Skip to contentSkip to conversation

hwmon/abituguru: handle sysfs errors

PATCH In review Jeff Garzik 4 participants 9 messages Oct 10, 2006 – Dec 2, 2006

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 hwmon/abituguru: handle sysfs errors

---

 drivers/hwmon/abituguru.c |   30 +++++++++++++++++++++++++-----
 1 file changed, 25 insertions(+), 5 deletions(-)

2b10f648c8ed965369976eb7925b922ee187ce21

drivers/hwmon/abituguru.c +25 -5
index e5cb0fd..3ded982 100644
@@ -1271,14 +1271,34 @@ static int __devinit abituguru_probe(str
 		res = PTR_ERR(data->class_dev);
 		goto abituguru_probe_error;
 	}
-	for (i = 0; i < sysfs_attr_i; i++)
-		device_create_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
-	for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++)
-		device_create_file(&pdev->dev,
-			&abituguru_sysfs_attr[i].dev_attr);
+	for (i = 0; i < sysfs_attr_i; i++) {
+		res = device_create_file(&pdev->dev,
+					 &data->sysfs_attr[i].dev_attr);
+		if (res) {
+			for (j = 0; j < i; j++)
+				device_remove_file(&pdev->dev,
+					 	&data->sysfs_attr[j].dev_attr);
+			goto err_devreg;
+		}
+	}
+	for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++) {
+		res = device_create_file(&pdev->dev,
+					 &abituguru_sysfs_attr[i].dev_attr);
+		if (res) {
+			for (j = 0; j < i; j++)
+				device_remove_file(&pdev->dev,
+					 &abituguru_sysfs_attr[j].dev_attr);
+			goto err_attr_i;
+		}
+	}
 
 	return 0;
 
+err_attr_i:
+	for (i = 0; i < sysfs_attr_i; i++)
+		device_remove_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
+err_devreg:
+	hwmon_device_unregister(data->class_dev);
 abituguru_probe_error:
 	kfree(data);
 	return res;
#1 lore raw

Hi Jean, Jeff,

You (Jean) already mailed me about this and it was on my todo list, but I'm currently rather busy with work. So it looks like Jeff beat me to it.

Jeff's patch looks fine, please apply. Thanks Jeff!

Regards,

Hans

---

drivers/hwmon/abituguru.c | 30 +++++++++++++++++++++++++-----

1 file changed, 25 insertions(+), 5 deletions(-)

2b10f648c8ed965369976eb7925b922ee187ce21

drivers/hwmon/abituguru.c +0 -0
index e5cb0fd..3ded982 100644
@@ -1271,14 +1271,34 @@ static int __devinit abituguru_probe(str
		res = PTR_ERR(data->class_dev);
		goto abituguru_probe_error;
	}

- for (i = 0; i < sysfs_attr_i; i++)
- device_create_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
- for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++)
- device_create_file(&pdev->dev,
- &abituguru_sysfs_attr[i].dev_attr);
+ for (i = 0; i < sysfs_attr_i; i++) {
+ res = device_create_file(&pdev->dev,
+ &data->sysfs_attr[i].dev_attr);
+ if (res) {
+ for (j = 0; j < i; j++)
+ device_remove_file(&pdev->dev,
+ &data->sysfs_attr[j].dev_attr);
+ goto err_devreg;
+ }
+ }
+ for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++) {
+ res = device_create_file(&pdev->dev,
+ &abituguru_sysfs_attr[i].dev_attr);
+ if (res) {
+ for (j = 0; j < i; j++)
+ device_remove_file(&pdev->dev,
+ &abituguru_sysfs_attr[j].dev_attr);
+ goto err_attr_i;
+ }
+ }

	return 0;

+err_attr_i:
+ for (i = 0; i < sysfs_attr_i; i++)
+ device_remove_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
+err_devreg:
+ hwmon_device_unregister(data->class_dev);
abituguru_probe_error:

	kfree(data);
	return res;
#2 lore raw

Hi Hans, Jeff,

Quoted 5 lines

You (Jean) already mailed me about this and it was on my todo list, but I'm currently rather busy with work. So it looks like Jeff beat me to it.

Jeff's patch looks fine, please apply. Thanks Jeff!

The patch isn't wrong per se, but it could be made more simple, and is incomplete in comparison to what was done for all other hardware monitoring drivers:

* We want to create all the files before registering with the hwmon class, this closes a race condition.
* We want to delete all the device files at regular cleanup time (after unregistering with the hwmon class).
* It's OK to call device_create_file() on a non-existent file, so the error path can be simplified.

I'd like the abituguru driver to behave the same as all other hardware monitoring drivers to lower the maintenance effort. Can either you or Jeff work up a compliant patch?

Thanks.

Quoted 48 lines
 drivers/hwmon/abituguru.c |   30 +++++++++++++++++++++++++-----
 1 file changed, 25 insertions(+), 5 deletions(-)

