));
> +	ret = spi_sync(spi, &m);
> +	if (ret == 0)
> +		*val = rxbuf[0];
> +
> +	return ret;
> +}
> +
> +static int penctrl_regs_write(struct penctrl_device *penctrl, u32 reg,
> +u32 val) {
> +	struct spi_device *spi = penctrl->spi;
> +	struct spi_transfer t = {};
> +	struct spi_message m;
> +	u8 txbuf[4];
> +
> +	txbuf[0] = PENCTRL_SPI_CMD_REGWR;
> +	txbuf[1] = reg;
> +	txbuf[2] = val;
> +	txbuf[3] = 0;
> +
> +	t.tx_buf = txbuf;
> +	t.len = sizeof(txbuf);
> +	spi_message_init_with_transfers(&m, &t, 1);
> +	return spi_sync(spi, &m);
> +}
> +
> +static int penctrl_reset_assert(struct reset_controller_dev *rcdev,
> +				unsigned long id)
> +{
> +	struct penctrl_device *penctrl =
> +		container_of(rcdev, struct penctrl_device, rcdev);
> +	struct spi_device *spi = penctrl->spi;
> +	unsigned int val;
> +	int ret;
> +
> +	mutex_lock(&spi_lock);
> +	spi->chip_select = 0;

Same here.

> +	spi->cs_gpiod = spi->controller->cs_gpiods[0];
> +	spi_setup(spi);
> +	ret = penctrl_regs_read(penctrl, PENCTRL_REG_CTRL0, &val);
> +	if (ret) {
> +		dev_err(&spi->dev, "error reading ctrl0 reg\n");
> +		goto out_unlock;
> +	}
> +
> +	val |= BIT(6);
> +	ret = penctrl_regs_write(penctrl, PENCTRL_REG_CTRL0, val);
> +	if (ret)
> +		dev_err(&spi->dev, "error writing ctrl0 reg\n");
> +
> +out_unlock:
> +	mutex_unlock(&spi_lock);
> +	return ret;
> +}
> +
> +static int penctrl_reset_deassert(struct reset_controller_dev *rcdev,
> +				  unsigned long id)
> +{
> +	struct penctrl_device *penctrl =
> +		container_of(rcdev, struct penctrl_device, rcdev);
> +	struct spi_device *spi = penctrl->spi;
> +	unsigned int val;
> +	int ret;
> +
> +	mutex_lock(&spi_lock);
> +	spi->chip_select = 0;

Same here.

Regards,
Amit

> +	spi->cs_gpiod = spi->controller->cs_gpiods[0];
> +	spi_setup(spi);
> +	ret = penctrl_regs_read(penctrl, PENCTRL_REG_CTRL0, &val);
> +	if (ret) {
> +		dev_err(&spi->dev, "error reading ctrl0 reg\n");
> +		goto out_unlock;
> +	}
> +
> +	val &= ~BIT(6);
> +	ret = penctrl_regs_write(penctrl, PENCTRL_REG_CTRL0, val);
> +	if (ret)
> +		dev_err(&spi->dev, "error writing ctrl0 reg\n");
> +
> +out_unlock:
> +	mutex_unlock(&spi_lock);
> +	return ret;
> +}
> +
> +static const struct reset_control_ops penctrl_reset_ops = {
> +	.assert = penctrl_reset_assert,
> +	.deassert = penctrl_reset_deassert,
> +};
> +
> +static int penctrl_spi_probe(struct spi_device *spi) {
> +	struct device *dev;
> +	struct cdev *cdev;
> +	u32 num_cs;
> +	int ret;
> +	u32 cs;
> +
> +	ret = device_property_read_u32(spi->dev.parent, "num-cs",
> &num_cs);
> +	if (ret)
> +		return dev_err_probe(&spi->dev, ret,
> +				     "number of chip-selects not defined\n");
> +
> +	ret = alloc_chrdev_region(&penctrl_devt, 0, num_cs, "penctrl");
> +	if (ret)
> +		return dev_err_probe(&spi->dev, ret,
> +				     "failed to alloc chrdev region\n");
> +
> +	penctrl_class = class_create("penctrl");
> +	if (IS_ERR(penctrl_class)) {
> +		ret = dev_err_probe(&spi->dev, PTR_ERR(penctrl_class),
> +				    "failed to create class\n");
> +		goto unregister_chrdev;
> +	}
> +
> +	cdev = cdev_alloc();
> +	if (!cdev) {
> +		ret = dev_err_probe(&spi->dev, -ENOMEM,
> +				    "allocation of cdev failed\n");
> +		goto destroy_class;
> +	}
> +	cdev->owner = THIS_MODULE;
> +	cdev_init(cdev, &penctrl_fops);
> +
> +	ret = cdev_add(cdev, penctrl_devt, num_cs);
> +	if (ret) {
> +		ret = dev_err_probe(&spi->dev, ret,
> +				    "register of cdev failed\n");
> +		goto free_cdev;
> +	}
> +
> +	/* Allocate driver data */
> +	penctrl = kzalloc(sizeof(*penctrl), GFP_KERNEL);
> +	if (!penctrl) {
> +		ret = -ENOMEM;
> +		goto free_cdev;
> +	}
> +	penctrl->spi = spi;
> +	mutex_init(&spi_lock);
> +
> +	/* Create a device for each chip select */
> +	for (cs = 0; cs < num_cs; cs++) {
> +		dev = device_create(penctrl_class,
> +				    &spi->dev,
> +				    MKDEV(MAJOR(penctrl_devt), cs),
> +				    penctrl,
> +				    "penctrl0.%d",
> +				    cs);
> +		if (IS_ERR(dev)) {
> +			ret = dev_err_probe(&spi->dev, PTR_ERR(dev),
> +					    "error creating device\n");
> +			goto destroy_device;
> +		}
> +		dev_dbg(&spi->dev, "created device major %u, minor
> %d\n",
> +			MAJOR(penctrl_devt), cs);
> +	}
> +
> +	/* Register reset controller */
> +	penctrl->rcdev.dev = &spi->dev;
> +	penctrl->rcdev.ops = &penctrl_reset_ops;
> +	penctrl->rcdev.owner = THIS_MODULE;
> +	penctrl->rcdev.of_node = spi->dev.of_node;
> +	penctrl->rcdev.nr_resets = 1;
> +
> +	ret = reset_controller_register(&penctrl->rcdev);
> +	if (ret)
> +		return dev_err_probe(&spi->dev, ret,
> +				     "failed to register reset controller\n");
> +	return 0;
> +
> +destroy_device:
> +	for (cs = 0; cs < num_cs; cs++)
> +		device_destroy(penctrl_class,
> MKDEV(MAJOR(penctrl_devt), cs));
> +	kfree(penctrl);
> +free_cdev:
> +	cdev_del(cdev);
> +destroy_class:
> +	class_destroy(penctrl_class);
> +unregister_chrdev:
> +	unregister_chrdev(MAJOR(penctrl_devt), "penctrl");
> +
> +	return ret;
> +}
> +
> +static const struct of_device_id penctrl_dt_match[] = {
> +	{ .compatible = "amd,pensando-elba-ctrl" },
> +	{ /* sentinel */ }
> +};
> +
> +static struct spi_driver penctrl_spi_driver = {
> +	.probe = penctrl_spi_probe,
> +	.driver = {
> +		.name = "pensando-elba-ctrl",
> +		.of_match_table = penctrl_dt_match,
> +	},
> +};
> +module_spi_driver(penctrl_spi_driver);
> +
> +MODULE_AUTHOR("Brad Larson <blarson@amd.com>");
> MODULE_DESCRIPTION("AMD
> +Pensando SoC Controller via SPI"); MODULE_LICENSE("GPL");
> diff --git a/include/uapi/linux/amd-pensando-ctrl.h
> b/include/uapi/linux/amd-pensando-ctrl.h
> new file mode 100644
> index 000000000000..e5f9f0dfe146
> --- /dev/null
> +++ b/include/uapi/linux/amd-pensando-ctrl.h
> @@ -0,0 +1,29 @@
> +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
> +/*
> + * Userspace interface for /dev/penctrl
> + *
> + * This file can be used by applications that need to communicate
> + * with the AMD Pensando SoC controller device via the ioctl interface.
> + */
> +#ifndef _UAPI_LINUX_AMD_PENSANDO_CTRL_H #define
> +_UAPI_LINUX_AMD_PENSANDO_CTRL_H
> +
> +#include <linux/types.h>
> +
> +#define PENCTRL_SPI_CMD_REGRD	0x0b
> +#define PENCTRL_SPI_CMD_REGWR	0x02
> +#define PENCTRL_IOC_MAGIC	'k'
> +#define PENCTRL_IOC_MAXNR	0
> +#define PENCTRL_MAX_MSG_LEN	16
> +#define PENCTRL_MAX_REG		0xff
> +#define PENCTRL_REG_CTRL0	0x10
> +
> +struct penctrl_spi_xfer {
> +	__u64 tx_buf;
> +	__u64 rx_buf;
> +	__u32 len;
> +	__u32 speed_hz;
> +	__u64 compat;
> +};
> +
> +#endif /* _UAPI_LINUX_AMD_PENSANDO_CTRL_H */
> --
> 2.17.1
  

On 5/16/23 07:19, Mahapatra, Amit Kumar wrote:
> 
> 
>> -----Original Message-----
>> From: Brad Larson <blarson@amd.com>
>> Sent: Monday, May 15, 2023 11:46 PM
>> To: linux-arm-kernel@lists.infradead.org
>> Cc: linux-kernel@vger.kernel.org; linux-mmc@vger.kernel.org; linux-
>> spi@vger.kernel.org; adrian.hunter@intel.com; alcooperx@gmail.com;
>> andy.shevchenko@gmail.com; arnd@arndb.de; Larson, Bradley
>> <Bradley.Larson@amd.com>; brendan.higgins@linux.dev;
>> briannorris@chromium.org; catalin.marinas@arm.com;
>> conor+dt@kernel.org; davidgow@google.com; gsomlo@gmail.com;
>> gerg@linux-m68k.org; hal.feng@starfivetech.com; hasegawa-
>> hitomi@fujitsu.com; j.neuschaefer@gmx.net; joel@jms.id.au;
>> kernel@esmil.dk; krzk@kernel.org; krzysztof.kozlowski+dt@linaro.org;
>> lee@kernel.org; lee.jones@linaro.org; broonie@kernel.org;
>> p.zabel@pengutronix.de; rdunlap@infradead.org; robh+dt@kernel.org;
>> samuel@sholland.org; fancer.lancer@gmail.com;
>> skhan@linuxfoundation.org; Suthikulpanit, Suravee
>> <Suravee.Suthikulpanit@amd.com>; Lendacky, Thomas
>> <Thomas.Lendacky@amd.com>; tonyhuang.sunplus@gmail.com;
>> ulf.hansson@linaro.org; vaishnav.a@ti.com; walker.chen@starfivetech.com;
>> will@kernel.org; zhuyinbo@loongson.cn; devicetree@vger.kernel.org
>> Subject: [PATCH v14 8/8] soc: amd: Add support for AMD Pensando SoC
>> Controller
>>
>> The Pensando SoC controller is a SPI connected companion device that is
>> present in all Pensando SoC board designs.  The essential board management
>> registers are accessed on chip select 0 with board mgmt IO support accessed
>> using additional chip selects.
>>
>> Signed-off-by: Brad Larson <blarson@amd.com>
>> ---
>>
>> v14 changes:
>> - Save 8 bytes of code size by swapping spi_device and reset_controller_dev
>>    in penctrl_device
>> - Code simplification and clarity from review inputs
>> - Set penctrl_spi_driver.driver.name to match compatible pensando-elba-ctrl
>> - Remove unused include in amd-pensando-ctrl.h
>> - Rebase to linux-next 6.4.0-rc1 class_create() API change
>>
>> v13 changes:
>> - Update include list in pensando-ctrl.c
>> - Change variable spi_dev to spi throughout
>> - Removed unneeded variable initialization, simplification of
>>    error checks, remove extra castings, and use dev_err_probe()
>> - Sort the includes in amd-pensando-ctrl.h
>> - Updates to cleanup if there is an error in penctrl_spi_probe()
>>
>> v12 changes:
>> - Fix gcc-12.1.0 warning:
>> Reported-by: kernel test robot <lkp@intel.com>
>> Link: https://lore.kernel.org/oe-kbuild-all/202303120925.SxLjwOd2-
>> lkp@intel.com/
>>
>> v11 changes:
>> - Fix the compatible to be specific 'amd,pensando-elba-ctrl'
>>
>> v10 changes:
>> - Different driver implementation specific to this Pensando controller device.
>> - Moved to soc/amd directory under new name based on guidance.  This
>> driver is
>>    of no use to any design other than all Pensando SoC based cards.
>> - Removed use of builtin_driver, can be built as a module.
>>
>> v9 changes:
>> - Previously patch 14/17
>> - After the change to the device tree node and squashing
>>    reset-cells into the parent simplified this to not use
>>    any MFD API and move it to drivers/spi/pensando-sr.c.
>> - Change the naming to remove elba since this driver is common
>>    for all Pensando SoC designs .
>> - Default yes SPI_PENSANDO_SR for ARCH_PENSANDO
>>
>> ---
>>   drivers/soc/Kconfig                    |   1 +
>>   drivers/soc/Makefile                   |   1 +
>>   drivers/soc/amd/Kconfig                |  16 ++
>>   drivers/soc/amd/Makefile               |   2 +
>>   drivers/soc/amd/pensando-ctrl.c        | 368 +++++++++++++++++++++++++
>>   include/uapi/linux/amd-pensando-ctrl.h |  29 ++
>>   6 files changed, 417 insertions(+)
>>   create mode 100644 drivers/soc/amd/Kconfig  create mode 100644
>> drivers/soc/amd/Makefile  create mode 100644 drivers/soc/amd/pensando-
>> ctrl.c  create mode 100644 include/uapi/linux/amd-pensando-ctrl.h
>>
>> diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig index
>> 4e176280113a..9e023f74e47c 100644
>> --- a/drivers/soc/Kconfig
>> +++ b/drivers/soc/Kconfig
>> @@ -2,6 +2,7 @@
>>   menu "SOC (System On Chip) specific Drivers"
>>
>>   source "drivers/soc/actions/Kconfig"
>> +source "drivers/soc/amd/Kconfig"
>>   source "drivers/soc/amlogic/Kconfig"
>>   source "drivers/soc/apple/Kconfig"
>>   source "drivers/soc/aspeed/Kconfig"
>> diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile index
>> 3b0f9fb3b5c8..8914530f2721 100644
>> --- a/drivers/soc/Makefile
>> +++ b/drivers/soc/Makefile
>> @@ -4,6 +4,7 @@
>>   #
>>
>>   obj-$(CONFIG_ARCH_ACTIONS)	+= actions/
>> +obj-y				+= amd/
>>   obj-y				+= apple/
>>   obj-y				+= aspeed/
>>   obj-$(CONFIG_ARCH_AT91)		+= atmel/
>> diff --git a/drivers/soc/amd/Kconfig b/drivers/soc/amd/Kconfig new file
>> mode 100644 index 000000000000..011d5339d14e
>> --- /dev/null
>> +++ b/drivers/soc/amd/Kconfig
>> @@ -0,0 +1,16 @@
>> +# SPDX-License-Identifier: GPL-2.0-only menu "AMD Pensando SoC drivers"
>> +
>> +config AMD_PENSANDO_CTRL
>> +	tristate "AMD Pensando SoC Controller"
>> +	depends on SPI_MASTER=y
>> +	depends on (ARCH_PENSANDO && OF) || COMPILE_TEST
>> +	default ARCH_PENSANDO
>> +	select REGMAP_SPI
>> +	select MFD_SYSCON
>> +	help
>> +	  Enables AMD Pensando SoC controller device support.  This is a SPI
>> +	  attached companion device in all Pensando SoC board designs which
>> +	  provides essential board control/status registers and management
>> IO
>> +	  support.
>> +endmenu
>> diff --git a/drivers/soc/amd/Makefile b/drivers/soc/amd/Makefile new file
>> mode 100644 index 000000000000..a2de0424f68d
