Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1565498 > unrolled thread

[PATCH V6 0/5] LPC: legacy ISA I/O support

Started by"zhichang.yuan" <yuanzhichang@hisilicon.com>
First post2017-01-24 07:40 +0100
Last post2017-02-04 14:30 +0100
Articles 4 on this page of 24 — 7 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH V6 0/5] LPC: legacy ISA I/O support "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-01-24 07:40 +0100
    [PATCH V6 4/5] LPC: Support the device-tree LPC host on Hip06/Hip07 "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-01-24 07:40 +0100
      Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 Rob Herring <robh@kernel.org> - 2017-01-27 23:20 +0100
      Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 Alexander Graf <agraf@suse.de> - 2017-01-30 21:20 +0100
        Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 John Garry <john.garry@huawei.com> - 2017-01-31 11:20 +0100
          Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 Alexander Graf <agraf@suse.de> - 2017-01-31 12:20 +0100
            Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 John Garry <john.garry@huawei.com> - 2017-01-31 13:00 +0100
            RE: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 Gabriele Paoloni <gabriele.paoloni@huawei.com> - 2017-01-31 13:00 +0100
        Re: [PATCH V6 4/5] LPC: Support the device-tree LPC host on  Hip06/Hip07 "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-02-13 15:50 +0100
    [PATCH V6 3/5] OF: Add missing I/O range exception for indirect-IO  devices "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-01-24 07:40 +0100
      Re: [PATCH V6 3/5] OF: Add missing I/O range exception for  indirect-IO  devices Rob Herring <robh@kernel.org> - 2017-01-27 23:10 +0100
        Re: [PATCH V6 3/5] OF: Add missing I/O range exception for  indirect-IO devices John Garry <john.garry@huawei.com> - 2017-01-30 10:10 +0100
          Re: [PATCH V6 3/5] OF: Add missing I/O range exception for  indirect-IO devices Arnd Bergmann <arnd@arndb.de> - 2017-01-30 11:20 +0100
    [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-01-24 07:40 +0100
      Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced Alexander Graf <agraf@suse.de> - 2017-01-30 18:20 +0100
        Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced John Garry <john.garry@huawei.com> - 2017-01-31 14:50 +0100
          Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced Alexander Graf <agraf@suse.de> - 2017-01-31 20:40 +0100
            RE: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced Gabriele Paoloni <gabriele.paoloni@huawei.com> - 2017-02-01 13:40 +0100
            Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-02-13 15:30 +0100
        Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-02-13 15:10 +0100
      Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced Bjorn Helgaas <helgaas@kernel.org> - 2017-01-31 01:20 +0100
        Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced John Garry <john.garry@huawei.com> - 2017-01-31 14:50 +0100
    [PATCH V5 5/5] LPC: Add the ACPI LPC support "zhichang.yuan" <yuanzhichang@hisilicon.com> - 2017-01-24 07:40 +0100
      Re: [PATCH V5 5/5] LPC: Add the ACPI LPC support John Garry <john.garry@huawei.com> - 2017-02-04 14:30 +0100

Page 2 of 2 — ← Prev page 1 [2]

#1570204 — Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced

FromBjorn Helgaas <helgaas@kernel.org>
Date2017-01-31 01:20 +0100
SubjectRe: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced
Message-ID<t5ukV-2bV-13@gated-at.bofh.it>
In reply to#1565502
On Tue, Jan 24, 2017 at 03:05:21PM +0800, zhichang.yuan wrote:
> Low-pin-count interface is integrated into some SoCs. The accesses to those
> peripherals under LPC make use of I/O ports rather than the memory mapped I/O.
> 
> To drive these devices, this patch introduces a method named indirect-IO.

It's slightly confusing to call this "indirect I/O" and then use
"extio" for the filename and function prefix.  It'd be nice to use
related names.

> +struct extio_node {
> +	unsigned long bus_start;	/* bus start address */
> +	unsigned long io_start;	/* io port token corresponding to bus_start */
> +	size_t range_size;	/* size of the extio node operating range */
> +	struct fwnode_handle *fwnode;
> +	struct list_head list;
> +	struct extio_ops *ops;	/* ops operating on this node */
> +	void *devpara;	/* private parameter of the host device */
> +};

I wish we didn't have both struct io_range and struct extio_node.  It
seems like they're both sort of trying to do the same thing.  Maybe
this is the same as what Alex is saying.

