: Do not free already unregistered net device.
Subsystems (2)
I do not understand why this change was accepted. All of the network drivers I just looked at in 2.6.25 do unregister_netdev() followed sometime soon by free_netdev(). Is there something different about ATM devices? I did not look at all the drivers.
Don
David Miller
maintainerFrom: Don Fry <[email protected]>
Date: Mon, 05 May 2008 09:44:40 -0700
Quoted 4 lines
I do not understand why this change was accepted. All of the network drivers I just looked at in 2.6.25 do
unregister_netdev()followed sometime soon byfree_netdev(). Is there something different about ATM devices? I did not look at all the drivers.
Indeed, this change does look bogus.
Pavel, please take a look, unregister_netdev() indeed does not free the netdevice, and something (usually the caller) needs to do it.
I think I should revert your change.
Quoted 8 lines · David Miller
From: Don Fry <[email protected]>
Date: Mon, 05 May 2008 09:44:40 -0700Quoted 4 lines
I do not understand why this change was accepted. All of the network drivers I just looked at in 2.6.25 do
unregister_netdev()followed sometime soon byfree_netdev(). Is there something different about ATM devices? I did not look at all the drivers.
Quoted 4 lines
Indeed, this change does look bogus.
Pavel, please take a look,
unregister_netdev()indeed does not free the netdevice, and something (usually the caller) needs to do it.
Indeed :( My bad, I've overlooked the fact, that this device doesn't have a ->destructor callback set (which most of the others set to the free_netdev), so the free_netdev is to be called manually.
Quoted 1 line
I think I should revert your change.
Agree. Sorry, for the confusion.
David Miller
maintainerFrom: Pavel Emelyanov <[email protected]>
Date: Tue, 06 May 2008 10:34:38 +0400
Quoted 4 lines
Quoted 1 line · David Miller
I think I should revert your change.
Agree. Sorry, for the confusion.
No worries, thanks for confirming. I'll revert the change right now.