2b10f648c8ed965369976eb7925b922ee187ce21

drivers/hwmon/abituguru.c +25 -5
index e5cb0fd..3ded982 100644
@@ -1271,14 +1271,34 @@ static int __devinit abituguru_probe(str
 		res = PTR_ERR(data->class_dev);
 		goto abituguru_probe_error;
 	}
-	for (i = 0; i < sysfs_attr_i; i++)
-		device_create_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
-	for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++)
-		device_create_file(&pdev->dev,
-			&abituguru_sysfs_attr[i].dev_attr);
+	for (i = 0; i < sysfs_attr_i; i++) {
+		res = device_create_file(&pdev->dev,
+					 &data->sysfs_attr[i].dev_attr);
+		if (res) {
+			for (j = 0; j < i; j++)
+				device_remove_file(&pdev->dev,
+					 	&data->sysfs_attr[j].dev_attr);
+			goto err_devreg;
+		}
+	}
+	for (i = 0; i < ARRAY_SIZE(abituguru_sysfs_attr); i++) {
+		res = device_create_file(&pdev->dev,
+					 &abituguru_sysfs_attr[i].dev_attr);
+		if (res) {
+			for (j = 0; j < i; j++)
+				device_remove_file(&pdev->dev,
+					 &abituguru_sysfs_attr[j].dev_attr);
+			goto err_attr_i;
+		}
+	}
 
 	return 0;
 
+err_attr_i:
+	for (i = 0; i < sysfs_attr_i; i++)
+		device_remove_file(&pdev->dev, &data->sysfs_attr[i].dev_attr);
+err_devreg:
+	hwmon_device_unregister(data->class_dev);
 abituguru_probe_error:
 	kfree(data);
 	return res;
Signature · 1 line
Jean Delvare
#3 lore raw
Quoted 6 lines · Jean Delvare

Hi Hans, Jeff,

Quoted 4 lines

You (Jean) already mailed me about this and it was on my todo list, but I'm currently rather busy with work. So it looks like Jeff beat me to it.

Quoted 8 lines
Quoted 1 line

Jeff's patch looks fine, please apply. Thanks Jeff!

The patch isn't wrong per se, but it could be made more simple, and is incomplete in comparison to what was done for all other hardware monitoring drivers:

* We want to create all the files before registering with the hwmon class, this closes a race condition.

OK

Quoted 2 lines

* We want to delete all the device files at regular cleanup time (after unregistering with the hwmon class).

Is this really nescesarry? AFAIK the files get deleted when the device gets deleted.

Quoted 3 lines

* It's OK to call device_create_file() on a non-existent file, so the error path can be simplified.

?? You mean device_remove_file I assume?

Quoted 4 lines

I'd like the abituguru driver to behave the same as all other hardware monitoring drivers to lower the maintenance effort. Can either you or Jeff work up a compliant patch?

I understand Jeff any chance you can do a new revision of your patch? Otherwise I'll take care of it as time permits.

Regards,

Hans

#4 lore raw

Hans,

Quoted 5 lines
Quoted 2 lines

* We want to delete all the device files at regular cleanup time (after unregistering with the hwmon class).

Is this really nescesarry? AFAIK the files get deleted when the device gets deleted.

The point is that the device shouldn't be deleted if we were following the device driver model. After all the uGuru device exists in your system before the abituguru driver is loaded, and is still there after the driver is unloaded.

Now I agree that for now it won't make a difference, be we are preparing for the future, and all other hardware monitoring drivers were updated that way already.

Quoted 4 lines
Quoted 2 lines

* It's OK to call device_create_file() on a non-existent file, so the error path can be simplified.

?? You mean device_remove_file I assume?

Oops, yes, that was a typo, sorry.

Thanks,

Signature · 1 line
Jean Delvare
#5 lore raw
Quoted 17 lines · Jean Delvare

Hans,

Quoted 5 lines
Quoted 2 lines

* We want to delete all the device files at regular cleanup time (after unregistering with the hwmon class).

Is this really nescesarry? AFAIK the files get deleted when the device gets deleted.

The point is that the device shouldn't be deleted if we were following the device driver model. After all the uGuru device exists in your system before the abituguru driver is loaded, and is still there after the driver is unloaded.

Now I agree that for now it won't make a difference, be we are preparing for the future, and all other hardware monitoring drivers were updated that way already.

Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

* It's OK to call device_create_file() on a non-existent file, so the error path can be simplified.

?? You mean device_remove_file I assume?

Oops, yes, that was a typo, sorry.

I know I sound like a roken record but this driver would benefit from using attribute_group.

Signature · 1 line
Dmitry
#6 lore raw

Hi Dmitry,

Quoted 2 lines · Dmitry Torokhov

I know I sound like a roken record but this driver would benefit from using attribute_group.

What about sending a patch then?

Signature · 1 line
Jean Delvare
#7 lore raw
Quoted 8 lines · Jean Delvare

Hi Dmitry,

Quoted 2 lines · Dmitry Torokhov

I know I sound like a roken record but this driver would benefit from using attribute_group.

What about sending a patch then?

Sorry, spoke too soon. Because the driver uses non-satic attributes converting it to use attribute_group would probably be messier than the present code.

Signature · 1 line
Dmitry
#8 lore raw
In reply to Hans de Goede

Hans,

Quoted 12 lines · Hans de Goede
Quoted 11 lines · Jean Delvare

The patch isn't wrong per se, but it could be made more simple, and is incomplete in comparison to what was done for all other hardware monitoring drivers:

* We want to create all the files before registering with the hwmon class, this closes a race condition.
* We want to delete all the device files at regular cleanup time (after unregistering with the hwmon class).
* It's OK to call device_remove_file() on a non-existent file, so the error path can be simplified.

Quoted 6 lines
Quoted 3 lines

I'd like the abituguru driver to behave the same as all other hardware monitoring drivers to lower the maintenance effort. Can either you or Jeff work up a compliant patch?

I understand Jeff any chance you can do a new revision of your patch? Otherwise I'll take care of it as time permits.

I'm still waiting for this patch. It'd be nice to have the abituguru driver fixed in 2.6.20.

Thanks,

Signature · 1 line
Jean Delvare