Bjorn

[toc] | [prev] | [next] | [standalone]

#1570763 — Re: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced

FromJohn Garry <john.garry@huawei.com>
Date2017-01-31 14:50 +0100
SubjectRe: [PATCH V6 1/5] LIB: Indirect ISA/LPC port IO introduced
Message-ID<t5GYO-1ja-15@gated-at.bofh.it>
In reply to#1570204
On 31/01/2017 00:09, Bjorn Helgaas wrote:
> On Tue, Jan 24, 2017 at 03:05:21PM +0800, zhichang.yuan wrote:
>> Low-pin-count interface is integrated into some SoCs. The accesses to those
>> peripherals under LPC make use of I/O ports rather than the memory mapped I/O.
>>
>> To drive these devices, this patch introduces a method named indirect-IO.
>
> It's slightly confusing to call this "indirect I/O" and then use
> "extio" for the filename and function prefix.  It'd be nice to use
> related names.

We will consider something more consistent.

>
>> +struct extio_node {
>> +	unsigned long bus_start;	/* bus start address */
>> +	unsigned long io_start;	/* io port token corresponding to bus_start */
>> +	size_t range_size;	/* size of the extio node operating range */
>> +	struct fwnode_handle *fwnode;
>> +	struct list_head list;
>> +	struct extio_ops *ops;	/* ops operating on this node */
>> +	void *devpara;	/* private parameter of the host device */
>> +};
>
> I wish we didn't have both struct io_range and struct extio_node.  It
> seems like they're both sort of trying to do the same thing.  Maybe
> this is the same as what Alex is saying.
>

I think so. I have just replied to Alex regarding this.

> Bjorn
>

Thanks,
John

> .
>

[toc] | [prev] | [next] | [standalone]

#1565503 — [PATCH V5 5/5] LPC: Add the ACPI LPC support

From"zhichang.yuan" <yuanzhichang@hisilicon.com>
Date2017-01-24 07:40 +0100
Subject[PATCH V5 5/5] LPC: Add the ACPI LPC support
Message-ID<t32VQ-Mf-25@gated-at.bofh.it>
In reply to#1565498
The patch update the _CRS of LPC children with the system logical I/O resource
after the translation from LPC-local I/O. Then the ACPI platform device
enumeration for LPC can apply the right I/O resource to request the system I/O
space.

Signed-off-by: zhichang.yuan <yuanzhichang@hisilicon.com>
---
 drivers/bus/hisi_lpc.c |  26 ++++++
 include/linux/extio.h  |   4 +
 lib/extio.c            | 228 +++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 258 insertions(+)

diff --git a/drivers/bus/hisi_lpc.c b/drivers/bus/hisi_lpc.c
index a96e384..8d52666 100644
--- a/drivers/bus/hisi_lpc.c
+++ b/drivers/bus/hisi_lpc.c
@@ -583,6 +583,32 @@ static int hisilpc_bus_platform_notify(struct notifier_block *nb,
 	/* register the linux virtual IO range node to list. */
 	register_extio(io_node);
 
+	/*
+	 * For ACPI children, translate the bus-local I/O range to logical
+	 * I/O range and set it as the current resource before the children
+	 * are enumerated.
+	 */
+	if (has_acpi_companion(dev)) {
+		struct acpi_device *root, *child;
+
+		root = to_acpi_device_node(dev->fwnode);
+		/* For hisilpc, only care about the sons of host. */
+		list_for_each_entry(child, &root->children, node) {
+			ret = acpi_set_extio_resource(child, root);
+			if (ret) {
+				dev_err(dev, "set resource failed..\n");
+				break;
+			}
+		}
+
+		if (ret) {
+			list_del(&io_node->list);
+			kfree(io_node);
+			dev_err(dev, "notify handling failed..\n");
+			return NOTIFY_DONE;
+		}
+	}
+
 	return NOTIFY_OK;
 }
 
diff --git a/include/linux/extio.h b/include/linux/extio.h
index 2ca7eab..c07607e 100644
--- a/include/linux/extio.h
+++ b/include/linux/extio.h
@@ -20,6 +20,7 @@
 