>> --- /dev/null
>> +++ b/drivers/soc/amd/Makefile
>> @@ -0,0 +1,2 @@
>> +# SPDX-License-Identifier: GPL-2.0-only
>> +obj-$(CONFIG_AMD_PENSANDO_CTRL)	+= pensando-ctrl.o
>> diff --git a/drivers/soc/amd/pensando-ctrl.c b/drivers/soc/amd/pensando-
>> ctrl.c new file mode 100644 index 000000000000..a7ddd181dfe8
>> --- /dev/null
>> +++ b/drivers/soc/amd/pensando-ctrl.c
>> @@ -0,0 +1,368 @@
>> +// SPDX-License-Identifier: GPL-2.0-or-later
>> +/*
>> + * AMD Pensando SoC Controller
>> + *
>> + * Userspace interface and reset driver support for SPI connected
>> +Pensando SoC
>> + * controller device.  This device is present in all Pensando SoC
>> +designs and
>> + * contains board control/status registers and management IO support.
>> + *
>> + * Copyright 2023 Advanced Micro Devices, Inc.
>> + */
>> +
>> +#include <linux/cdev.h>
>> +#include <linux/device.h>
>> +#include <linux/err.h>
>> +#include <linux/init.h>
>> +#include <linux/mod_devicetable.h>
>> +#include <linux/module.h>
>> +#include <linux/mutex.h>
>> +#include <linux/of.h>
>> +#include <linux/reset-controller.h>
>> +#include <linux/spi/spi.h>
>> +
>> +#include <linux/amd-pensando-ctrl.h>
>> +
>> +struct penctrl_device {
>> +	struct reset_controller_dev rcdev;
>> +	struct spi_device *spi;
>> +};
>> +
>> +static DEFINE_MUTEX(spi_lock);
>> +static dev_t penctrl_devt;
>> +static struct penctrl_device *penctrl;
>> +static struct class *penctrl_class;
>> +
>> +static long
>> +penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) {
>> +	void __user *in_arg = (void __user *)arg;
>> +	struct penctrl_device *penctrl;
>> +	u8 tx_buf[PENCTRL_MAX_MSG_LEN];
>> +	u8 rx_buf[PENCTRL_MAX_MSG_LEN];
>> +	struct spi_transfer t[2] = {};
>> +	struct penctrl_spi_xfer *msg;
>> +	struct spi_device *spi;
>> +	unsigned int num_msgs;
>> +	struct spi_message m;
>> +	u32 size;
>> +	int ret;
>> +
>> +	/* Check for a valid command */
>> +	if (_IOC_TYPE(cmd) != PENCTRL_IOC_MAGIC)
>> +		return -ENOTTY;
>> +
>> +	if (_IOC_NR(cmd) > PENCTRL_IOC_MAXNR)
>> +		return -ENOTTY;
>> +
>> +	if (((_IOC_DIR(cmd) & _IOC_READ)) && !access_ok(in_arg,
>> _IOC_SIZE(cmd)))
>> +		return -EFAULT;
>> +
>> +	if (((_IOC_DIR(cmd) & _IOC_WRITE)) && !access_ok(in_arg,
>> _IOC_SIZE(cmd)))
>> +		return -EFAULT;
>> +
>> +	/* Get a reference to the SPI device */
>> +	penctrl = filp->private_data;
>> +	if (!penctrl)
>> +		return -ESHUTDOWN;
>> +
>> +	spi = spi_dev_get(penctrl->spi);
>> +	if (!spi)
>> +		return -ESHUTDOWN;
>> +
>> +	/* Verify and prepare SPI message */
>> +	size = _IOC_SIZE(cmd);
>> +	num_msgs = size / sizeof(struct penctrl_spi_xfer);
>> +	if (num_msgs > 2 || size == 0 || size % sizeof(struct
>> penctrl_spi_xfer)) {
>> +		ret = -EINVAL;
>> +		goto out_unlock;
>> +	}
>> +	msg = memdup_user((struct penctrl_spi_xfer *)arg, size);
>> +	if (IS_ERR(msg)) {
>> +		ret = PTR_ERR(msg);
>> +		goto out_unlock;
>> +	}
>> +	if (msg->len > PENCTRL_MAX_MSG_LEN) {
>> +		ret = -EINVAL;
>> +		goto out_unlock;
>> +	}
>> +
>> +	t[0].tx_buf = tx_buf;
>> +	t[0].len = msg->len;
>> +	if (copy_from_user(tx_buf, (void __user *)msg->tx_buf, msg->len))
>> {
>> +		ret = -EFAULT;
>> +		goto out_unlock;
>> +	}
>> +	if (num_msgs > 1) {
>> +		msg++;
>> +		if (msg->len > PENCTRL_MAX_MSG_LEN) {
>> +			ret = -EINVAL;
>> +			goto out_unlock;
>> +		}
>> +		t[1].rx_buf = rx_buf;
>> +		t[1].len = msg->len;
>> +	}
>> +	spi_message_init_with_transfers(&m, t, num_msgs);
>> +
>> +	/* Perform the transfer */
>> +	mutex_lock(&spi_lock);
>> +	ret = spi_sync(spi, &m);
>> +	mutex_unlock(&spi_lock);
>> +
>> +	if (ret || (num_msgs == 1))
>> +		goto out_unlock;
>> +
>> +	if (copy_to_user((void __user *)msg->rx_buf, rx_buf, msg->len))
>> +		ret = -EFAULT;
>> +
>> +out_unlock:
>> +	spi_dev_put(spi);
>> +	return ret;
>> +}
>> +
>> +static int penctrl_open(struct inode *inode, struct file *filp) {
>> +	struct spi_device *spi;
>> +	u8 current_cs;
>> +
>> +	filp->private_data = penctrl;
>> +	current_cs = iminor(inode);
>> +	spi = penctrl->spi;
>> +	spi->chip_select = current_cs;
> 
> New set/get APIs for accessing spi->chip_select were introduced by
> https://github.com/torvalds/linux/commit/303feb3cc06ac0665d0ee9c1414941200e60e8a3
> please use these APIs instead of accessing spi->chip_select dricetly.
> 
>> +	spi->cs_gpiod = spi->controller->cs_gpiods[current_cs];
>> +	spi_setup(spi);
>> +	return stream_open(inode, filp);
>> +}
>> +
>> +static int penctrl_release(struct inode *inode, struct file *filp) {
>> +	filp->private_data = NULL;
>> +	return 0;
>> +}
>> +
>> +static const struct file_operations penctrl_fops = {
>> +	.owner		= THIS_MODULE,
>> +	.unlocked_ioctl = penctrl_ioctl,
>> +	.open		= penctrl_open,
>> +	.release	= penctrl_release,
>> +	.llseek		= no_llseek,
>> +};
>> +
>> +static int penctrl_regs_read(struct penctrl_device *penctrl, u32 reg,
>> +u32 *val) {
>> +	struct spi_device *spi = penctrl->spi;
>> +	struct spi_transfer t[2] = {};
>> +	struct spi_message m;
>> +	u8 txbuf[3];
>> +	u8 rxbuf[1];
>> +	int ret;
>> +
>> +	txbuf[0] = PENCTRL_SPI_CMD_REGRD;
>> +	txbuf[1] = reg;
>> +	txbuf[2] = 0;
>> +	t[0].tx_buf = txbuf;
>> +	t[0].len = sizeof(txbuf);
>> +
>> +	rxbuf[0] = 0;
>> +	t[1].rx_buf = rxbuf;
>> +	t[1].len = sizeof(rxbuf);
>> +
>> +	spi_message_init_with_transfers(&m, t, ARRAY_SIZE(t));
>> +	ret = spi_sync(spi, &m);
>> +	if (ret == 0)
>> +		*val = rxbuf[0];
>> +
>> +	return ret;
>> +}
>> +
>> +static int penctrl_regs_write(struct penctrl_device *penctrl, u32 reg,
>> +u32 val) {
>> +	struct spi_device *spi = penctrl->spi;
>> +	struct spi_transfer t = {};
>> +	struct spi_message m;
>> +	u8 txbuf[4];
>> +
>> +	txbuf[0] = PENCTRL_SPI_CMD_REGWR;
