[v2,2/3] irqchip: Add the Sophgo SG2042 MSI interrupt controller

Commit Message

Chen Wang Dec. 9, 2024, 7:12 a.m. UTC

From: Chen Wang <[e-mail]>

Add driver for Sophgo SG2042 MSI interrupt controller.

Signed-off-by: Chen Wang <[e-mail]>
---
 drivers/irqchip/Kconfig          |  12 ++
 drivers/irqchip/Makefile         |   1 +
 drivers/irqchip/irq-sg2042-msi.c | 285 +++++++++++++++++++++++++++++++
 3 files changed, 298 insertions(+)
 create mode 100644 drivers/irqchip/irq-sg2042-msi.c

Comments

Krzysztof Kozlowski Dec. 9, 2024, 9:34 a.m. UTC | #1

On Mon, Dec 09, 2024 at 03:12:00PM +0800, Chen Wang wrote:
> +static void sg2042_msi_irq_ack(struct irq_data *d)
> +{
> +	struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
> +	int bit_off = d->hwirq - data->irq_first;
> +
> +	writel(1 << bit_off, (unsigned int *)data->reg_clr);
> +
> +	irq_chip_ack_parent(d);
> +}
> +
> +static void sg2042_msi_irq_compose_msi_msg(struct irq_data *data,
> +					   struct msi_msg *msg)
> +{
> +	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(data);
> +
> +	msg->address_hi = upper_32_bits(priv->doorbell_addr);
> +	msg->address_lo = lower_32_bits(priv->doorbell_addr);
> +	msg->data = 1 << (data->hwirq - priv->irq_first);
> +
> +	pr_debug("%s hwirq[%ld]: address_hi[%#x], address_lo[%#x], data[%#x]\n",
> +		 __func__, data->hwirq, msg->address_hi, msg->address_lo, msg->data);

Don't print addresses, it is useless - it will be a constant.

> +}
> +
> +static struct irq_chip sg2042_msi_middle_irq_chip = {
> +	.name			= "SG2042 MSI",
> +	.irq_ack		= sg2042_msi_irq_ack,
> +	.irq_mask		= irq_chip_mask_parent,
> +	.irq_unmask		= irq_chip_unmask_parent,
> +#ifdef CONFIG_SMP
> +	.irq_set_affinity	= irq_chip_set_affinity_parent,
> +#endif
> +	.irq_compose_msi_msg	= sg2042_msi_irq_compose_msi_msg,
> +};

...

> +static int sg2042_msi_probe(struct platform_device *pdev)
> +{
> +	struct of_phandle_args args = {};
> +	struct sg2042_msi_data *data;
> +	int ret;
> +
> +	data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), GFP_KERNEL);
> +	if (!data)
> +		return -ENOMEM;
> +
> +	data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
> +	if (IS_ERR(data->reg_clr)) {
> +		dev_err(&pdev->dev, "Failed to map clear register\n");
> +		return PTR_ERR(data->reg_clr);
> +	}
> +
> +	if (of_property_read_u64(pdev->dev.of_node, "sophgo,msi-doorbell-addr",
> +				 &data->doorbell_addr)) {
> +		dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
> +		return -EINVAL;
> +	}
> +
> +	ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
> +					 "#interrupt-cells", 0, &args);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
> +		return ret;
> +	}

You leak the phandle. You leak much more, btw...

> +	data->irq_first = (u32)args.args[0];
> +
> +	ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
> +					 args.args_count + 1, &data->num_irqs);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
> +		return ret;
> +	}
> +
> +	if (data->irq_first < SG2042_VECTOR_MIN ||
> +	    (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
> +		dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
> +		return -EINVAL;
> +	}
> +
> +	mutex_init(&data->msi_map_lock);
> +
> +	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);

This also leaks during removal.

> +	if (!data->msi_map)
> +		return -ENOMEM;
> +
> +	ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
> +	if (ret)
> +		bitmap_free(data->msi_map);
> +
> +	return ret;
> +}
> +
> +static const struct of_device_id sg2042_msi_of_match[] = {
> +	{ .compatible	= "sophgo,sg2042-msi" },
> +	{}
> +};
> +
> +static struct platform_driver sg2042_msi_driver = {
> +	.driver = {
> +		.name		= "sg2042-msi",
> +		.of_match_table	= of_match_ptr(sg2042_msi_of_match),

Drop of_match_ptr(), unnecessary and might lead to warnings even if this
is not a module.

Best regards,
Krzysztof

kernel test robot Dec. 10, 2024, 7:37 a.m. UTC | #2

Hi Chen,

kernel test robot noticed the following build warnings:

[auto build test WARNING on fac04efc5c793dccbd07e2d59af9f90b7fc0dca4]

url:    https://github.com/intel-lab-lkp/linux/commits/Chen-Wang/dt-bindings-interrupt-controller-Add-Sophgo-SG2042-MSI/20241209-151429
base:   fac04efc5c793dccbd07e2d59af9f90b7fc0dca4
patch link:    https://lore.kernel.org/r/c882fe329932409131be76ce47b81a6155595ce4.1733726057.git.unicorn_wang%40outlook.com
patch subject: [PATCH v2 2/3] irqchip: Add the Sophgo SG2042 MSI interrupt controller
config: arm-randconfig-r131-20241210 (https://download.01.org/0day-ci/archive/20241210/[e-mail]/config)
compiler: arm-linux-gnueabi-gcc (GCC) 14.2.0
reproduce: (https://download.01.org/0day-ci/archive/20241210/[e-mail]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[e-mail]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[e-mail]/

sparse warnings: (new ones prefixed by >>)
>> drivers/irqchip/irq-sg2042-msi.c:64:9: sparse: sparse: cast removes address space '__iomem' of expression
>> drivers/irqchip/irq-sg2042-msi.c:64:9: sparse: sparse: incorrect type in argument 2 (different address spaces) @@     expected void volatile [noderef] __iomem *addr @@     got unsigned int * @@
   drivers/irqchip/irq-sg2042-msi.c:64:9: sparse:     expected void volatile [noderef] __iomem *addr
   drivers/irqchip/irq-sg2042-msi.c:64:9: sparse:     got unsigned int *

vim +/__iomem +64 drivers/irqchip/irq-sg2042-msi.c

    58	
    59	static void sg2042_msi_irq_ack(struct irq_data *d)
    60	{
    61		struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
    62		int bit_off = d->hwirq - data->irq_first;
    63	
  > 64		writel(1 << bit_off, (unsigned int *)data->reg_clr);
    65	
    66		irq_chip_ack_parent(d);
    67	}
    68

kernel test robot Dec. 10, 2024, 6:34 p.m. UTC | #3

Hi Chen,

kernel test robot noticed the following build warnings:

