[PATCH] HID: i2c-hid: add mutex protecting open/close race

We should not enter close function while someone else is in open.
This mutex prevents this race.

There is also no need to override the ret value with -EIO in case of
a failure of i2c_hid_set_power.

Signed-off-by: Benjamin Tissoires <redacted>
---

Hi Jiri, hi Jean,

this is the last patch that came from the last review. Sorry I took so long to
do it, but I thought it would have raised other problems. In the end, the
changes are pretty straightforward, and it would be great to have this one
in 3.8-rc0 too.
Anyway, Jiri, the choice is yours.

Cheers,
Benjamin

 drivers/hid/i2c-hid/i2c-hid.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/hid/i2c-hid/i2c-hid.c b/drivers/hid/i2c-hid/i2c-hid.c
index 6e1774c..74dfc33 100644
--- a/drivers/hid/i2c-hid/i2c-hid.c
+++ b/drivers/hid/i2c-hid/i2c-hid.c

@@ -117,6 +117,8 @@ static const struct i2c_hid_cmd hid_set_power_cmd =	{ I2C_HID_CMD(0x08) };
  * static const struct i2c_hid_cmd hid_set_protocol_cmd = { I2C_HID_CMD(0x07) };
  */
 
+static DEFINE_MUTEX(i2c_hid_open_mut);
+
 /* The main device structure */
 struct i2c_hid {
 	struct i2c_client	*client;	/* i2c client */

@@ -641,17 +643,20 @@ static int i2c_hid_open(struct hid_device *hid)
 {
 	struct i2c_client *client = hid->driver_data;
 	struct i2c_hid *ihid = i2c_get_clientdata(client);
-	int ret;
+	int ret = 0;
 
+	mutex_lock(&i2c_hid_open_mut);
 	if (!hid->open++) {
 		ret = i2c_hid_set_power(client, I2C_HID_PWR_ON);
 		if (ret) {
 			hid->open--;
-			return -EIO;
+			goto done;
 		}
 		set_bit(I2C_HID_STARTED, &ihid->flags);
 	}
-	return 0;
+done:
+	mutex_unlock(&i2c_hid_open_mut);
+	return ret;
 }
 
 static void i2c_hid_close(struct hid_device *hid)

@@ -663,12 +668,14 @@ static void i2c_hid_close(struct hid_device *hid)
 	 * data acquistion due to a resumption we no longer
 	 * care about
 	 */
+	mutex_lock(&i2c_hid_open_mut);
 	if (!--hid->open) {
 		clear_bit(I2C_HID_STARTED, &ihid->flags);
 
 		/* Save some power */
 		i2c_hid_set_power(client, I2C_HID_PWR_SLEEP);
 	}
+	mutex_unlock(&i2c_hid_open_mut);
 }
 
 static int i2c_hid_power(struct hid_device *hid, int lvl)

-- 
1.8.0.2

Hi Benjamin,

On Wed, 12 Dec 2012 17:12:16 +0100, Benjamin Tissoires wrote:

We should not enter close function while someone else is in open.
This mutex prevents this race.

There is also no need to override the ret value with -EIO in case of
a failure of i2c_hid_set_power.

Signed-off-by: Benjamin Tissoires <redacted>
---

Looks good to me except that you forgot to include <linux/mutex.h> and
this could cause build failures on some architectures/configs.

Once this is fixed, you can add:

Reviewed-by: Jean Delvare <redacted>

-- 
Jean Delvare

On Wed, Dec 12, 2012 at 5:34 PM, Jean Delvare [off-list ref] wrote:

Hi Benjamin,

On Wed, 12 Dec 2012 17:12:16 +0100, Benjamin Tissoires wrote:

quoted

We should not enter close function while someone else is in open.
This mutex prevents this race.

There is also no need to override the ret value with -EIO in case of
a failure of i2c_hid_set_power.

Signed-off-by: Benjamin Tissoires <redacted>
---

Looks good to me except that you forgot to include <linux/mutex.h> and
this could cause build failures on some architectures/configs.

Once this is fixed, you can add:

Reviewed-by: Jean Delvare <redacted>

Thanks, I will resend it right now.

Cheers,
Benjamin

--
Jean Delvare