>> +	txbuf[1] = reg;
>> +	txbuf[2] = val;
>> +	txbuf[3] = 0;
>> +
>> +	t.tx_buf = txbuf;
>> +	t.len = sizeof(txbuf);
>> +	spi_message_init_with_transfers(&m, &t, 1);
>> +	return spi_sync(spi, &m);
>> +}
>> +
>> +static int penctrl_reset_assert(struct reset_controller_dev *rcdev,
>> +				unsigned long id)
>> +{
>> +	struct penctrl_device *penctrl =
>> +		container_of(rcdev, struct penctrl_device, rcdev);
>> +	struct spi_device *spi = penctrl->spi;
>> +	unsigned int val;
>> +	int ret;
>> +
>> +	mutex_lock(&spi_lock);
>> +	spi->chip_select = 0;
> 
> Same here.
> 
>> +	spi->cs_gpiod = spi->controller->cs_gpiods[0];
>> +	spi_setup(spi);
>> +	ret = penctrl_regs_read(penctrl, PENCTRL_REG_CTRL0, &val);
>> +	if (ret) {
>> +		dev_err(&spi->dev, "error reading ctrl0 reg\n");
>> +		goto out_unlock;
>> +	}
>> +
>> +	val |= BIT(6);
>> +	ret = penctrl_regs_write(penctrl, PENCTRL_REG_CTRL0, val);
>> +	if (ret)
>> +		dev_err(&spi->dev, "error writing ctrl0 reg\n");
>> +
>> +out_unlock:
>> +	mutex_unlock(&spi_lock);
>> +	return ret;
>> +}
>> +
>> +static int penctrl_reset_deassert(struct reset_controller_dev *rcdev,
>> +				  unsigned long id)
>> +{
>> +	struct penctrl_device *penctrl =
>> +		container_of(rcdev, struct penctrl_device, rcdev);
>> +	struct spi_device *spi = penctrl->spi;
>> +	unsigned int val;
>> +	int ret;
>> +
>> +	mutex_lock(&spi_lock);
>> +	spi->chip_select = 0;
> 
> Same here.
> 
> Regards,
> Amit
> 
>> +	spi->cs_gpiod = spi->controller->cs_gpiods[0];
>> +	spi_setup(spi);
>> +	ret = penctrl_regs_read(penctrl, PENCTRL_REG_CTRL0, &val);
>> +	if (ret) {
>> +		dev_err(&spi->dev, "error reading ctrl0 reg\n");
>> +		goto out_unlock;
>> +	}
>> +
>> +	val &= ~BIT(6);
>> +	ret = penctrl_regs_write(penctrl, PENCTRL_REG_CTRL0, val);
>> +	if (ret)
>> +		dev_err(&spi->dev, "error writing ctrl0 reg\n");
>> +
>> +out_unlock:
>> +	mutex_unlock(&spi_lock);
>> +	return ret;
>> +}
>> +
>> +static const struct reset_control_ops penctrl_reset_ops = {
>> +	.assert = penctrl_reset_assert,
>> +	.deassert = penctrl_reset_deassert,
>> +};
>> +
>> +static int penctrl_spi_probe(struct spi_device *spi) {
>> +	struct device *dev;
>> +	struct cdev *cdev;
>> +	u32 num_cs;
>> +	int ret;
>> +	u32 cs;
>> +
>> +	ret = device_property_read_u32(spi->dev.parent, "num-cs",
>> &num_cs);
>> +	if (ret)
>> +		return dev_err_probe(&spi->dev, ret,
>> +				     "number of chip-selects not defined\n");
>> +
>> +	ret = alloc_chrdev_region(&penctrl_devt, 0, num_cs, "penctrl");
>> +	if (ret)
>> +		return dev_err_probe(&spi->dev, ret,
>> +				     "failed to alloc chrdev region\n");
>> +
>> +	penctrl_class = class_create("penctrl");
>> +	if (IS_ERR(penctrl_class)) {
>> +		ret = dev_err_probe(&spi->dev, PTR_ERR(penctrl_class),
>> +				    "failed to create class\n");
>> +		goto unregister_chrdev;
>> +	}
>> +
>> +	cdev = cdev_alloc();
>> +	if (!cdev) {
>> +		ret = dev_err_probe(&spi->dev, -ENOMEM,
>> +				    "allocation of cdev failed\n");
>> +		goto destroy_class;
>> +	}
>> +	cdev->owner = THIS_MODULE;
>> +	cdev_init(cdev, &penctrl_fops);
>> +
>> +	ret = cdev_add(cdev, penctrl_devt, num_cs);
>> +	if (ret) {
>> +		ret = dev_err_probe(&spi->dev, ret,
>> +				    "register of cdev failed\n");
>> +		goto free_cdev;
>> +	}
>> +
>> +	/* Allocate driver data */
>> +	penctrl = kzalloc(sizeof(*penctrl), GFP_KERNEL);
>> +	if (!penctrl) {
>> +		ret = -ENOMEM;
>> +		goto free_cdev;
>> +	}
>> +	penctrl->spi = spi;
>> +	mutex_init(&spi_lock);
>> +
>> +	/* Create a device for each chip select */
>> +	for (cs = 0; cs < num_cs; cs++) {
>> +		dev = device_create(penctrl_class,
>> +				    &spi->dev,
>> +				    MKDEV(MAJOR(penctrl_devt), cs),
>> +				    penctrl,
>> +				    "penctrl0.%d",
>> +				    cs);
>> +		if (IS_ERR(dev)) {
>> +			ret = dev_err_probe(&spi->dev, PTR_ERR(dev),
>> +					    "error creating device\n");
>> +			goto destroy_device;
>> +		}
>> +		dev_dbg(&spi->dev, "created device major %u, minor
>> %d\n",
>> +			MAJOR(penctrl_devt), cs);
>> +	}
>> +
>> +	/* Register reset controller */
>> +	penctrl->rcdev.dev = &spi->dev;
>> +	penctrl->rcdev.ops = &penctrl_reset_ops;
>> +	penctrl->rcdev.owner = THIS_MODULE;
>> +	penctrl->rcdev.of_node = spi->dev.of_node;
>> +	penctrl->rcdev.nr_resets = 1;
>> +
>> +	ret = reset_controller_register(&penctrl->rcdev);
>> +	if (ret)
>> +		return dev_err_probe(&spi->dev, ret,
>> +				     "failed to register reset controller\n");
>> +	return 0;
>> +
>> +destroy_device:
>> +	for (cs = 0; cs < num_cs; cs++)
>> +		device_destroy(penctrl_class,
>> MKDEV(MAJOR(penctrl_devt), cs));
>> +	kfree(penctrl);
>> +free_cdev:
>> +	cdev_del(cdev);
>> +destroy_class:
>> +	class_destroy(penctrl_class);
>> +unregister_chrdev:
>> +	unregister_chrdev(MAJOR(penctrl_devt), "penctrl");
>> +
>> +	return ret;
>> +}
>> +
>> +static const struct of_device_id penctrl_dt_match[] = {
>> +	{ .compatible = "amd,pensando-elba-ctrl" },
>> +	{ /* sentinel */ }
>> +};
>> +
>> +static struct spi_driver penctrl_spi_driver = {
>> +	.probe = penctrl_spi_probe,
>> +	.driver = {
>> +		.name = "pensando-elba-ctrl",
>> +		.of_match_table = penctrl_dt_match,
>> +	},
>> +};
>> +module_spi_driver(penctrl_spi_driver);
>> +
>> +MODULE_AUTHOR("Brad Larson <blarson@amd.com>");
>> MODULE_DESCRIPTION("AMD
>> +Pensando SoC Controller via SPI"); MODULE_LICENSE("GPL");
>> diff --git a/include/uapi/linux/amd-pensando-ctrl.h
>> b/include/uapi/linux/amd-pensando-ctrl.h
>> new file mode 100644
>> index 000000000000..e5f9f0dfe146
>> --- /dev/null
>> +++ b/include/uapi/linux/amd-pensando-ctrl.h
>> @@ -0,0 +1,29 @@
>> +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
>> +/*
>> + * Userspace interface for /dev/penctrl

You are missing AMD copyright here.
Also in DT patches I have seen that you didn't switch to 2023 year yet.

Thanks,
Michal
  

Hi Brad,

kernel test robot noticed the following build warnings:

[auto build test WARNING on e922ba281a8d84f640d8c8e18a385d032c19e185]

url:    https://github.com/intel-lab-lkp/linux/commits/Brad-Larson/dt-bindings-arm-add-AMD-Pensando-boards/20230516-032312
base:   e922ba281a8d84f640d8c8e18a385d032c19e185
patch link:    https://lore.kernel.org/r/20230515181606.65953-9-blarson%40amd.com
patch subject: [PATCH v14 8/8] soc: amd: Add support for AMD Pensando SoC Controller
config: sh-allmodconfig (https://download.01.org/0day-ci/archive/20230516/202305161642.PZQ4S8o2-lkp@intel.com/config)
compiler: sh4-linux-gcc (GCC) 12.1.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/intel-lab-lkp/linux/commit/48a90df35083d2f3788e171ff0af01ddc8cd871b
        git remote add linux-review https://github.com/intel-lab-lkp/linux
        git fetch --no-tags linux-review Brad-Larson/dt-bindings-arm-add-AMD-Pensando-boards/20230516-032312
        git checkout 48a90df35083d2f3788e171ff0af01ddc8cd871b
        # save the config file
        mkdir build_dir && cp config build_dir/.config
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=sh olddefconfig
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=sh SHELL=/bin/bash drivers/soc/amd/