[auto build test WARNING on fac04efc5c793dccbd07e2d59af9f90b7fc0dca4]

url:    https://github.com/intel-lab-lkp/linux/commits/Chen-Wang/dt-bindings-interrupt-controller-Add-Sophgo-SG2042-MSI/20241209-151429
base:   fac04efc5c793dccbd07e2d59af9f90b7fc0dca4
patch link:    https://lore.kernel.org/r/c882fe329932409131be76ce47b81a6155595ce4.1733726057.git.unicorn_wang%40outlook.com
patch subject: [PATCH v2 2/3] irqchip: Add the Sophgo SG2042 MSI interrupt controller
config: arm-randconfig-r131-20241210 (https://download.01.org/0day-ci/archive/20241211/[e-mail]/config)
compiler: arm-linux-gnueabi-gcc (GCC) 14.2.0
reproduce: (https://download.01.org/0day-ci/archive/20241211/[e-mail]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[e-mail]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[e-mail]/

sparse warnings: (new ones prefixed by >>)
>> drivers/irqchip/irq-sg2042-msi.c:64:9: sparse: sparse: cast removes address space '__iomem' of expression
>> drivers/irqchip/irq-sg2042-msi.c:64:9: sparse: sparse: incorrect type in argument 2 (different address spaces) @@     expected void volatile [noderef] __iomem *addr @@     got unsigned int * @@
   drivers/irqchip/irq-sg2042-msi.c:64:9: sparse:     expected void volatile [noderef] __iomem *addr
   drivers/irqchip/irq-sg2042-msi.c:64:9: sparse:     got unsigned int *

vim +/__iomem +64 drivers/irqchip/irq-sg2042-msi.c

    58	
    59	static void sg2042_msi_irq_ack(struct irq_data *d)
    60	{
    61		struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
    62		int bit_off = d->hwirq - data->irq_first;
    63	
  > 64		writel(1 << bit_off, (unsigned int *)data->reg_clr);
    65	
    66		irq_chip_ack_parent(d);
    67	}
    68

kernel test robot Dec. 10, 2024, 6:44 p.m. UTC | #4

Hi Chen,

kernel test robot noticed the following build warnings:

[auto build test WARNING on fac04efc5c793dccbd07e2d59af9f90b7fc0dca4]

url:    https://github.com/intel-lab-lkp/linux/commits/Chen-Wang/dt-bindings-interrupt-controller-Add-Sophgo-SG2042-MSI/20241209-151429
base:   fac04efc5c793dccbd07e2d59af9f90b7fc0dca4
patch link:    https://lore.kernel.org/r/c882fe329932409131be76ce47b81a6155595ce4.1733726057.git.unicorn_wang%40outlook.com
patch subject: [PATCH v2 2/3] irqchip: Add the Sophgo SG2042 MSI interrupt controller
config: alpha-randconfig-r063-20241211 (https://download.01.org/0day-ci/archive/20241211/[e-mail]/config)
compiler: alpha-linux-gcc (GCC) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241211/[e-mail]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[e-mail]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[e-mail]/

All warnings (new ones prefixed by >>):

>> drivers/irqchip/irq-sg2042-msi.c:273:34: warning: 'sg2042_msi_of_match' defined but not used [-Wunused-const-variable=]
     273 | static const struct of_device_id sg2042_msi_of_match[] = {
         |                                  ^~~~~~~~~~~~~~~~~~~


vim +/sg2042_msi_of_match +273 drivers/irqchip/irq-sg2042-msi.c

   272	
 > 273	static const struct of_device_id sg2042_msi_of_match[] = {
   274		{ .compatible	= "sophgo,sg2042-msi" },
   275		{}
   276	};
   277

Samuel Holland Dec. 10, 2024, 11:13 p.m. UTC | #5

On 2024-12-09 1:12 AM, Chen Wang wrote:
> From: Chen Wang <[e-mail]>
> 
> Add driver for Sophgo SG2042 MSI interrupt controller.
> 
> Signed-off-by: Chen Wang <[e-mail]>
> ---
>  drivers/irqchip/Kconfig          |  12 ++
>  drivers/irqchip/Makefile         |   1 +
>  drivers/irqchip/irq-sg2042-msi.c | 285 +++++++++++++++++++++++++++++++
>  3 files changed, 298 insertions(+)
>  create mode 100644 drivers/irqchip/irq-sg2042-msi.c
> 
> diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> index 9bee02db1643..161fb5df857f 100644
> --- a/drivers/irqchip/Kconfig
> +++ b/drivers/irqchip/Kconfig
> @@ -749,6 +749,18 @@ config MCHP_EIC
>  	help
>  	  Support for Microchip External Interrupt Controller.
>  
> +config SOPHGO_SG2042_MSI
> +	bool "Sophgo SG2042 MSI Controller"
> +	depends on ARCH_SOPHGO || COMPILE_TEST
> +	depends on PCI
> +	select IRQ_DOMAIN_HIERARCHY
> +	select IRQ_MSI_LIB
> +	select PCI_MSI
> +	help
> +	  Support for the Sophgo SG2042 MSI Controller.
> +	  This on-chip interrupt controller enables MSI sources to be
> +	  routed to the primary PLIC controller on SoC.
> +
>  config SUNPLUS_SP7021_INTC
>  	bool "Sunplus SP7021 interrupt controller" if COMPILE_TEST
>  	default SOC_SP7021
> diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
> index 25e9ad29b8c4..dd60e597491d 100644
> --- a/drivers/irqchip/Makefile
> +++ b/drivers/irqchip/Makefile
> @@ -128,4 +128,5 @@ obj-$(CONFIG_WPCM450_AIC)		+= irq-wpcm450-aic.o
>  obj-$(CONFIG_IRQ_IDT3243X)		+= irq-idt3243x.o
>  obj-$(CONFIG_APPLE_AIC)			+= irq-apple-aic.o
>  obj-$(CONFIG_MCHP_EIC)			+= irq-mchp-eic.o
> +obj-$(CONFIG_SOPHGO_SG2042_MSI)		+= irq-sg2042-msi.o
>  obj-$(CONFIG_SUNPLUS_SP7021_INTC)	+= irq-sp7021-intc.o
> diff --git a/drivers/irqchip/irq-sg2042-msi.c b/drivers/irqchip/irq-sg2042-msi.c
> new file mode 100644
> index 000000000000..495ee2b45eb2
> --- /dev/null
> +++ b/drivers/irqchip/irq-sg2042-msi.c
> @@ -0,0 +1,285 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * SG2042 MSI Controller
> + *
> + * Copyright (C) 2024 Sophgo Technology Inc.
> + * Copyright (C) 2024 Chen Wang <[e-mail]>
> + */
> +
> +#include <linux/cleanup.h>
> +#include <linux/io.h>
> +#include <linux/irq.h>
> +#include <linux/irqdomain.h>
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/msi.h>
> +#include <linux/of.h>
> +#include <linux/of_irq.h>
> +#include <linux/of_pci.h>
> +#include <linux/of_platform.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +
> +#include "irq-msi-lib.h"
> +
> +#define SG2042_VECTOR_MIN	64
> +#define SG2042_VECTOR_MAX	95
> +
> +struct sg2042_msi_data {
> +	void __iomem	*reg_clr;	// clear reg, see TRM, 10.1.33, GP_INTR0_CLR
> +
> +	u64		doorbell_addr;	// see TRM, 10.1.32, GP_INTR0_SET
> +
> +	u32		irq_first;	// The vector number that MSIs starts
> +	u32		num_irqs;	// The number of vectors for MSIs
> +
> +	unsigned long	*msi_map;
> +	struct mutex	msi_map_lock;	// lock for msi_map
> +};
> +
> +static int sg2042_msi_allocate_hwirq(struct sg2042_msi_data *priv, int num_req)
> +{
> +	int first;
> +
> +	guard(mutex)(&priv->msi_map_lock);
> +	first = bitmap_find_free_region(priv->msi_map, priv->num_irqs,
> +					get_count_order(num_req));
> +	return first >= 0 ? priv->irq_first + first : -ENOSPC;

How does this work? The irqdomain is instantiated with size == priv->num_irqs,
so hwirqs must be less than priv->num_irqs. It also quite simplifies the code
for hwirq be a number between 0 and priv->num_irqs. You only need to apply the
offset when allocating the parent IRQ:

	fwspec.param[0] = priv->irq_first + hwirq;

