I would say maybe. Because it is just suppressing some warnings during umount for "sleep during spinlock". As far as I can tell, in this particular situation, the warnings should be harmless as we won't be entering this same function concurrently from any other thread during umount. The warning that is printed is valid, we should not potentially sleep while holding a spinlock because that can cause a different thread to spin for very long trying to grab that same spinlock. But in this case I don't think it can happen. So the patch is I think mostly cosmetic. Prevent a warning message from being printed to dmesg which in 99% of the cases are something that can cause real issues and in very few special cases, like this, is still a bug but benign. But lets push it to stable. Even if benign in this particular case, stable-users do not deserve to see these messages in the log. On Sat, Mar 19, 2022 at 2:10 PM Steve French <smfrench@xxxxxxxxx> wrote: > > Since "cifs: we do not need a spinlock around the tree access during > umount" fixes "cifs: fix handlecache and multiuser" (commit > 47178c7722ac528ea08aa82c3ef9ffa178962d7a) which was marked for stable. > I marked "cifs: we do not need a spinlock around the tree access > during umount" for stable as well. > > Let me know if you want me to change that. > > > -- > Thanks, > > Steve