If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Link: https://lore.kernel.org/oe-kbuild-all/202305161642.PZQ4S8o2-lkp@intel.com/

All warnings (new ones prefixed by >>):

   drivers/soc/amd/pensando-ctrl.c: In function 'penctrl_ioctl':
>> drivers/soc/amd/pensando-ctrl.c:91:36: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
      91 |         if (copy_from_user(tx_buf, (void __user *)msg->tx_buf, msg->len)) {
         |                                    ^
   drivers/soc/amd/pensando-ctrl.c:114:26: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
     114 |         if (copy_to_user((void __user *)msg->rx_buf, rx_buf, msg->len))
         |                          ^


vim +91 drivers/soc/amd/pensando-ctrl.c

    34	
    35	static long
    36	penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
    37	{
    38		void __user *in_arg = (void __user *)arg;
    39		struct penctrl_device *penctrl;
    40		u8 tx_buf[PENCTRL_MAX_MSG_LEN];
    41		u8 rx_buf[PENCTRL_MAX_MSG_LEN];
    42		struct spi_transfer t[2] = {};
    43		struct penctrl_spi_xfer *msg;
    44		struct spi_device *spi;
    45		unsigned int num_msgs;
    46		struct spi_message m;
    47		u32 size;
    48		int ret;
    49	
    50		/* Check for a valid command */
    51		if (_IOC_TYPE(cmd) != PENCTRL_IOC_MAGIC)
    52			return -ENOTTY;
    53	
    54		if (_IOC_NR(cmd) > PENCTRL_IOC_MAXNR)
    55			return -ENOTTY;
    56	
    57		if (((_IOC_DIR(cmd) & _IOC_READ)) && !access_ok(in_arg, _IOC_SIZE(cmd)))
    58			return -EFAULT;
    59	
    60		if (((_IOC_DIR(cmd) & _IOC_WRITE)) && !access_ok(in_arg, _IOC_SIZE(cmd)))
    61			return -EFAULT;
    62	
    63		/* Get a reference to the SPI device */
    64		penctrl = filp->private_data;
    65		if (!penctrl)
    66			return -ESHUTDOWN;
    67	
    68		spi = spi_dev_get(penctrl->spi);
    69		if (!spi)
    70			return -ESHUTDOWN;
    71	
    72		/* Verify and prepare SPI message */
    73		size = _IOC_SIZE(cmd);
    74		num_msgs = size / sizeof(struct penctrl_spi_xfer);
    75		if (num_msgs > 2 || size == 0 || size % sizeof(struct penctrl_spi_xfer)) {
    76			ret = -EINVAL;
    77			goto out_unlock;
    78		}
    79		msg = memdup_user((struct penctrl_spi_xfer *)arg, size);
    80		if (IS_ERR(msg)) {
    81			ret = PTR_ERR(msg);
    82			goto out_unlock;
    83		}
    84		if (msg->len > PENCTRL_MAX_MSG_LEN) {
    85			ret = -EINVAL;
    86			goto out_unlock;
    87		}
    88	
    89		t[0].tx_buf = tx_buf;
    90		t[0].len = msg->len;
  > 91		if (copy_from_user(tx_buf, (void __user *)msg->tx_buf, msg->len)) {
    92			ret = -EFAULT;
    93			goto out_unlock;
    94		}
    95		if (num_msgs > 1) {
    96			msg++;
    97			if (msg->len > PENCTRL_MAX_MSG_LEN) {
    98				ret = -EINVAL;
    99				goto out_unlock;
   100			}
   101			t[1].rx_buf = rx_buf;
   102			t[1].len = msg->len;
   103		}
   104		spi_message_init_with_transfers(&m, t, num_msgs);
   105	
   106		/* Perform the transfer */
   107		mutex_lock(&spi_lock);
   108		ret = spi_sync(spi, &m);
   109		mutex_unlock(&spi_lock);
   110	
   111		if (ret || (num_msgs == 1))
   112			goto out_unlock;
   113	
   114		if (copy_to_user((void __user *)msg->rx_buf, rx_buf, msg->len))
   115			ret = -EFAULT;
   116	
   117	out_unlock:
   118		spi_dev_put(spi);
   119		return ret;
   120	}
   121
  

On Mon, May 15, 2023, at 20:16, Brad Larson wrote:
> The Pensando SoC controller is a SPI connected companion device
> that is present in all Pensando SoC board designs.  The essential
> board management registers are accessed on chip select 0 with
> board mgmt IO support accessed using additional chip selects.
>
> Signed-off-by: Brad Larson <blarson@amd.com>

Hi Brad,

I'm sorry I wasn't paying enough attention to this driver as the
past 13 revisions went by.

> v10 changes:
> - Different driver implementation specific to this Pensando controller device.
> - Moved to soc/amd directory under new name based on guidance.  This driver is
>   of no use to any design other than all Pensando SoC based cards.
> - Removed use of builtin_driver, can be built as a module.

it looks like this was a fundamental change that I failed to see.

> +# SPDX-License-Identifier: GPL-2.0-only
> +menu "AMD Pensando SoC drivers"
> +
> +config AMD_PENSANDO_CTRL
> +	tristate "AMD Pensando SoC Controller"
> +	depends on SPI_MASTER=y
> +	depends on (ARCH_PENSANDO && OF) || COMPILE_TEST
> +	default ARCH_PENSANDO
> +	select REGMAP_SPI
> +	select MFD_SYSCON
> +	help
> +	  Enables AMD Pensando SoC controller device support.  This is a SPI
> +	  attached companion device in all Pensando SoC board designs which
> +	  provides essential board control/status registers and management IO
> +	  support.

So generally speaking, I don't want custom user interfaces in
drivers/soc. It looks like this one has internal interfaces for
a reset controller and the regmap, so those parts are fine, but
I'm confused about the purpose of the ioctl interface:

> +static long
> +penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
> +{

> +	if (num_msgs > 1) {
> +		msg++;
> +		if (msg->len > PENCTRL_MAX_MSG_LEN) {
> +			ret = -EINVAL;
> +			goto out_unlock;
> +		}
> +		t[1].rx_buf = rx_buf;
> +		t[1].len = msg->len;
> +	}
> +	spi_message_init_with_transfers(&m, t, num_msgs);

This seems to be just a passthrough of user space messages, which
is what the spidev driver already provides, but using a different
ioctl interface. I don't really want any user-level interfaces
in drivers/soc as a rule, but having one that duplicates existing
functionality seems particularly wrong.

Can you explain what the purpose is? Is this about serializing
access between the in-kernel reset control and the user-side
access?

Also, can you explain why this needs a low-lever user interface
in the first place, rather than something that can be expressed
using high-level abstractions as you already do with the reset
control?