> +}
> +
> +static void sg2042_msi_free_hwirq(struct sg2042_msi_data *priv,
> +				  int hwirq, int num_req)
> +{
> +	int first = hwirq - priv->irq_first;
> +
> +	guard(mutex)(&priv->msi_map_lock);
> +	bitmap_release_region(priv->msi_map, first, get_count_order(num_req));
> +}
> +
> +static void sg2042_msi_irq_ack(struct irq_data *d)
> +{
> +	struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
> +	int bit_off = d->hwirq - data->irq_first;
> +
> +	writel(1 << bit_off, (unsigned int *)data->reg_clr);
> +
> +	irq_chip_ack_parent(d);
> +}
> +
> +static void sg2042_msi_irq_compose_msi_msg(struct irq_data *data,
> +					   struct msi_msg *msg)
> +{
> +	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(data);
> +
> +	msg->address_hi = upper_32_bits(priv->doorbell_addr);
> +	msg->address_lo = lower_32_bits(priv->doorbell_addr);
> +	msg->data = 1 << (data->hwirq - priv->irq_first);
> +
> +	pr_debug("%s hwirq[%ld]: address_hi[%#x], address_lo[%#x], data[%#x]\n",
> +		 __func__, data->hwirq, msg->address_hi, msg->address_lo, msg->data);
> +}
> +
> +static struct irq_chip sg2042_msi_middle_irq_chip = {
> +	.name			= "SG2042 MSI",
> +	.irq_ack		= sg2042_msi_irq_ack,
> +	.irq_mask		= irq_chip_mask_parent,
> +	.irq_unmask		= irq_chip_unmask_parent,
> +#ifdef CONFIG_SMP
> +	.irq_set_affinity	= irq_chip_set_affinity_parent,
> +#endif
> +	.irq_compose_msi_msg	= sg2042_msi_irq_compose_msi_msg,
> +};
> +
> +static int sg2042_msi_parent_domain_alloc(struct irq_domain *domain,
> +					  unsigned int virq, int hwirq)
> +{
> +	struct irq_fwspec fwspec;
> +	struct irq_data *d;
> +	int ret;
> +
> +	fwspec.fwnode = domain->parent->fwnode;
> +	fwspec.param_count = 2;
> +	fwspec.param[0] = hwirq;
> +	fwspec.param[1] = IRQ_TYPE_EDGE_RISING;
> +
> +	ret = irq_domain_alloc_irqs_parent(domain, virq, 1, &fwspec);
> +	if (ret)
> +		return ret;
> +
> +	d = irq_domain_get_irq_data(domain->parent, virq);
> +	return d->chip->irq_set_type(d, IRQ_TYPE_EDGE_RISING);
> +}
> +
> +static int sg2042_msi_middle_domain_alloc(struct irq_domain *domain,
> +					  unsigned int virq,
> +					  unsigned int nr_irqs, void *args)
> +{
> +	struct sg2042_msi_data *priv = domain->host_data;
> +	int hwirq, err, i;
> +
> +	hwirq = sg2042_msi_allocate_hwirq(priv, nr_irqs);
> +	if (hwirq < 0)
> +		return hwirq;
> +
> +	for (i = 0; i < nr_irqs; i++) {
> +		err = sg2042_msi_parent_domain_alloc(domain, virq + i, hwirq + i);
> +		if (err)
> +			goto err_hwirq;
> +
> +		pr_debug("%s: virq[%d], hwirq[%d]\n", __func__, virq + i, hwirq + i);
> +
> +		irq_domain_set_hwirq_and_chip(domain, virq + i, hwirq + i,
> +					      &sg2042_msi_middle_irq_chip, priv);
> +	}
> +
> +	return 0;
> +
> +err_hwirq:
> +	sg2042_msi_free_hwirq(priv, hwirq, nr_irqs);
> +	irq_domain_free_irqs_parent(domain, virq, i);
> +
> +	return err;
> +}
> +
> +static void sg2042_msi_middle_domain_free(struct irq_domain *domain,
> +					  unsigned int virq,
> +					  unsigned int nr_irqs)
> +{
> +	struct irq_data *d = irq_domain_get_irq_data(domain, virq);
> +	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(d);
> +
> +	irq_domain_free_irqs_parent(domain, virq, nr_irqs);
> +	sg2042_msi_free_hwirq(priv, d->hwirq, nr_irqs);
> +}
> +
> +static const struct irq_domain_ops sg2042_msi_middle_domain_ops = {
> +	.alloc	= sg2042_msi_middle_domain_alloc,
> +	.free	= sg2042_msi_middle_domain_free,
> +	.select	= msi_lib_irq_domain_select,
> +};
> +
> +#define SG2042_MSI_FLAGS_REQUIRED (MSI_FLAG_USE_DEF_DOM_OPS |	\
> +				   MSI_FLAG_USE_DEF_CHIP_OPS)
> +
> +#define SG2042_MSI_FLAGS_SUPPORTED MSI_GENERIC_FLAGS_MASK
> +
> +static struct msi_parent_ops sg2042_msi_parent_ops = {
> +	.required_flags		= SG2042_MSI_FLAGS_REQUIRED,
> +	.supported_flags	= SG2042_MSI_FLAGS_SUPPORTED,
> +	.bus_select_mask	= MATCH_PCI_MSI,
> +	.bus_select_token	= DOMAIN_BUS_NEXUS,
> +	.prefix			= "SG2042-",
> +	.init_dev_msi_info	= msi_lib_init_dev_msi_info,
> +};
> +
> +static int sg2042_msi_init_domains(struct sg2042_msi_data *priv,
> +				   struct device_node *node)
> +{
> +	struct fwnode_handle *fwnode = of_node_to_fwnode(node);
> +	struct irq_domain *plic_domain, *middle_domain;
> +	struct device_node *plic_node;
> +
> +	if (!of_find_property(node, "interrupt-parent", NULL)) {
> +		pr_err("Can't find interrupt-parent!\n");
> +		return -EINVAL;
> +	}
> +
> +	plic_node = of_irq_find_parent(node);
> +	if (!plic_node) {
> +		pr_err("Failed to find the PLIC node!\n");
> +		return -ENXIO;
> +	}
> +
> +	plic_domain = irq_find_host(plic_node);
> +	of_node_put(plic_node);
> +	if (!plic_domain) {
> +		pr_err("Failed to find the PLIC domain\n");
> +		return -ENXIO;
> +	}
> +
> +	middle_domain = irq_domain_create_hierarchy(plic_domain, 0, priv->num_irqs,
> +						    fwnode,
> +						    &sg2042_msi_middle_domain_ops,
> +						    priv);
> +	if (!middle_domain) {
> +		pr_err("Failed to create the MSI middle domain\n");
> +		return -ENOMEM;
> +	}
> +
> +	irq_domain_update_bus_token(middle_domain, DOMAIN_BUS_NEXUS);
> +
> +	middle_domain->flags |= IRQ_DOMAIN_FLAG_MSI_PARENT;
> +	middle_domain->msi_parent_ops = &sg2042_msi_parent_ops;
> +
> +	return 0;
> +}
> +
> +static int sg2042_msi_probe(struct platform_device *pdev)
> +{
> +	struct of_phandle_args args = {};
> +	struct sg2042_msi_data *data;
> +	int ret;
> +
> +	data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), GFP_KERNEL);
> +	if (!data)
> +		return -ENOMEM;
> +
> +	data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
> +	if (IS_ERR(data->reg_clr)) {
> +		dev_err(&pdev->dev, "Failed to map clear register\n");
> +		return PTR_ERR(data->reg_clr);
> +	}
> +
> +	if (of_property_read_u64(pdev->dev.of_node, "sophgo,msi-doorbell-addr",
> +				 &data->doorbell_addr)) {
> +		dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
> +		return -EINVAL;
> +	}
> +
> +	ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
> +					 "#interrupt-cells", 0, &args);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
> +		return ret;
> +	}
> +	data->irq_first = (u32)args.args[0];
> +
> +	ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
> +					 args.args_count + 1, &data->num_irqs);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
> +		return ret;
> +	}
> +
> +	if (data->irq_first < SG2042_VECTOR_MIN ||
> +	    (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
> +		dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
> +		return -EINVAL;
> +	}
> +
> +	mutex_init(&data->msi_map_lock);
> +
> +	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);
> +	if (!data->msi_map)
> +		return -ENOMEM;
> +
> +	ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
> +	if (ret)
> +		bitmap_free(data->msi_map);
> +
> +	return ret;
> +}
> +
> +static const struct of_device_id sg2042_msi_of_match[] = {
> +	{ .compatible	= "sophgo,sg2042-msi" },
> +	{}
> +};
> +
> +static struct platform_driver sg2042_msi_driver = {
> +	.driver = {
> +		.name		= "sg2042-msi",
> +		.of_match_table	= of_match_ptr(sg2042_msi_of_match),
> +	},
> +	.probe = sg2042_msi_probe,
> +};
> +builtin_platform_driver(sg2042_msi_driver);