 #ifdef __KERNEL__
 
+#include <linux/acpi.h>
 #include <linux/fwnode.h>
 
 struct extio_ops {
@@ -81,5 +82,8 @@ static inline struct extio_node *extio_find_node(struct fwnode_handle *node)
 #endif
 extern void register_extio(struct extio_node *node);
 
+extern int acpi_set_extio_resource(struct acpi_device *adev,
+		struct acpi_device *host);
+
 #endif /* __KERNEL__ */
 #endif /* __LINUX_EXTIO_H */
diff --git a/lib/extio.c b/lib/extio.c
index 46228de..47f5913 100644
--- a/lib/extio.c
+++ b/lib/extio.c
@@ -75,6 +75,234 @@ unsigned long extio_translate(struct fwnode_handle *node,
 	return port_id;
 }
 
+static inline bool acpi_extio_supported_resource(struct acpi_resource *res)
+{
+	switch (res->type) {
+	case ACPI_RESOURCE_TYPE_ADDRESS16:
+	case ACPI_RESOURCE_TYPE_ADDRESS32:
+	case ACPI_RESOURCE_TYPE_ADDRESS64:
+		return true;
+	}
+	return false;
+}
+
+static acpi_status acpi_count_extiores(struct acpi_resource *res,
+					   void *data)
+{
+	int *res_cnt = data;
+
+	if (acpi_extio_supported_resource(res) &&
+		!acpi_dev_filter_resource_type(res, IORESOURCE_IO))
+		(*res_cnt)++;
+
+	return AE_OK;
+}
+
+static acpi_status acpi_read_one_extiores(struct acpi_resource *res,
+		void *data)
+{
+	struct acpi_resource **resource = data;
+
+	if (acpi_extio_supported_resource(res) &&
+		!acpi_dev_filter_resource_type(res, IORESOURCE_IO)) {
+		memcpy((*resource), res, sizeof(struct acpi_resource));
+		(*resource)->length = sizeof(struct acpi_resource);
+		(*resource)->type = res->type;
+		(*resource)++;
+	}
+
+	return AE_OK;
+}
+
+static acpi_status
+acpi_build_extiores_template(struct acpi_device *adev,
+			struct acpi_buffer *buffer)
+{
+	acpi_handle handle = adev->handle;
+	struct acpi_resource *resource;
+	acpi_status status;
+	int res_cnt = 0;
+
+	status = acpi_walk_resources(handle, METHOD_NAME__PRS,
+				     acpi_count_extiores, &res_cnt);
+	if (ACPI_FAILURE(status) || !res_cnt) {
+		dev_err(&adev->dev, "can't evaluate _CRS: %d\n", status);
+		return -EINVAL;
+	}
+
+	buffer->length = sizeof(struct acpi_resource) * (res_cnt + 1) + 1;
+	buffer->pointer = kzalloc(buffer->length - 1, GFP_KERNEL);
+	if (!buffer->pointer)
+		return -ENOMEM;
+
+	resource = (struct acpi_resource *)buffer->pointer;
+	status = acpi_walk_resources(handle, METHOD_NAME__PRS,
+				     acpi_read_one_extiores, &resource);
+	if (ACPI_FAILURE(status)) {
+		kfree(buffer->pointer);
+		dev_err(&adev->dev, "can't evaluate _PRS: %d\n", status);
+		return -EINVAL;
+	}
+
+	resource->type = ACPI_RESOURCE_TYPE_END_TAG;
+	resource->length = sizeof(struct acpi_resource);
+
+	return 0;
+}
+
+static int acpi_translate_extiores(struct acpi_device *adev,
+		struct acpi_device *host, struct acpi_buffer *buffer)
+{
+	int res_cnt = (buffer->length - 1) / sizeof(struct acpi_resource) - 1;
+	struct acpi_resource *resource = buffer->pointer;
+	struct acpi_resource_address64 addr;
+	unsigned long sys_port;
+	struct device *dev = &adev->dev;
+
+	/* only one I/O resource now */
+	if (res_cnt != 1) {
+		dev_err(dev, "encode %d resources whose type is(%d)!\n",
+			res_cnt, resource->type);
+		return -EINVAL;
+	}
+
+	if (ACPI_FAILURE(acpi_resource_to_address64(resource, &addr))) {
+		dev_err(dev, "convert acpi resource(%d) as addr64 FAIL!\n",
+			resource->type);
+		return -EFAULT;
