Kay Sievers [off-list ref] writes:

Kay you are full of it.  My argument is that this a clear case
where aiming for perfection is the enemy of the good.

quoted

--- a/drivers/base/core.c
+++ b/drivers/base/core.c

@@ -673,7 +673,7 @@ static struct kobject *get_device_parent(struct device *dev,

                */
               if (parent == NULL)
                       parent_kobj = virtual_device_parent(dev);
-               else if (parent->class)
+               else if (parent->class && (strcmp(dev->class->name, "net") != 0))

Since the issue is not a regression, and not even a bug in the core,
it should not be done this way for mainline.

Yes the issue is a namespace regression and it has been a regression
for the bluetooth network drivers since CONFIG_SYSFS_DEPRECATED was
introduced.  "ip link set name blarg"  Where blarg is an attribute of
the bluteooth parent device fails only with CONFIG_SYSFS_DEPRECATED
disabled.

I reported that regression for the majority of the network devices and
the class directories were introduced to avoid this problem, except
convoluted logic of when to create a class directory fails to cover
a few flavors of network drivers.

For the mac80211_hwsim driver I don't care.  That isn't real hardware
and it is only interesting for testing.  For the bluetooth network driver
that is real hardware that needs to be made to work.

Please try to fix these drivers instead, or mark the broken for
namespaces, if nobody can fix them right now.

Let me get this straight.  You believe a non-trival non-obvious
complete rewrite of entire subsystems after rc1 is better than
a one line bug fix that let's real hardware work for real
people?

You have had 3 years to do better a better job with the bluetooth
driver.

Furthermore we fixed the rest of this regression in the core I don't
see why we can fix the entirety of it this namespace regression in the
core.  Especially since it is an obviously correct one line change.

Eric