Chen Wang Dec. 11, 2024, 9:10 a.m. UTC | #6

On 2024/12/9 17:34, Krzysztof Kozlowski wrote:
> On Mon, Dec 09, 2024 at 03:12:00PM +0800, Chen Wang wrote:
>> +static void sg2042_msi_irq_ack(struct irq_data *d)
>> +{
>> +	struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
>> +	int bit_off = d->hwirq - data->irq_first;
>> +
>> +	writel(1 << bit_off, (unsigned int *)data->reg_clr);
>> +
>> +	irq_chip_ack_parent(d);
>> +}
>> +
>> +static void sg2042_msi_irq_compose_msi_msg(struct irq_data *data,
>> +					   struct msi_msg *msg)
>> +{
>> +	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(data);
>> +
>> +	msg->address_hi = upper_32_bits(priv->doorbell_addr);
>> +	msg->address_lo = lower_32_bits(priv->doorbell_addr);
>> +	msg->data = 1 << (data->hwirq - priv->irq_first);
>> +
>> +	pr_debug("%s hwirq[%ld]: address_hi[%#x], address_lo[%#x], data[%#x]\n",
>> +		 __func__, data->hwirq, msg->address_hi, msg->address_lo, msg->data);
> Don't print addresses, it is useless - it will be a constant.
Ok.
>
>> +}
>> +
>> +static struct irq_chip sg2042_msi_middle_irq_chip = {
>> +	.name			= "SG2042 MSI",
>> +	.irq_ack		= sg2042_msi_irq_ack,
>> +	.irq_mask		= irq_chip_mask_parent,
>> +	.irq_unmask		= irq_chip_unmask_parent,
>> +#ifdef CONFIG_SMP
>> +	.irq_set_affinity	= irq_chip_set_affinity_parent,
>> +#endif
>> +	.irq_compose_msi_msg	= sg2042_msi_irq_compose_msi_msg,
>> +};
> ...
>
>> +static int sg2042_msi_probe(struct platform_device *pdev)
>> +{
>> +	struct of_phandle_args args = {};
>> +	struct sg2042_msi_data *data;
>> +	int ret;
>> +
>> +	data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), GFP_KERNEL);
>> +	if (!data)
>> +		return -ENOMEM;
>> +
>> +	data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
>> +	if (IS_ERR(data->reg_clr)) {
>> +		dev_err(&pdev->dev, "Failed to map clear register\n");
>> +		return PTR_ERR(data->reg_clr);
>> +	}
>> +
>> +	if (of_property_read_u64(pdev->dev.of_node, "sophgo,msi-doorbell-addr",
>> +				 &data->doorbell_addr)) {
>> +		dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
>> +					 "#interrupt-cells", 0, &args);
>> +	if (ret) {
>> +		dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
>> +		return ret;
>> +	}
> You leak the phandle. You leak much more, btw...
I will double-check, thanks.
>
>> +	data->irq_first = (u32)args.args[0];
>> +
>> +	ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
>> +					 args.args_count + 1, &data->num_irqs);
>> +	if (ret) {
>> +		dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
>> +		return ret;
>> +	}
>> +
>> +	if (data->irq_first < SG2042_VECTOR_MIN ||
>> +	    (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
>> +		dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	mutex_init(&data->msi_map_lock);
>> +
>> +	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);
> This also leaks during removal.
Got, thanks.
>
>> +	if (!data->msi_map)
>> +		return -ENOMEM;
>> +
>> +	ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
>> +	if (ret)
>> +		bitmap_free(data->msi_map);
>> +
>> +	return ret;
>> +}
>> +
>> +static const struct of_device_id sg2042_msi_of_match[] = {
>> +	{ .compatible	= "sophgo,sg2042-msi" },
>> +	{}
>> +};
>> +
>> +static struct platform_driver sg2042_msi_driver = {
>> +	.driver = {
>> +		.name		= "sg2042-msi",
>> +		.of_match_table	= of_match_ptr(sg2042_msi_of_match),
> Drop of_match_ptr(), unnecessary and might lead to warnings even if this
> is not a module.

Got, I will remove it, thanks.


>
> Best regards,
> Krzysztof
>

Chen Wang Dec. 11, 2024, 9:12 a.m. UTC | #7

On 2024/12/11 7:13, Samuel Holland wrote:
> On 2024-12-09 1:12 AM, Chen Wang wrote:
[......]
>> +
>> +static int sg2042_msi_allocate_hwirq(struct sg2042_msi_data *priv, int num_req)
>> +{
>> +	int first;
>> +
>> +	guard(mutex)(&priv->msi_map_lock);
>> +	first = bitmap_find_free_region(priv->msi_map, priv->num_irqs,
>> +					get_count_order(num_req));
>> +	return first >= 0 ? priv->irq_first + first : -ENOSPC;
> How does this work? The irqdomain is instantiated with size == priv->num_irqs,
> so hwirqs must be less than priv->num_irqs. It also quite simplifies the code
> for hwirq be a number between 0 and priv->num_irqs. You only need to apply the
> offset when allocating the parent IRQ:
>
> 	fwspec.param[0] = priv->irq_first + hwirq;

I will double check, seems this need to be improved.

Thanks,

Chen

[......]

Christophe JAILLET Dec. 11, 2024, 4:32 p.m. UTC | #8

Le 09/12/2024 à 08:12, Chen Wang a écrit :
> From: Chen Wang <[e-mail]>
> 
> Add driver for Sophgo SG2042 MSI interrupt controller.
> 
> Signed-off-by: Chen Wang <[e-mail]>

...

> +#define SG2042_VECTOR_MIN	64
> +#define SG2042_VECTOR_MAX	95

...

> +static struct irq_chip sg2042_msi_middle_irq_chip = {

const?