+	}
+
+	/* For indirect-IO, addr length must be fixed. (>0, 0, 0) */
+	if (!addr.address.address_length || addr.min_address_fixed ||
+		addr.max_address_fixed) {
+		dev_warn(dev, "variable I/O resource is invalid!\n");
+		return -EINVAL;
+	}
+
+	sys_port = extio_translate(&host->fwnode, addr.address.minimum);
+	if (sys_port == -1) {
+		dev_err(dev, "translate bus-addr(0x%llx) fail!\n",
+			addr.address.minimum);
+		return -EFAULT;
+	}
+
+	switch (resource->type) {
+	case ACPI_RESOURCE_TYPE_ADDRESS16:
+	{
+		struct acpi_resource_address16 *out_res;
+
+		out_res = &resource->data.address16;
+		out_res->address.minimum = sys_port;
+		out_res->address.maximum = sys_port +
+			addr.address.address_length - 1;
+
+		dev_info(dev, "_SRS 16IO: [0x%x - 0x%x]\n",
+			out_res->address.minimum,
+			out_res->address.maximum);
+
+		break;
+	}
+
+	case ACPI_RESOURCE_TYPE_ADDRESS32:
+	{
+		struct acpi_resource_address32 *out_res;
+
+		out_res = &resource->data.address32;
+		out_res->address.minimum = sys_port;
+		out_res->address.maximum = sys_port +
+			addr.address.address_length - 1;
+
+		dev_info(dev, "_SRS 32IO: [0x%x - 0x%x]\n",
+			out_res->address.minimum,
+			out_res->address.maximum);
+
+		break;
+	}
+
+	case ACPI_RESOURCE_TYPE_ADDRESS64:
+	{
+		struct acpi_resource_address64 *out_res;
+
+		out_res = &resource->data.address64;
+		out_res->address.minimum = sys_port;
+		out_res->address.maximum = sys_port +
+			addr.address.address_length - 1;
+
+		dev_info(dev, "_SRS 64IO: [0x%llx - 0x%llx]\n",
+			out_res->address.minimum,
+			out_res->address.maximum);
+
+		break;
+	}
+
+	default:
+		return -EINVAL;
+
+	}
+
+	return 0;
+}
+
+/*
+ * update/set the current I/O resource of the designated device node.
+ * after this calling, the enumeration can be started as the I/O resource
+ * had been translated to logicial I/O from bus-local I/O.
+ *
+ * @adev: the device node to be updated the I/O resource;
+ * @host: the device node where 'adev' is attached, which can be not
+ *	the parent of 'adev';
+ *
+ * return 0 when successful, negative is for failure.
+ */
+int acpi_set_extio_resource(struct acpi_device *adev,
+		struct acpi_device *host)
+{
+	struct device *dev = &adev->dev;
+	struct acpi_buffer buffer;
+	acpi_status status;
+	int ret;
+
+	if (!host)
+		return -EINVAL;
+
+	/* check the device state */
+	if (!adev->status.present) {
+		dev_info(dev, "ACPI: device is not present!\n");
+		return 0;
+	}
+	/* whether the child had been enumerated? */
+	if (acpi_device_enumerated(adev)) {
+		dev_info(dev, "ACPI: had been enumerated!\n");
+		return 0;
+	}
+
+	/* read the _PRS and convert as acpi_buffer */
+	status = acpi_build_extiores_template(adev, &buffer);
+	if (ACPI_FAILURE(status)) {
+		dev_warn(dev, "Failure evaluating %s\n",
+				METHOD_NAME__PRS);
+		return -ENODEV;
+	}
+
+	/* translate the I/O resources */
+	ret = acpi_translate_extiores(adev, host, &buffer);
+	if (ret) {
+		kfree(buffer.pointer);
+		dev_err(dev, "Translate I/O range FAIL!\n");
+		return ret;
+	}
+
+	/* set current resource... */
+	status = acpi_set_current_resources(adev->handle, &buffer);
+	kfree(buffer.pointer);
+	if (ACPI_FAILURE(status)) {
+		dev_err(dev, "Error evaluating _SRS (0x%x)\n", status);
+		ret = -EIO;
+	}
+
+	return ret;
+}
+
 #ifdef PCI_IOBASE
 
