Re: asus_atk0110 not working on Asus P7P55D PRO

[Luca Tettamanti]

Il Sun, Sep 20, 2009 at 11:47:25AM -0600, Robert Hancock ha scritto: 
> On Sun, Sep 20, 2009 at 10:23 AM, Luca Tettamanti <kronos.it@xxxxxxxxx> wrote:
> > Il Wed, Sep 16, 2009 at 11:28:39PM -0600, Robert Hancock ha scritto:
> >> Just built a new system with an Asus P7P55D PRO motherboard. The
> >> ATK0110 driver doesn't seem to be able to retrieve any hardware
> >> monitoring parameters successfully, sensors gives:
> >>
> >> atk0110-acpi-0
> >> Adapter: ACPI interface
> >> ERROR: Can't get value of subfeature in0_input: I/O error
> >> Vcore Voltage:      +0.00 V  (min =  +0.80 V, max =  +1.60 V)
> >> ERROR: Can't get value of subfeature in1_input: I/O error
> >> +3.3V Voltage:      +0.00 V  (min =  +2.97 V, max =  +3.63 V)
> >> ERROR: Can't get value of subfeature in2_input: I/O error
> >> +5V Voltage:        +0.00 V  (min =  +4.50 V, max =  +5.50 V)
> >>
> >> etc. and dmesg spits out a bunch of these:
> >>
> >> ATK0110 ATK0110:00: atk_read_value_new: ACPI exception: AE_BUFFER_OVERFLOW
> >>
> >> I'm guessing this board uses a different format than what the driver
> >> is expecting. I'm attaching the gzipped decompiled DSDT from the
> >> board, hopefully it's useful to somebody..
> >
> > Please try the following patch, it should detect the proper buffer size.
> >
> 
> Obviously something not quite right:

Ah yes, the pointer for the output buffer was pointing to the wrong
variable. Sorry for that ;)


diff --git a/drivers/hwmon/asus_atk0110.c b/drivers/hwmon/asus_atk0110.c
index fe4fa29..f1056cf 100644
--- a/drivers/hwmon/asus_atk0110.c
+++ b/drivers/hwmon/asus_atk0110.c
@@ -75,6 +75,13 @@ enum atk_pack_member {
 #define _HWMON_OLD_PACK_ENABLE	4
 
 
+struct atk_acpi_out_buffer {
+	union acpi_object buf;
+	u32 flags;
+	u32 value;
+	u8 data[];
+};
+
 struct atk_data {
 	struct device *hwmon_dev;
 	acpi_handle atk_handle;
@@ -94,6 +101,10 @@ struct atk_data {
 	int temperature_count;
 	int fan_count;
 	struct list_head sensor_list;
+
+	struct atk_acpi_out_buffer *buffer;
+	size_t buffer_size;
+	struct mutex buffer_lock;
 };
 
 
@@ -129,11 +140,6 @@ struct atk_sensor_data {
 	char const *acpi_name;
 };
 
-struct atk_acpi_buffer_u64 {
-	union acpi_object buf;
-	u64 value;
-};
-
 static int atk_add(struct acpi_device *device);
 static int atk_remove(struct acpi_device *device, int type);
 static void atk_print_sensor(struct atk_data *data, union acpi_object *obj);
@@ -446,8 +452,8 @@ static int atk_read_value_new(struct atk_sensor_data *sensor, u64 *value)
 	struct acpi_object_list params;
 	struct acpi_buffer ret;
 	union acpi_object id;
-	struct atk_acpi_buffer_u64 tmp;
 	acpi_status status;
+	int err = 0;
 
 	id.type = ACPI_TYPE_INTEGER;
 	id.integer.value = sensor->id;
@@ -455,36 +461,35 @@ static int atk_read_value_new(struct atk_sensor_data *sensor, u64 *value)
 	params.count = 1;
 	params.pointer = &id;
 
-	tmp.buf.type = ACPI_TYPE_BUFFER;
-	tmp.buf.buffer.pointer = (u8 *)&tmp.value;
-	tmp.buf.buffer.length = sizeof(u64);
-	ret.length = sizeof(tmp);
-	ret.pointer = &tmp;
+	mutex_lock(&data->buffer_lock);
+
+	ret.length = data->buffer_size;
+	ret.pointer = data->buffer;
 
 	status = acpi_evaluate_object_typed(data->read_handle, NULL, &params,
 			&ret, ACPI_TYPE_BUFFER);
 	if (status != AE_OK) {
 		dev_warn(dev, "%s: ACPI exception: %s\n", __func__,
 				acpi_format_exception(status));
-		return -EIO;
+		err = -EIO;
+		goto out;
 	}
 
-	/* Return buffer format:
-	 * [0-3] "value" is valid flag
-	 * [4-7] value
-	 */
-	if (!(tmp.value & 0xffffffff)) {
+	if (!data->buffer->flags) {
 		/* The reading is not valid, possible causes:
 		 * - sensor failure
 		 * - enumeration was FUBAR (and we didn't notice)
 		 */
-		dev_info(dev, "Failure: %#llx\n", tmp.value);
-		return -EIO;
-	}
+		dev_info(dev, "Failure: %#x\n", data->buffer->flags);
 
-	*value = (tmp.value & 0xffffffff00000000ULL) >> 32;
+		err = -EIO;
+		goto out;
+	}
 
-	return 0;
+	*value = data->buffer->value;
+out:
+	mutex_unlock(&data->buffer_lock);
+	return err;
 }
 
 static int atk_read_value(struct atk_sensor_data *sensor, u64 *value)
@@ -721,11 +726,40 @@ static int atk_enumerate_new_hwmon(struct atk_data *data)
 	struct acpi_object_list params;
 	union acpi_object id;
 	union acpi_object *pack;
+	union acpi_object *asbf;
 	int err;
 	int i;
 
 	dev_dbg(dev, "Enumerating hwmon sensors\n");
 
+	/* Probe ASBF */
+	buf.length = ACPI_ALLOCATE_BUFFER;
+	ret = acpi_evaluate_object_typed(data->atk_handle, "ASBF", NULL, &buf,
+			ACPI_TYPE_BUFFER);
+	if (ret != AE_OK) {
+		dev_warn(dev, "Failed to evaluate ASBF: %s\n",
+				acpi_format_exception(ret));
+		return -ENODEV;
+	}
+	asbf = buf.pointer;
+	data->buffer_size = asbf->buffer.length;
+	ACPI_FREE(buf.pointer);
+	buf.pointer = NULL;
+
+	dev_dbg(dev, "ASBF buffer size: %zu\n", data->buffer_size);
+	/* Sanity check */
+	if (data->buffer_size < 8 || data->buffer_size > 512) {
+		dev_warn(dev, "Invalid ASBF buffer size: %zu\n",
+				data->buffer_size);
+		return -EINVAL;
+	}
+	data->buffer_size += sizeof(union acpi_object);
+
+	data->buffer = kzalloc(data->buffer_size, GFP_KERNEL);
+	if (!data->buffer)
+		return -ENOMEM;
+	mutex_init(&data->buffer_lock);
+
 	id.type = ACPI_TYPE_INTEGER;
 	id.integer.value = ATK_MUX_HWMON;
 	params.count = 1;
@@ -737,7 +771,8 @@ static int atk_enumerate_new_hwmon(struct atk_data *data)
 	if (ret != AE_OK) {
 		dev_warn(dev, METHOD_ENUMERATE ": ACPI exception: %s\n",
 				acpi_format_exception(ret));
-		return -ENODEV;
+		err = -ENODEV;
+		goto out;
 	}
 
 	/* Result must be a package */
@@ -759,6 +794,8 @@ static int atk_enumerate_new_hwmon(struct atk_data *data)
 	err = data->voltage_count + data->temperature_count + data->fan_count;
 
 out:
+	if (err < 0)
+		kfree(data->buffer);
 	ACPI_FREE(buf.pointer);
 	return err;
 }
@@ -988,6 +1025,7 @@ static int atk_remove(struct acpi_device *device, int type)
 	atk_free_sensors(data);
 	hwmon_device_unregister(data->hwmon_dev);
 
+	kfree(data->buffer);
 	kfree(data);
 
 	return 0;


Luca
-- 
The trouble with computers is that they do what you tell them, 
not what you want.
D. Cohen
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/