> +	.name			= "SG2042 MSI",
> +	.irq_ack		= sg2042_msi_irq_ack,
> +	.irq_mask		= irq_chip_mask_parent,
> +	.irq_unmask		= irq_chip_unmask_parent,
> +#ifdef CONFIG_SMP
> +	.irq_set_affinity	= irq_chip_set_affinity_parent,
> +#endif
> +	.irq_compose_msi_msg	= sg2042_msi_irq_compose_msi_msg,
> +};

...

> +static int sg2042_msi_probe(struct platform_device *pdev)
> +{
> +	struct of_phandle_args args = {};
> +	struct sg2042_msi_data *data;
> +	int ret;
> +
> +	data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), GFP_KERNEL);
> +	if (!data)
> +		return -ENOMEM;
> +
> +	data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
> +	if (IS_ERR(data->reg_clr)) {
> +		dev_err(&pdev->dev, "Failed to map clear register\n");
> +		return PTR_ERR(data->reg_clr);
> +	}
> +
> +	if (of_property_read_u64(pdev->dev.of_node, "sophgo,msi-doorbell-addr",
> +				 &data->doorbell_addr)) {
> +		dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
> +		return -EINVAL;
> +	}
> +
> +	ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
> +					 "#interrupt-cells", 0, &args);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
> +		return ret;
> +	}
> +	data->irq_first = (u32)args.args[0];
> +
> +	ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
> +					 args.args_count + 1, &data->num_irqs);
> +	if (ret) {
> +		dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
> +		return ret;
> +	}
> +
> +	if (data->irq_first < SG2042_VECTOR_MIN ||
> +	    (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
> +		dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
> +		return -EINVAL;
> +	}
> +
> +	mutex_init(&data->msi_map_lock);
> +
> +	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);

IIUC, num_irqs is between 0 and (SG2042_VECTOR_MAX - SG2042_VECTOR_MIN) 
(maybe + or -1).
So around 32.

Would it make sence to use DECLARE_BITMAP(msi_map, <correct_size>) in 
sg2042_msi_data to avoid this allocation and an indirection at runtime?

> +	if (!data->msi_map)
> +		return -ENOMEM;
> +
> +	ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
> +	if (ret)
> +		bitmap_free(data->msi_map);
> +
> +	return ret;
> +}

...

CJ

Chen Wang Dec. 12, 2024, 12:19 a.m. UTC | #9

On 2024/12/12 0:32, Christophe JAILLET wrote:
> Le 09/12/2024 à 08:12, Chen Wang a écrit :
>> From: Chen Wang <[e-mail]>
>>
>> Add driver for Sophgo SG2042 MSI interrupt controller.
>>
>> Signed-off-by: Chen Wang <[e-mail]>
>
> ...
>
>> +#define SG2042_VECTOR_MIN    64
>> +#define SG2042_VECTOR_MAX    95
>
> ...
>
>> +static struct irq_chip sg2042_msi_middle_irq_chip = {
>
> const?
Yes, I will add this in next version, thanks.
>
>> +    .name            = "SG2042 MSI",
>> +    .irq_ack        = sg2042_msi_irq_ack,
>> +    .irq_mask        = irq_chip_mask_parent,
>> +    .irq_unmask        = irq_chip_unmask_parent,
>> +#ifdef CONFIG_SMP
>> +    .irq_set_affinity    = irq_chip_set_affinity_parent,
>> +#endif
>> +    .irq_compose_msi_msg    = sg2042_msi_irq_compose_msi_msg,
>> +};
>
> ...
>
>> +static int sg2042_msi_probe(struct platform_device *pdev)
>> +{
>> +    struct of_phandle_args args = {};
>> +    struct sg2042_msi_data *data;
>> +    int ret;
>> +
>> +    data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), 
>> GFP_KERNEL);
>> +    if (!data)
>> +        return -ENOMEM;
>> +
>> +    data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
>> +    if (IS_ERR(data->reg_clr)) {
>> +        dev_err(&pdev->dev, "Failed to map clear register\n");
>> +        return PTR_ERR(data->reg_clr);
>> +    }
>> +
>> +    if (of_property_read_u64(pdev->dev.of_node, 
>> "sophgo,msi-doorbell-addr",
>> +                 &data->doorbell_addr)) {
>> +        dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
>> +        return -EINVAL;
>> +    }
>> +
>> +    ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
>> +                     "#interrupt-cells", 0, &args);
>> +    if (ret) {
>> +        dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
>> +        return ret;
>> +    }
>> +    data->irq_first = (u32)args.args[0];
>> +
>> +    ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
>> +                     args.args_count + 1, &data->num_irqs);
>> +    if (ret) {
>> +        dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
>> +        return ret;
>> +    }
>> +
>> +    if (data->irq_first < SG2042_VECTOR_MIN ||
>> +        (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
>> +        dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
>> +        return -EINVAL;
>> +    }
>> +
>> +    mutex_init(&data->msi_map_lock);
>> +
>> +    data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);
>
> IIUC, num_irqs is between 0 and (SG2042_VECTOR_MAX - 
> SG2042_VECTOR_MIN) (maybe + or -1).
> So around 32.
>
> Would it make sence to use DECLARE_BITMAP(msi_map, <correct_size>) in 
> sg2042_msi_data to avoid this allocation and an indirection at runtime?

This is also a good choice. I will double check this.

Thanks,

Chen

>
>> +    if (!data->msi_map)
>> +        return -ENOMEM;
>> +
>> +    ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
>> +    if (ret)
>> +        bitmap_free(data->msi_map);
>> +
>> +    return ret;
>> +}
>
> ...
>
> CJ

Patch

diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
index 9bee02db1643..161fb5df857f 100644
--- a/drivers/irqchip/Kconfig
+++ b/drivers/irqchip/Kconfig
@@ -749,6 +749,18 @@  config MCHP_EIC
 	help
 	  Support for Microchip External Interrupt Controller.
 