 #define BUILD_EXTIO(bw, type)						\
-- 
1.9.1

[toc] | [prev] | [next] | [standalone]

#1573670 — Re: [PATCH V5 5/5] LPC: Add the ACPI LPC support

FromJohn Garry <john.garry@huawei.com>
Date2017-02-04 14:30 +0100
SubjectRe: [PATCH V5 5/5] LPC: Add the ACPI LPC support
Message-ID<t78zD-10H-3@gated-at.bofh.it>
In reply to#1565503
+ linux-acpi

Adding linux-acpi list, which should have been originally included.

The patchset threads have had much discussion, here is a pointer: 
http://www.spinics.net/lists/devicetree/msg160611.html

John

On 24/01/2017 07:05, zhichang.yuan wrote:
> The patch update the _CRS of LPC children with the system logical I/O resource
> after the translation from LPC-local I/O. Then the ACPI platform device
> enumeration for LPC can apply the right I/O resource to request the system I/O
> space.
>
> Signed-off-by: zhichang.yuan <yuanzhichang@hisilicon.com>
> ---
>  drivers/bus/hisi_lpc.c |  26 ++++++
>  include/linux/extio.h  |   4 +
>  lib/extio.c            | 228 +++++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 258 insertions(+)
>
> diff --git a/drivers/bus/hisi_lpc.c b/drivers/bus/hisi_lpc.c
> index a96e384..8d52666 100644
> --- a/drivers/bus/hisi_lpc.c
> +++ b/drivers/bus/hisi_lpc.c
> @@ -583,6 +583,32 @@ static int hisilpc_bus_platform_notify(struct notifier_block *nb,
>  	/* register the linux virtual IO range node to list. */
>  	register_extio(io_node);
>
> +	/*
> +	 * For ACPI children, translate the bus-local I/O range to logical
> +	 * I/O range and set it as the current resource before the children
> +	 * are enumerated.
> +	 */
> +	if (has_acpi_companion(dev)) {
> +		struct acpi_device *root, *child;
> +
> +		root = to_acpi_device_node(dev->fwnode);
> +		/* For hisilpc, only care about the sons of host. */
> +		list_for_each_entry(child, &root->children, node) {
> +			ret = acpi_set_extio_resource(child, root);
> +			if (ret) {
> +				dev_err(dev, "set resource failed..\n");
> +				break;
> +			}
> +		}
> +
> +		if (ret) {
> +			list_del(&io_node->list);
> +			kfree(io_node);
> +			dev_err(dev, "notify handling failed..\n");
> +			return NOTIFY_DONE;
> +		}
> +	}
> +
>  	return NOTIFY_OK;
>  }
>
> diff --git a/include/linux/extio.h b/include/linux/extio.h
> index 2ca7eab..c07607e 100644
> --- a/include/linux/extio.h
> +++ b/include/linux/extio.h
> @@ -20,6 +20,7 @@
>
>  #ifdef __KERNEL__
>
> +#include <linux/acpi.h>
>  #include <linux/fwnode.h>
>
>  struct extio_ops {
> @@ -81,5 +82,8 @@ static inline struct extio_node *extio_find_node(struct fwnode_handle *node)
>  #endif
>  extern void register_extio(struct extio_node *node);
>
> +extern int acpi_set_extio_resource(struct acpi_device *adev,
> +		struct acpi_device *host);
> +
>  #endif /* __KERNEL__ */
>  #endif /* __LINUX_EXTIO_H */
> diff --git a/lib/extio.c b/lib/extio.c
> index 46228de..47f5913 100644
> --- a/lib/extio.c
> +++ b/lib/extio.c
> @@ -75,6 +75,234 @@ unsigned long extio_translate(struct fwnode_handle *node,
>  	return port_id;
>  }
>
> +static inline bool acpi_extio_supported_resource(struct acpi_resource *res)
> +{
> +	switch (res->type) {
> +	case ACPI_RESOURCE_TYPE_ADDRESS16:
> +	case ACPI_RESOURCE_TYPE_ADDRESS32:
> +	case ACPI_RESOURCE_TYPE_ADDRESS64:
> +		return true;
> +	}
> +	return false;
> +}
> +
> +static acpi_status acpi_count_extiores(struct acpi_resource *res,
> +					   void *data)
> +{
> +	int *res_cnt = data;
> +
> +	if (acpi_extio_supported_resource(res) &&
> +		!acpi_dev_filter_resource_type(res, IORESOURCE_IO))
> +		(*res_cnt)++;
> +
> +	return AE_OK;
> +}
> +
> +static acpi_status acpi_read_one_extiores(struct acpi_resource *res,
> +		void *data)
> +{
> +	struct acpi_resource **resource = data;
> +
> +	if (acpi_extio_supported_resource(res) &&
> +		!acpi_dev_filter_resource_type(res, IORESOURCE_IO)) {
> +		memcpy((*resource), res, sizeof(struct acpi_resource));
> +		(*resource)->length = sizeof(struct acpi_resource);
> +		(*resource)->type = res->type;
> +		(*resource)++;
> +	}
> +
> +	return AE_OK;
> +}
> +
> +static acpi_status
> +acpi_build_extiores_template(struct acpi_device *adev,
> +			struct acpi_buffer *buffer)
> +{
> +	acpi_handle handle = adev->handle;
> +	struct acpi_resource *resource;
> +	acpi_status status;
> +	int res_cnt = 0;
> +
> +	status = acpi_walk_resources(handle, METHOD_NAME__PRS,
> +				     acpi_count_extiores, &res_cnt);
> +	if (ACPI_FAILURE(status) || !res_cnt) {
> +		dev_err(&adev->dev, "can't evaluate _CRS: %d\n", status);
> +		return -EINVAL;
> +	}
> +
> +	buffer->length = sizeof(struct acpi_resource) * (res_cnt + 1) + 1;
> +	buffer->pointer = kzalloc(buffer->length - 1, GFP_KERNEL);
> +	if (!buffer->pointer)
> +		return -ENOMEM;
> +
> +	resource = (struct acpi_resource *)buffer->pointer;
> +	status = acpi_walk_resources(handle, METHOD_NAME__PRS,
> +				     acpi_read_one_extiores, &resource);
> +	if (ACPI_FAILURE(status)) {
> +		kfree(buffer->pointer);
> +		dev_err(&adev->dev, "can't evaluate _PRS: %d\n", status);
> +		return -EINVAL;
> +	}
> +
> +	resource->type = ACPI_RESOURCE_TYPE_END_TAG;
> +	resource->length = sizeof(struct acpi_resource);
> +
> +	return 0;
> +}
> +
> +static int acpi_translate_extiores(struct acpi_device *adev,
> +		struct acpi_device *host, struct acpi_buffer *buffer)
> +{
> +	int res_cnt = (buffer->length - 1) / sizeof(struct acpi_resource) - 1;
> +	struct acpi_resource *resource = buffer->pointer;
> +	struct acpi_resource_address64 addr;
> +	unsigned long sys_port;
> +	struct device *dev = &adev->dev;
> +
> +	/* only one I/O resource now */
> +	if (res_cnt != 1) {
> +		dev_err(dev, "encode %d resources whose type is(%d)!\n",
> +			res_cnt, resource->type);
> +		return -EINVAL;
> +	}
> +
> +	if (ACPI_FAILURE(acpi_resource_to_address64(resource, &addr))) {
> +		dev_err(dev, "convert acpi resource(%d) as addr64 FAIL!\n",
> +			resource->type);
> +		return -EFAULT;
> +	}
> +
> +	/* For indirect-IO, addr length must be fixed. (>0, 0, 0) */
> +	if (!addr.address.address_length || addr.min_address_fixed ||
> +		addr.max_address_fixed) {
> +		dev_warn(dev, "variable I/O resource is invalid!\n");
> +		return -EINVAL;
> +	}
> +
> +	sys_port = extio_translate(&host->fwnode, addr.address.minimum);
> +	if (sys_port == -1) {
> +		dev_err(dev, "translate bus-addr(0x%llx) fail!\n",
> +			addr.address.minimum);
> +		return -EFAULT;
> +	}
> +
> +	switch (resource->type) {