All of the above should be part of the changelog text to get a
driver like this merged. I don't think we can get a quick
solution here though, so maybe you can start by removing the
ioctl side and having the rest of the driver in drivers/reset?

     Arnd
  

Hi Michal,

On Tue, May 16, 2023 at 9:42 AM Michal Simek <michal.simek@amd.com> wrote:
> Also in DT patches I have seen that you didn't switch to 2023 year yet.

If no substantial changes were made in 2023, there is no need nor
desire to update the copyright year.

Gr{oetje,eeting}s,

                        Geert
  

Hi Andy,

On Tue, May 16, 2023 at 00:05:32 Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
> On Mon, May 15, 2023 at 9:18 PM Brad Larson <blarson@amd.com> wrote:
>>
>> The Pensando SoC controller is a SPI connected companion device
>> that is present in all Pensando SoC board designs.  The essential
>> board management registers are accessed on chip select 0 with
>> board mgmt IO support accessed using additional chip selects.
>
> ...
>
>> +#include <linux/cdev.h>
>> +#include <linux/device.h>
>> +#include <linux/err.h>
>> +#include <linux/init.h>
>> +#include <linux/mod_devicetable.h>
>> +#include <linux/module.h>
>> +#include <linux/mutex.h>
>
>> +#include <linux/of.h>
>
> Unneeded inclusion.

Removed

>> +#include <linux/reset-controller.h>
>> +#include <linux/spi/spi.h>
>
> ...
>
>
>> +       u8 tx_buf[PENCTRL_MAX_MSG_LEN];
>> +       u8 rx_buf[PENCTRL_MAX_MSG_LEN];
>
> Does it need to be DMA-capable?

Doesn't need to be DMA-capable

> ...
>
>> +       spi->chip_select = current_cs;
>> +       spi->cs_gpiod = spi->controller->cs_gpiods[current_cs];
>
> Nowadays these require API calls instead of direct assignments.

Changed to:
	spi_set_csgpiod(spi, 0, spi->controller->cs_gpiods[current_cs]);

> ...
>
>> +static int penctrl_release(struct inode *inode, struct file *filp)
>> +{
>> +       filp->private_data = NULL;
>> +       return 0;
>> +}
>
> Is it possible to unload the module without releasing the device node?

If the refcount is not zero the kernel prevents the module from being unloaded.

> ...
>
>> +       u8 txbuf[3];
>> +       u8 rxbuf[1];
>
> Same question about DMA.

Not DMA-capable

> ...
>
>> +       ret = spi_sync(spi, &m);
>
>> +       if (ret == 0)
>> +               *val = rxbuf[0];
>> +
>> +       return ret;
>
> Can also be written in more usual way:
>
>  if (ret)
>    return ret;
>  ...
>  return 0;

Yes, changed to:
        ret = spi_sync(spi, &m);
        if (ret)
                return ret;

        *val = rxbuf[0];
        return 0;

> ...
>
>> +       u8 txbuf[4];
>
> DMA?

Not DMA-capable

> ...
>
>> +       spi->chip_select = 0;
>> +       spi->cs_gpiod = spi->controller->cs_gpiods[0];
>
> Setter APIs.

Changed to:
	spi_set_csgpiod(spi, 0, spi->controller->cs_gpiods[0]);

>
> ...
>
>> +       spi->chip_select = 0;
>> +       spi->cs_gpiod = spi->controller->cs_gpiods[0];
>
> Ditto.

Changed to:
	spi_set_csgpiod(spi, 0, spi->controller->cs_gpiods[0]);

>> +       ret = device_property_read_u32(spi->dev.parent, "num-cs", &num_cs);
>> +       if (ret)
>> +               return dev_err_probe(&spi->dev, ret,
>> +                                    "number of chip-selects not defined\n");
>
> Hmm... Shouldn't SPI core take care of this in a generic way? Yes, I
> understand that you need the number for the allocation, but I would
> expect something like spi_fw_get_num_cs() to exist (seems not?).
>

No need to look into the parent node, changed to this:

        num_cs = spi->controller->num_chipselect;

> ...
>
>> +       penctrl->rcdev.of_node = spi->dev.of_node;
>
> Use device_set_node(). It helps to modify the data types beneath.

Added:
	device_set_node(penctrl->rcdev.dev, dev_fwnode(&spi->dev));

Regards,
Brad
  

Hi Arnd,

> On Mon, May 15, 2023, at 20:16, Brad Larson wrote:
>> The Pensando SoC controller is a SPI connected companion device
>> that is present in all Pensando SoC board designs.  The essential
>> board management registers are accessed on chip select 0 with
>> board mgmt IO support accessed using additional chip selects.
>>
>> Signed-off-by: Brad Larson <blarson@amd.com>
>
> Hi Brad,
>
> I'm sorry I wasn't paying enough attention to this driver as the
> past 13 revisions went by.
>
No worries, bit of a saga.  See explanation below.

>> v10 changes:
>> - Different driver implementation specific to this Pensando controller device.
>> - Moved to soc/amd directory under new name based on guidance.  This driver is
>>   of no use to any design other than all Pensando SoC based cards.
>> - Removed use of builtin_driver, can be built as a module.
>
> it looks like this was a fundamental change that I failed to see.

See explanation below.

>> +# SPDX-License-Identifier: GPL-2.0-only
>> +menu "AMD Pensando SoC drivers"
>> +
>> +config AMD_PENSANDO_CTRL
>> +	tristate "AMD Pensando SoC Controller"
>> +	depends on SPI_MASTER=y
>> +	depends on (ARCH_PENSANDO && OF) || COMPILE_TEST
>> +	default ARCH_PENSANDO
>> +	select REGMAP_SPI
>> +	select MFD_SYSCON
>> +	help
>> +	  Enables AMD Pensando SoC controller device support.  This is a SPI
>> +	  attached companion device in all Pensando SoC board designs which
>> +	  provides essential board control/status registers and management IO
>> +	  support.
>
> So generally speaking, I don't want custom user interfaces in
> drivers/soc. It looks like this one has internal interfaces for
> a reset controller and the regmap, so those parts are fine, but
> I'm confused about the purpose of the ioctl interface:
>
>> +static long
>> +penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>> +{
>
>> +	if (num_msgs > 1) {
>> +		msg++;
>> +		if (msg->len > PENCTRL_MAX_MSG_LEN) {
>> +			ret = -EINVAL;
>> +			goto out_unlock;
>> +		}
>> +		t[1].rx_buf = rx_buf;
>> +		t[1].len = msg->len;
>> +	}
>> +	spi_message_init_with_transfers(&m, t, num_msgs);
>
> This seems to be just a passthrough of user space messages, which
> is what the spidev driver already provides, but using a different
> ioctl interface. I don't really want any user-level interfaces
> in drivers/soc as a rule, but having one that duplicates existing
> functionality seems particularly wrong.
>
> Can you explain what the purpose is? Is this about serializing
> access between the in-kernel reset control and the user-side
> access?
>
> Also, can you explain why this needs a low-lever user interface
> in the first place, rather than something that can be expressed
> using high-level abstractions as you already do with the reset
> control?
>
> All of the above should be part of the changelog text to get a
> driver like this merged. I don't think we can get a quick
> solution here though, so maybe you can start by removing the
> ioctl side and having the rest of the driver in drivers/reset?

