Skip to contentSkip to conversation

leds-lp3944: properly handle lp3944_configure fail in lp3944_probe

PATCH Superseded Axel Lin 3 participants 4 messages May 17, 2010 LED

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Axel Lin

author
#0 lore raw

PATCH 1/1 leds-lp3944: properly handle lp3944_configure fail in lp3944_probe

In current implementation, lp3944_probe return 0 even if lp3944_configure fail.
Therefore, led_classdev_unregister will be executed twice ( in error handling of lp3944_configure and lp3944_remove ). This patch properly handles lp3944_configure fail in lp3944_probe.

---
 drivers/leds/leds-lp3944.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)
drivers/leds/leds-lp3944.c +9 -1
index 8d5ecce..03a24d0 100644
@@ -379,6 +379,7 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 {
 	struct lp3944_platform_data *lp3944_pdata = client->dev.platform_data;
 	struct lp3944_data *data;
+	int err;
 
 	if (lp3944_pdata == NULL) {
 		dev_err(&client->dev, "no platform data\n");
@@ -403,8 +404,15 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 
 	dev_info(&client->dev, "lp3944 enabled\n");
 
-	lp3944_configure(client, data, lp3944_pdata);
+	err = lp3944_configure(client, data, lp3944_pdata);
+	if (err < 0)
+		goto err_configure;
+
 	return 0;
+
+err_configure:
+	kfree(data);
+	return err;
 }
 
 static int __devexit lp3944_remove(struct i2c_client *client)
Signature · 1 line
1.5.4.3
#1 lore raw
Quoted 4 lines · On Mon, 17 May 2010 09:34:02 +0800 Axel Lin <[email protected]> wrote:

In current implementation, lp3944_probe return 0 even if lp3944_configure fail.
Therefore, led_classdev_unregister will be executed twice ( in error handling of lp3944_configure and lp3944_remove ). This patch properly handles lp3944_configure fail in lp3944_probe.

Hi Axel, thanks for the fix, I agree it's needed.
There are some minor comments inlined below.

Plus, when possible, I prefer commit messages to be wrapped to 72/80 characters per line. Please, consider this if you are sending a v2.

Ah, may I ask where you are using this driver? Just curious.

Quoted 18 lines
---
 drivers/leds/leds-lp3944.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)
drivers/leds/leds-lp3944.c +1 -0
index 8d5ecce..03a24d0 100644
@@ -379,6 +379,7 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 {
 	struct lp3944_platform_data *lp3944_pdata = client->dev.platform_data;
 	struct lp3944_data *data;
+	int err;
 
 	if (lp3944_pdata == NULL) {
 		dev_err(&client->dev, "no platform data\n");

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 410

Quoted 8 lines
drivers/leds/leds-lp3944.c +4 -1
@@ -403,8 +404,15 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 
 	dev_info(&client->dev, "lp3944 enabled\n");
 
-	lp3944_configure(client, data, lp3944_pdata);
+	err = lp3944_configure(client, data, lp3944_pdata);
+	if (err < 0)
+		goto err_configure;
+

The dev_info(&client->dev, "lp3944 enabled\n"); could now go right here, before the return 0, so we don't report the driver as enabled even in the case its probe fails.

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 414

Quoted 4 lines
drivers/leds/leds-lp3944.c +3 -0
 	return 0;
+
+err_configure:
+	kfree(data);

add i2c_set_clientdata(client, NULL) here just like in lp3944_remove()

Quoted 7 lines
drivers/leds/leds-lp3944.c +1 -0
+	return err;
 }
 
 static int __devexit lp3944_remove(struct i2c_client *client)
Signature · 1 line
1.5.4.3

Regards,
Antonio

Signature · 10 lines
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
#2 lore raw
Quoted 47 lines · Antonio Ospite
Quoted 4 lines · On Mon, 17 May 2010 09:34:02 +0800 Axel Lin <[email protected]> wrote:

In current implementation, lp3944_probe return 0 even if lp3944_configure fail.
Therefore, led_classdev_unregister will be executed twice ( in error handling of lp3944_configure and lp3944_remove ). This patch properly handles lp3944_configure fail in lp3944_probe.

Hi Axel, thanks for the fix, I agree it's needed.
There are some minor comments inlined below.

Plus, when possible, I prefer commit messages to be wrapped to 72/80 characters per line. Please, consider this if you are sending a v2.

