Tejun Heo <htejun@gmail.com> writes:

>> +{
>> +	struct sysfs_rename_struct *srs;
>> +	struct super_block *sb;
>> +	struct dentry *dentry;
>> +	int error;
>> +
>> +	list_for_each_entry(sb, &sysfs_fs_type.fs_supers, s_instances) {
>> +		dentry = sysfs_get_dentry(sb, sd);
>> +		if (!dentry)
>> +			continue;
>
> sysfs_get_dentry() return ERR_PTR() value.  Oops, sysfs_get_dentry()
> implementation is wrong too.  Also, please move
> sysfs_grab/release_supers() near this patch and add (a lot of)
> comments there.
>
> Other than that, I think this is as clean as this can be.  Great.

Welcome.  I will see what I can do with respect to cleaning up
the names.

As for the return value of sysfs_get_dentry that is tricky.  In particular
I have three specific cases the code needs to deal with.

- We got the dentry.
- We did not get the dentry because for this super block there never
  ever will be a dentry.
- Some kind of error occurred in attempting to get the dentry.

Not getting a dentry because it is impossible I am currently handling
with a NULL return.  I can equally use a specific error code to mean
that as well.  It doesn't much matter.  So I guess the hunk in
question could read:

>> +	list_for_each_entry(sb, &sysfs_fs_type.fs_supers, s_instances) {
>> +		dentry = sysfs_get_dentry(sb, sd);
>> +		if (dentry == ERR_PTR(-ENOENT))
>> +			continue;

As long as we handle that class of error differently I really don't
care.

Eric
_______________________________________________
Containers mailing list
Containers@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers