On Sat, 2020-11-07 at 11:05 -0800, Jakub Kicinski wrote: > On Sat, 7 Nov 2020 17:21:31 +0000 Taehee Yoo wrote: > > When debugfs file is opened, its module should not be removed until > > it's closed. > > Because debugfs internally uses the module's data. > > So, it could access freed memory. > > > > In order to avoid panic, it just sets .owner to THIS_MODULE. > > So that all modules will be held when its debugfs file is opened. > > Hm, looks like some of the patches need to be revised because > .owner is already set in the ops, and a warning gets generated. > > Also it'd be good to mention why Johannes's approach was abandoned. Well, I had two. One was awful, and worked in all cases. The other was less awful, and didn't work in all cases. I think both gave Al Viro hives ;-) > Patch 1 needs to be split in two. Patches 2 and 3 would go via Johannes. FWIW, I'm happy for you to take patches 2 and 3 as well, but I guess if patch 1 needs to be split there's a resend coming anyway, so then I'll be happy to take the patches 2/3 from a separate set. johannes