Ah, may I ask where you are using this driver? Just curious.

Quoted 18 lines
---
 drivers/leds/leds-lp3944.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)
drivers/leds/leds-lp3944.c +1 -0
index 8d5ecce..03a24d0 100644
@@ -379,6 +379,7 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 {
 	struct lp3944_platform_data *lp3944_pdata = client->dev.platform_data;
 	struct lp3944_data *data;
+	int err;
 
 	if (lp3944_pdata == NULL) {
 		dev_err(&client->dev, "no platform data\n");

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 410

Quoted 8 lines
drivers/leds/leds-lp3944.c +4 -1
@@ -403,8 +404,15 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 
 	dev_info(&client->dev, "lp3944 enabled\n");
 
-	lp3944_configure(client, data, lp3944_pdata);
+	err = lp3944_configure(client, data, lp3944_pdata);
+	if (err < 0)
+		goto err_configure;
+

The dev_info(&client->dev, "lp3944 enabled\n"); could now go right here, before the return 0, so we don't report the driver as enabled even in the case its probe fails.

Quoted 6 lines

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 414

Quoted 4 lines
drivers/leds/leds-lp3944.c +3 -0
 	return 0;
+
+err_configure:
+	kfree(data);

add i2c_set_clientdata(client, NULL) here just like in lp3944_remove()

Just to note:
This is not needed anymore, the i2c-core does this meanwhile for the drivers. I am preparing a cleanup patch fixing this in all drivers for -rc1.

Signature · 2 lines
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
author
#3 lore raw
In reply to Antonio Ospite

2010/5/17 Antonio Ospite <[email protected]>:

Quoted 8 lines
Quoted 4 lines · On Mon, 17 May 2010 09:34:02 +0800 Axel Lin <[email protected]> wrote:

In current implementation, lp3944_probe return 0 even if lp3944_configure fail.
Therefore, led_classdev_unregister will be executed twice ( in error handling of lp3944_configure and lp3944_remove ). This patch properly handles lp3944_configure fail in lp3944_probe.

Quoted 5 lines

Hi Axel, thanks for the fix, I agree it's needed.
There are some minor comments inlined below.

Plus, when possible, I prefer commit messages to be wrapped to 72/80 characters per line. Please, consider this if you are sending a v2.

ok. done in v2.

Quoted 24 lines

Ah, may I ask where you are using this driver? Just curious.

Quoted 21 lines

---
 drivers/leds/leds-lp3944.c |   10 +++++++++-

 1 files changed, 9 insertions(+), 1 deletions(-)
drivers/leds/leds-lp3944.c +1 -0
index 8d5ecce..03a24d0 100644
@@ -379,6 +379,7 @@ static int __devinit lp3944_probe(struct i2c_client *client,
  {
       struct lp3944_platform_data *lp3944_pdata = client->dev.platform_data;
       struct lp3944_data *data;
+     int err;
 
       if (lp3944_pdata == NULL) {
               dev_err(&client->dev, "no platform data\n");
@@ -403,8 +404,15 @@ static int __devinit lp3944_probe(struct i2c_client *client,
 
       dev_info(&client->dev, "lp3944 enabled\n");
 
Quoted 9 lines

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 410

Quoted 5 lines
drivers/leds/leds-lp3944.c +4 -1
-     lp3944_configure(client, data, lp3944_pdata);
+     err = lp3944_configure(client, data, lp3944_pdata);
+     if (err < 0)
+             goto err_configure;
+

The dev_info(&client->dev, "lp3944 enabled\n"); could now go right here, before the return 0, so we don't report the driver as enabled even in the case its probe fails.

done in v2.

Quoted 7 lines

drivers/leds/leds-lp3944.c static int __devinit lp3944_probe(struct i2c_client *client, · line 414

Quoted 5 lines
drivers/leds/leds-lp3944.c +3 -0
       return 0;
+
+err_configure:
+     kfree(data);

add i2c_set_clientdata(client, NULL) here just like in lp3944_remove()

per Wolfram Sang's comment, I skip this part in v2.

Quoted 24 lines
Quoted 8 lines
drivers/leds/leds-lp3944.c +1 -1
+     return err;
  }
 
  static int __devexit lp3944_remove(struct i2c_client *client)
--

1.5.4.3

Regards,
  Antonio

--
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text. See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?