On Mon, May 25, 2026 at 02:32:35PM +0800, Rui Qi wrote: > From: "Rui Qi" <[email protected]> > > Fix a bug where rcu_read_unlock() was used instead of srcu_read_unlock() > in handle_read_event_rsp() when ipmi_alloc_recv_msg() fails. > > This mismatch leads to an SRCU read-side critical section imbalance: the > entry uses srcu_read_lock(&intf->users_srcu) but the error path > incorrectly calls rcu_read_unlock(), which is a no-op for SRCU and > leaves the SRCU lock held. > > The offending code was restructured in mainline by commit 3be997d5a64a > ("ipmi:msghandler: Remove srcu from the ipmi user structure"), which > replaced the SRCU locking with a mutex in this function, effectively > eliminating the mismatch. However, that commit is part of a larger > SRCU removal series that is not suitable for stable backport. This > minimal fix addresses the SRCU imbalance for 6.12 and earlier stable > branches that still carry the original locking scheme. > > Fixes: e86ee2d44b44 ("ipmi: Rework locking and shutdown for hot remove") > Cc: [email protected] > Signed-off-by: Rui Qi <[email protected]> > --- > drivers/char/ipmi/ipmi_msghandler.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/char/ipmi/ipmi_msghandler.c > b/drivers/char/ipmi/ipmi_msghandler.c > index 188722ec0337..41ae4dac4eeb 100644 > --- a/drivers/char/ipmi/ipmi_msghandler.c > +++ b/drivers/char/ipmi/ipmi_msghandler.c > @@ -4395,7 +4395,7 @@ static int handle_read_event_rsp(struct ipmi_smi *intf, > > recv_msg = ipmi_alloc_recv_msg(user); > if (IS_ERR(recv_msg)) { > - rcu_read_unlock(); > + srcu_read_unlock(&intf->users_srcu, index); > list_for_each_entry_safe(recv_msg, recv_msg2, &msgs, > link) { > list_del(&recv_msg->link); > -- > 2.20.1 >
Does not apply to the tree, please rebase and resend. thanks, greg k-h _______________________________________________ Openipmi-developer mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openipmi-developer