> +	case ACPI_RESOURCE_TYPE_ADDRESS16:
> +	{
> +		struct acpi_resource_address16 *out_res;
> +
> +		out_res = &resource->data.address16;
> +		out_res->address.minimum = sys_port;
> +		out_res->address.maximum = sys_port +
> +			addr.address.address_length - 1;
> +
> +		dev_info(dev, "_SRS 16IO: [0x%x - 0x%x]\n",
> +			out_res->address.minimum,
> +			out_res->address.maximum);
> +
> +		break;
> +	}
> +
> +	case ACPI_RESOURCE_TYPE_ADDRESS32:
> +	{
> +		struct acpi_resource_address32 *out_res;
> +
> +		out_res = &resource->data.address32;
> +		out_res->address.minimum = sys_port;
> +		out_res->address.maximum = sys_port +
> +			addr.address.address_length - 1;
> +
> +		dev_info(dev, "_SRS 32IO: [0x%x - 0x%x]\n",
> +			out_res->address.minimum,
> +			out_res->address.maximum);
> +
> +		break;
> +	}
> +
> +	case ACPI_RESOURCE_TYPE_ADDRESS64:
> +	{
> +		struct acpi_resource_address64 *out_res;
> +
> +		out_res = &resource->data.address64;
> +		out_res->address.minimum = sys_port;
> +		out_res->address.maximum = sys_port +
> +			addr.address.address_length - 1;
> +
> +		dev_info(dev, "_SRS 64IO: [0x%llx - 0x%llx]\n",
> +			out_res->address.minimum,
> +			out_res->address.maximum);
> +
> +		break;
> +	}
> +
> +	default:
> +		return -EINVAL;
> +
> +	}
> +
> +	return 0;
> +}
> +
> +/*
> + * update/set the current I/O resource of the designated device node.
> + * after this calling, the enumeration can be started as the I/O resource
> + * had been translated to logicial I/O from bus-local I/O.
> + *
> + * @adev: the device node to be updated the I/O resource;
> + * @host: the device node where 'adev' is attached, which can be not
> + *	the parent of 'adev';
> + *
> + * return 0 when successful, negative is for failure.
> + */
> +int acpi_set_extio_resource(struct acpi_device *adev,
> +		struct acpi_device *host)
> +{
> +	struct device *dev = &adev->dev;
> +	struct acpi_buffer buffer;
> +	acpi_status status;
> +	int ret;
> +
> +	if (!host)
> +		return -EINVAL;
> +
> +	/* check the device state */
> +	if (!adev->status.present) {
> +		dev_info(dev, "ACPI: device is not present!\n");
> +		return 0;
> +	}
> +	/* whether the child had been enumerated? */
> +	if (acpi_device_enumerated(adev)) {
> +		dev_info(dev, "ACPI: had been enumerated!\n");
> +		return 0;
> +	}
> +
> +	/* read the _PRS and convert as acpi_buffer */
> +	status = acpi_build_extiores_template(adev, &buffer);
> +	if (ACPI_FAILURE(status)) {
> +		dev_warn(dev, "Failure evaluating %s\n",
> +				METHOD_NAME__PRS);
> +		return -ENODEV;
> +	}
> +
> +	/* translate the I/O resources */
> +	ret = acpi_translate_extiores(adev, host, &buffer);
> +	if (ret) {
> +		kfree(buffer.pointer);
> +		dev_err(dev, "Translate I/O range FAIL!\n");
> +		return ret;
> +	}
> +
> +	/* set current resource... */
> +	status = acpi_set_current_resources(adev->handle, &buffer);
> +	kfree(buffer.pointer);
> +	if (ACPI_FAILURE(status)) {
> +		dev_err(dev, "Error evaluating _SRS (0x%x)\n", status);
> +		ret = -EIO;
> +	}
> +
> +	return ret;
> +}
> +
>  #ifdef PCI_IOBASE
>
>  #define BUILD_EXTIO(bw, type)						\
>

[toc] | [prev] | [standalone]

Page 2 of 2 — ← Prev page 1 [2]

Back to top | Article view | linux.kernel

csiph-web