+config SOPHGO_SG2042_MSI
+	bool "Sophgo SG2042 MSI Controller"
+	depends on ARCH_SOPHGO || COMPILE_TEST
+	depends on PCI
+	select IRQ_DOMAIN_HIERARCHY
+	select IRQ_MSI_LIB
+	select PCI_MSI
+	help
+	  Support for the Sophgo SG2042 MSI Controller.
+	  This on-chip interrupt controller enables MSI sources to be
+	  routed to the primary PLIC controller on SoC.
+
 config SUNPLUS_SP7021_INTC
 	bool "Sunplus SP7021 interrupt controller" if COMPILE_TEST
 	default SOC_SP7021
diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index 25e9ad29b8c4..dd60e597491d 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -128,4 +128,5 @@  obj-$(CONFIG_WPCM450_AIC)		+= irq-wpcm450-aic.o
 obj-$(CONFIG_IRQ_IDT3243X)		+= irq-idt3243x.o
 obj-$(CONFIG_APPLE_AIC)			+= irq-apple-aic.o
 obj-$(CONFIG_MCHP_EIC)			+= irq-mchp-eic.o
+obj-$(CONFIG_SOPHGO_SG2042_MSI)		+= irq-sg2042-msi.o
 obj-$(CONFIG_SUNPLUS_SP7021_INTC)	+= irq-sp7021-intc.o
diff --git a/drivers/irqchip/irq-sg2042-msi.c b/drivers/irqchip/irq-sg2042-msi.c
new file mode 100644
index 000000000000..495ee2b45eb2
--- /dev/null
+++ b/drivers/irqchip/irq-sg2042-msi.c
@@ -0,0 +1,285 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * SG2042 MSI Controller
+ *
+ * Copyright (C) 2024 Sophgo Technology Inc.
+ * Copyright (C) 2024 Chen Wang <[e-mail]>
+ */
+
+#include <linux/cleanup.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/msi.h>
+#include <linux/of.h>
+#include <linux/of_irq.h>
+#include <linux/of_pci.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+#include "irq-msi-lib.h"
+
+#define SG2042_VECTOR_MIN	64
+#define SG2042_VECTOR_MAX	95
+
+struct sg2042_msi_data {
+	void __iomem	*reg_clr;	// clear reg, see TRM, 10.1.33, GP_INTR0_CLR
+
+	u64		doorbell_addr;	// see TRM, 10.1.32, GP_INTR0_SET
+
+	u32		irq_first;	// The vector number that MSIs starts
+	u32		num_irqs;	// The number of vectors for MSIs
+
+	unsigned long	*msi_map;
+	struct mutex	msi_map_lock;	// lock for msi_map
+};
+
+static int sg2042_msi_allocate_hwirq(struct sg2042_msi_data *priv, int num_req)
+{
+	int first;
+
+	guard(mutex)(&priv->msi_map_lock);
+	first = bitmap_find_free_region(priv->msi_map, priv->num_irqs,
+					get_count_order(num_req));
+	return first >= 0 ? priv->irq_first + first : -ENOSPC;
+}
+
+static void sg2042_msi_free_hwirq(struct sg2042_msi_data *priv,
+				  int hwirq, int num_req)
+{
+	int first = hwirq - priv->irq_first;
+
+	guard(mutex)(&priv->msi_map_lock);
+	bitmap_release_region(priv->msi_map, first, get_count_order(num_req));
+}
+
+static void sg2042_msi_irq_ack(struct irq_data *d)
+{
+	struct sg2042_msi_data *data  = irq_data_get_irq_chip_data(d);
+	int bit_off = d->hwirq - data->irq_first;
+
+	writel(1 << bit_off, (unsigned int *)data->reg_clr);
+
+	irq_chip_ack_parent(d);
+}
+
+static void sg2042_msi_irq_compose_msi_msg(struct irq_data *data,
+					   struct msi_msg *msg)
+{
+	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(data);
+
+	msg->address_hi = upper_32_bits(priv->doorbell_addr);
+	msg->address_lo = lower_32_bits(priv->doorbell_addr);
+	msg->data = 1 << (data->hwirq - priv->irq_first);
+
+	pr_debug("%s hwirq[%ld]: address_hi[%#x], address_lo[%#x], data[%#x]\n",
+		 __func__, data->hwirq, msg->address_hi, msg->address_lo, msg->data);
+}
+
+static struct irq_chip sg2042_msi_middle_irq_chip = {
+	.name			= "SG2042 MSI",
+	.irq_ack		= sg2042_msi_irq_ack,
+	.irq_mask		= irq_chip_mask_parent,
+	.irq_unmask		= irq_chip_unmask_parent,
+#ifdef CONFIG_SMP
+	.irq_set_affinity	= irq_chip_set_affinity_parent,
+#endif
+	.irq_compose_msi_msg	= sg2042_msi_irq_compose_msi_msg,
+};
+
+static int sg2042_msi_parent_domain_alloc(struct irq_domain *domain,
+					  unsigned int virq, int hwirq)
+{
+	struct irq_fwspec fwspec;
+	struct irq_data *d;
+	int ret;
+
+	fwspec.fwnode = domain->parent->fwnode;
+	fwspec.param_count = 2;
+	fwspec.param[0] = hwirq;
+	fwspec.param[1] = IRQ_TYPE_EDGE_RISING;
+
+	ret = irq_domain_alloc_irqs_parent(domain, virq, 1, &fwspec);
+	if (ret)
+		return ret;
+
+	d = irq_domain_get_irq_data(domain->parent, virq);
+	return d->chip->irq_set_type(d, IRQ_TYPE_EDGE_RISING);
+}
+
+static int sg2042_msi_middle_domain_alloc(struct irq_domain *domain,
+					  unsigned int virq,
+					  unsigned int nr_irqs, void *args)
+{
+	struct sg2042_msi_data *priv = domain->host_data;
+	int hwirq, err, i;
+
+	hwirq = sg2042_msi_allocate_hwirq(priv, nr_irqs);
+	if (hwirq < 0)
+		return hwirq;
+
+	for (i = 0; i < nr_irqs; i++) {
