Skip to contentSkip to conversation

hwmon: fix unchecked returncodes in w83791d

PATCH New Jean Delvare 2 participants 3 messages Sep 25, 2006

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 hwmon: fix unchecked returncodes in w83791d

Quoted 10 lines

Fix unchecked return-status by replacing all unchecked calls to device_create_file with a single group declaration,
and one call to sysfs_create_group, and check that one return status.

Signed-off-by Jim Cromie <jim.cromie at gmail.com>

---

$ diffstat pc-set/hwmon-unchecked-return-status-fixes-w83791d.patch
 w83791d.c |   83 +++++++++++++++++++++++++++++++++++++++++---------------------
 1 files changed, 55 insertions(+), 28 deletions(-)
Quoted 6 lines

---

Its impressive (or embarrassing) how much slop you detect. I trust this comment is in right place for auto-strip to keep it out of Changelog. Is any of this automated, and if so, are we/I getting it right (or close) ?

Yes, anything between "---" and the beginning of the actual patch is dropped upstream. But I'm still editing things manually anyway so it's no big deal where you add your comments.

Review static int w83791d_detach_client(struct · line 1060

Quoted 8 lines
(quoted diff) +2 -0
@@ -1029,6 +1054,8 @@ static int w83791d_detach_client(struct 
 	if (data)
 		hwmon_device_unregister(data->class_dev);
 
+	sysfs_remove_group(&client->dev.kobj, &w83791d_group);
+
 	if ((err = i2c_detach_client(client)))
 		return err;

The "if (data)" is used to differenciate between the real client and the subclients, exactly as is done in the w83781d driver. As subclients have no files, the call to sysfs_remove_group() should be made conditional as well. If not, it'll still work, but with a significant performance drop.

Thanks,

Signature · 1 line
Jean Delvare
#1 lore raw

PATCH 1/1 hwmon: fix unchecked returncodes in w83791d

Quoted 27 lines · Jean Delvare
Quoted 10 lines

Fix unchecked return-status by replacing all unchecked calls to device_create_file with a single group declaration,
and one call to sysfs_create_group, and check that one return status.

Signed-off-by Jim Cromie <jim.cromie at gmail.com>

---

$ diffstat pc-set/hwmon-unchecked-return-status-fixes-w83791d.patch
 w83791d.c |   83 +++++++++++++++++++++++++++++++++++++++++---------------------
 1 files changed, 55 insertions(+), 28 deletions(-)
Quoted 7 lines

---

Its impressive (or embarrassing) how much slop you detect. I trust this comment is in right place for auto-strip to keep it out of Changelog. Is any of this automated, and if so, are we/I getting it right (or close) ?

Yes, anything between "---" and the beginning of the actual patch is dropped upstream. But I'm still editing things manually anyway so it's no big deal where you add your comments.

Quoted 4 lines
(quoted diff) +0 -0
@@ -1029,6 +1054,8 @@ static int w83791d_detach_client(struct 
 	if (data)
 		hwmon_device_unregister(data->class_dev);
 
Quoted 13 lines

Review static int w83791d_detach_client(struct · line 1061

Quoted 5 lines
(quoted diff) +2 -0
+	sysfs_remove_group(&client->dev.kobj, &w83791d_group);
+
 	if ((err = i2c_detach_client(client)))
 		return err;
 

The "if (data)" is used to differenciate between the real client and the subclients, exactly as is done in the w83781d driver. As subclients have no files, the call to sysfs_remove_group() should be made conditional as well. If not, it'll still work, but with a significant performance drop.

on this item, I actually checked :-)

w83791d_detect will fail unless data is sucessfully allocated,
so it must be there if detach_client is called.

but I suppose its safer (less action-at-a-distance) your way, esp wrt any future changes.
redo attached.

Quoted 2 lines

Thanks,

-------------- next part --------------
A non-text attachment was scrubbed...
Name: hwmon-unchecked-return-status-fixes-w83791d.patch
Type: text/x-patch
Size: 3343 bytes
Desc: not available
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20060925/773e9a77/attachment.bin

#2 lore raw

PATCH 1/1 hwmon: fix unchecked returncodes in w83791d

In reply to Jean Delvare

Hi Jim,

Quoted 15 lines
Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -1029,6 +1054,8 @@ static int w83791d_detach_client(struct 
 	if (data)
 		hwmon_device_unregister(data->class_dev);
 
Quoted 10 lines

Review static int w83791d_detach_client(struct · line 1060

Quoted 4 lines
(quoted diff) +2 -0
+	sysfs_remove_group(&client->dev.kobj, &w83791d_group);
+
 	if ((err = i2c_detach_client(client)))
 		return err;

The "if (data)" is used to differenciate between the real client and the subclients, exactly as is done in the w83781d driver. As subclients have no files, the call to sysfs_remove_group() should be made conditional as well. If not, it'll still work, but with a significant performance drop.

Quoted 8 lines

on this item, I actually checked :-)

w83791d_detect will fail unless data is sucessfully allocated,
so it must be there if detach_client is called.

but I suppose its safer (less action-at-a-distance) your way, esp wrt any future changes.
redo attached.

I'm sorry to insist, but you overlooked another function which also attaches clients: w83791d_create_subclient(). And this one does _not_ associate data to the client. All three clients (one main and two subclients) have the same driver, so w83791d_detach_client() is called for all of them. The "if (data)" test is there exactly for this reason, so this test is really needed.

The new patch looks OK to me, thanks.

Signature · 1 line
Jean Delvare