In the original patchset I added a pensando compatible to spidev and that
was nacked in review and reusing some random compatible that made it into 
spidev was just wrong.  Further it was recommended this should be a system 
specific driver and don't rely on a debug driver like spidev.  I changed 
over to a platform specific driver and at that time I also needed to include 
a reset controller (emmc reset only).  I put these in drivers/mfd and 
drivers/reset.  Review of the device tree for this approach went back and 
forth to _not_ have four child nodes on the spi device each with the same 
compatible. Decision was to squash the child nodes into the parent and put 
the reset-controller there also.  One driver and since its pensando
specific its currently in drivers/soc/amd.

There are five different user processes and some utilities that access the 
functionality in the cpld/fpga.  You're correct, its passing messages that 
are specific to the IP accessed via chip-select.  No Elba system will boot 
without this driver providing ioctl access.

Regards,
Brad
  

On Wed, May 24, 2023, at 00:11, Brad Larson wrote:
>> On Mon, May 15, 2023, at 20:16, Brad Larson wrote:
>
>> Also, can you explain why this needs a low-lever user interface
>> in the first place, rather than something that can be expressed
>> using high-level abstractions as you already do with the reset
>> control?
>>
>> All of the above should be part of the changelog text to get a
>> driver like this merged. I don't think we can get a quick
>> solution here though, so maybe you can start by removing the
>> ioctl side and having the rest of the driver in drivers/reset?
>
> In the original patchset I added a pensando compatible to spidev and that
> was nacked in review and reusing some random compatible that made it into 
> spidev was just wrong.  Further it was recommended this should be a system 
> specific driver and don't rely on a debug driver like spidev.  I changed 
> over to a platform specific driver and at that time I also needed to include 
> a reset controller (emmc reset only).  I put these in drivers/mfd and 
> drivers/reset.  Review of the device tree for this approach went back and 
> forth to _not_ have four child nodes on the spi device each with the same 
> compatible. Decision was to squash the child nodes into the parent and put 
> the reset-controller there also.  One driver and since its pensando
> specific its currently in drivers/soc/amd.
>
> There are five different user processes and some utilities that access the 
> functionality in the cpld/fpga.  You're correct, its passing messages that 
> are specific to the IP accessed via chip-select.  No Elba system will boot 
> without this driver providing ioctl access.

Thank you for the detailed summary. Moving away from spidev and
from mfd seems all reasonable here. I'm still a bit confused by
why you have multiple chipselects here that are for different
subdevices but ended with a single user interface for all of them,
but that's not a big deal.

The main bit that sticks out about this high-level design is how
it relies on user space utilities at all to understand the message
format. From what I understand about the actual functionality of
this device, it most closely resembles an embedded controller that
you might find in a laptop or server machine, and those usually
have kernel drivers in drivers/platform/ to interact with the
device.

Has anyone tried to do it like that? Maybe it would help
to see what the full protocol and the user space side looks
like, in order to move some or all of it into the kernel.

     Arnd
  

Hi Arnd,

> On Wed, May 24, 2023, at 00:11, Brad Larson wrote:
>>> On Mon, May 15, 2023, at 20:16, Brad Larson wrote:
...
>>> Also, can you explain why this needs a low-lever user interface
>>> in the first place, rather than something that can be expressed
>>> using high-level abstractions as you already do with the reset
>>> control?
>>>
>>> All of the above should be part of the changelog text to get a
>>> driver like this merged. I don't think we can get a quick
>>> solution here though, so maybe you can start by removing the
>>> ioctl side and having the rest of the driver in drivers/reset?

Might be best to pull the whole thing for now until an acceptable 
solution is reached.  The reset function is a recovery mechanisim rarely
used where the byte access to the different IP at the 4 chip-selects
is needed for a system to boot.

>> In the original patchset I added a pensando compatible to spidev and that
>> was nacked in review and reusing some random compatible that made it into 
>> spidev was just wrong.  Further it was recommended this should be a system 
>> specific driver and don't rely on a debug driver like spidev.  I changed 
>> over to a platform specific driver and at that time I also needed to include 
>> a reset controller (emmc reset only).  I put these in drivers/mfd and 
>> drivers/reset.  Review of the device tree for this approach went back and 
>> forth to _not_ have four child nodes on the spi device each with the same 
>> compatible. Decision was to squash the child nodes into the parent and put 
>> the reset-controller there also.  One driver and since its pensando
>> specific its currently in drivers/soc/amd.
>>
>> There are five different user processes and some utilities that access the 
>> functionality in the cpld/fpga.  You're correct, its passing messages that 
>> are specific to the IP accessed via chip-select.  No Elba system will boot 
>> without this driver providing ioctl access.

> Thank you for the detailed summary. Moving away from spidev and
> from mfd seems all reasonable here. I'm still a bit confused by
> why you have multiple chipselects here that are for different
> subdevices but ended with a single user interface for all of them,
> but that's not a big deal.

The goal is to isolate the the kernel from device and platform specific
changes.  All the IO to the spi connected CPLD/FPGA (design/cost dependent)
is a byte at a time or up to 16 bytes for internal flash mgmt.  Performance
is not an issue and spidev was sufficient.

Maybe this paints the right picture to zero in on a correct approach.
Internal and external IP can be present at CS1/CS2 depending on the design
where the CS0 board controller registers get additions over time in a
backward compatible manner.

Design 1: FPGA 
CS0: Board controller registers
CS1: Designware SPI to I2C to board peripherals
CS2: Lattice dual I2C master
CS3: Internal storage

Design 2: CPLD
CS0: Board controller registers
CS1: Not used or some other board specific registers
CS2: Lattice dual I2C master
CS3: Internal storage

> The main bit that sticks out about this high-level design is how
> it relies on user space utilities at all to understand the message
> format. From what I understand about the actual functionality of
> this device, it most closely resembles an embedded controller that
> you might find in a laptop or server machine, and those usually
> have kernel drivers in drivers/platform/ to interact with the
> device.

The dozens of registers at CS0 for board management are defined in
userspace programs or script.  Only the regsiter offset/bit for
emmc reset is needed for the reset function in the patches.

> Has anyone tried to do it like that? Maybe it would help
> to see what the full protocol and the user space side looks
> like, in order to move some or all of it into the kernel.

Looking at drivers/platform its pretty sparse.  What do you 
recommend based on the design 1/2 variations?

Regards,
Brad