+		err = sg2042_msi_parent_domain_alloc(domain, virq + i, hwirq + i);
+		if (err)
+			goto err_hwirq;
+
+		pr_debug("%s: virq[%d], hwirq[%d]\n", __func__, virq + i, hwirq + i);
+
+		irq_domain_set_hwirq_and_chip(domain, virq + i, hwirq + i,
+					      &sg2042_msi_middle_irq_chip, priv);
+	}
+
+	return 0;
+
+err_hwirq:
+	sg2042_msi_free_hwirq(priv, hwirq, nr_irqs);
+	irq_domain_free_irqs_parent(domain, virq, i);
+
+	return err;
+}
+
+static void sg2042_msi_middle_domain_free(struct irq_domain *domain,
+					  unsigned int virq,
+					  unsigned int nr_irqs)
+{
+	struct irq_data *d = irq_domain_get_irq_data(domain, virq);
+	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(d);
+
+	irq_domain_free_irqs_parent(domain, virq, nr_irqs);
+	sg2042_msi_free_hwirq(priv, d->hwirq, nr_irqs);
+}
+
+static const struct irq_domain_ops sg2042_msi_middle_domain_ops = {
+	.alloc	= sg2042_msi_middle_domain_alloc,
+	.free	= sg2042_msi_middle_domain_free,
+	.select	= msi_lib_irq_domain_select,
+};
+
+#define SG2042_MSI_FLAGS_REQUIRED (MSI_FLAG_USE_DEF_DOM_OPS |	\
+				   MSI_FLAG_USE_DEF_CHIP_OPS)
+
+#define SG2042_MSI_FLAGS_SUPPORTED MSI_GENERIC_FLAGS_MASK
+
+static struct msi_parent_ops sg2042_msi_parent_ops = {
+	.required_flags		= SG2042_MSI_FLAGS_REQUIRED,
+	.supported_flags	= SG2042_MSI_FLAGS_SUPPORTED,
+	.bus_select_mask	= MATCH_PCI_MSI,
+	.bus_select_token	= DOMAIN_BUS_NEXUS,
+	.prefix			= "SG2042-",
+	.init_dev_msi_info	= msi_lib_init_dev_msi_info,
+};
+
+static int sg2042_msi_init_domains(struct sg2042_msi_data *priv,
+				   struct device_node *node)
+{
+	struct fwnode_handle *fwnode = of_node_to_fwnode(node);
+	struct irq_domain *plic_domain, *middle_domain;
+	struct device_node *plic_node;
+
+	if (!of_find_property(node, "interrupt-parent", NULL)) {
+		pr_err("Can't find interrupt-parent!\n");
+		return -EINVAL;
+	}
+
+	plic_node = of_irq_find_parent(node);
+	if (!plic_node) {
+		pr_err("Failed to find the PLIC node!\n");
+		return -ENXIO;
+	}
+
+	plic_domain = irq_find_host(plic_node);
+	of_node_put(plic_node);
+	if (!plic_domain) {
+		pr_err("Failed to find the PLIC domain\n");
+		return -ENXIO;
+	}
+
+	middle_domain = irq_domain_create_hierarchy(plic_domain, 0, priv->num_irqs,
+						    fwnode,
+						    &sg2042_msi_middle_domain_ops,
+						    priv);
+	if (!middle_domain) {
+		pr_err("Failed to create the MSI middle domain\n");
+		return -ENOMEM;
+	}
+
+	irq_domain_update_bus_token(middle_domain, DOMAIN_BUS_NEXUS);
+
+	middle_domain->flags |= IRQ_DOMAIN_FLAG_MSI_PARENT;
+	middle_domain->msi_parent_ops = &sg2042_msi_parent_ops;
+
+	return 0;
+}
+
+static int sg2042_msi_probe(struct platform_device *pdev)
+{
+	struct of_phandle_args args = {};
+	struct sg2042_msi_data *data;
+	int ret;
+
+	data = devm_kzalloc(&pdev->dev, sizeof(struct sg2042_msi_data), GFP_KERNEL);
+	if (!data)
+		return -ENOMEM;
+
+	data->reg_clr = devm_platform_ioremap_resource_byname(pdev, "clr");
+	if (IS_ERR(data->reg_clr)) {
+		dev_err(&pdev->dev, "Failed to map clear register\n");
+		return PTR_ERR(data->reg_clr);
+	}
+
+	if (of_property_read_u64(pdev->dev.of_node, "sophgo,msi-doorbell-addr",
+				 &data->doorbell_addr)) {
+		dev_err(&pdev->dev, "Unable to parse MSI doorbell addr\n");
+		return -EINVAL;
+	}
+
+	ret = of_parse_phandle_with_args(pdev->dev.of_node, "msi-ranges",
+					 "#interrupt-cells", 0, &args);
+	if (ret) {
+		dev_err(&pdev->dev, "Unable to parse MSI vec base\n");
+		return ret;
+	}
+	data->irq_first = (u32)args.args[0];
+
+	ret = of_property_read_u32_index(pdev->dev.of_node, "msi-ranges",
+					 args.args_count + 1, &data->num_irqs);
+	if (ret) {
+		dev_err(&pdev->dev, "Unable to parse MSI vec number\n");
+		return ret;
+	}
+
+	if (data->irq_first < SG2042_VECTOR_MIN ||
+	    (data->irq_first + data->num_irqs - 1) > SG2042_VECTOR_MAX) {
+		dev_err(&pdev->dev, "msi-ranges is incorrect!\n");
+		return -EINVAL;
+	}
+
+	mutex_init(&data->msi_map_lock);
+
+	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);
+	if (!data->msi_map)
+		return -ENOMEM;
+
+	ret = sg2042_msi_init_domains(data, pdev->dev.of_node);
+	if (ret)
+		bitmap_free(data->msi_map);
+
+	return ret;
+}
+
+static const struct of_device_id sg2042_msi_of_match[] = {
+	{ .compatible	= "sophgo,sg2042-msi" },
+	{}
+};
+
+static struct platform_driver sg2042_msi_driver = {
+	.driver = {
+		.name		= "sg2042-msi",
+		.of_match_table	= of_match_ptr(sg2042_msi_of_match),
+	},
+	.probe = sg2042_msi_probe,
+};
+builtin_platform_driver(sg2042_msi_driver);