| Message ID | 1346352988-32444-3-git-send-email-paulmck@linux.vnet.ibm.com |
|---|---|
| State | Accepted |
| Commit | 1e3fd2b38cea41f5386bf23440f2cbdd74cf13d0 |
| Headers | show |
On Thu, Aug 30, 2012 at 11:56:16AM -0700, Paul E. McKenney wrote: > From: "Paul E. McKenney" <paul.mckenney@linaro.org> > > When rcu_preempt_offline_tasks() clears tasks from a leaf rcu_node > structure, it does not NULL out the structure's ->boost_tasks field. > This commit therefore fixes this issue. > > Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org> > --- > kernel/rcutree_plugin.h | 7 ++++--- > 1 files changed, 4 insertions(+), 3 deletions(-) > > diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h > index 7f3244c..b1b4851 100644 > --- a/kernel/rcutree_plugin.h > +++ b/kernel/rcutree_plugin.h > @@ -584,8 +584,11 @@ static int rcu_preempt_offline_tasks(struct rcu_state *rsp, > raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */ > } > > + rnp->gp_tasks = NULL; > + rnp->exp_tasks = NULL; > #ifdef CONFIG_RCU_BOOST > - /* In case root is being boosted and leaf is not. */ > + rnp->boost_tasks = NULL; > + /* In case root is being boosted and leaf was not. */ > raw_spin_lock(&rnp_root->lock); /* irqs already disabled */ > if (rnp_root->boost_tasks != NULL && > rnp_root->boost_tasks != rnp_root->gp_tasks) > @@ -593,8 +596,6 @@ static int rcu_preempt_offline_tasks(struct rcu_state *rsp, > raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */ > #endif /* #ifdef CONFIG_RCU_BOOST */ > > - rnp->gp_tasks = NULL; > - rnp->exp_tasks = NULL; > return retval; > } > > -- > 1.7.8 >
On Thu, 2012-08-30 at 11:56 -0700, Paul E. McKenney wrote: > When rcu_preempt_offline_tasks() clears tasks from a leaf rcu_node > structure, it does not NULL out the structure's ->boost_tasks field. > This commit therefore fixes this issue. What would have been the side-effects of this? Would rcu-boosting have been able to go funny on hotplug, and if so, how?
On Thu, Sep 06, 2012 at 04:40:38PM +0200, Peter Zijlstra wrote: > On Thu, 2012-08-30 at 11:56 -0700, Paul E. McKenney wrote: > > When rcu_preempt_offline_tasks() clears tasks from a leaf rcu_node > > structure, it does not NULL out the structure's ->boost_tasks field. > > This commit therefore fixes this issue. > > What would have been the side-effects of this? Would rcu-boosting have > been able to go funny on hotplug, and if so, how? In some circumstances, this could prevent any future RCU boosting. The ->boost_tasks field would be non-NULL, so it wouldn't ever try boosting again, having been fooled into thinking that the previous boost attempt was still in progress. The expected segfault is prevented by the fact that an attempt to initiate a boost first checks for ->boost_tasks being non-NULL, and if so, declines to wake up the RCU-boost kthread. Thanx, Paul
diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 7f3244c..b1b4851 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -584,8 +584,11 @@ static int rcu_preempt_offline_tasks(struct rcu_state *rsp, raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */ } + rnp->gp_tasks = NULL; + rnp->exp_tasks = NULL; #ifdef CONFIG_RCU_BOOST - /* In case root is being boosted and leaf is not. */ + rnp->boost_tasks = NULL; + /* In case root is being boosted and leaf was not. */ raw_spin_lock(&rnp_root->lock); /* irqs already disabled */ if (rnp_root->boost_tasks != NULL && rnp_root->boost_tasks != rnp_root->gp_tasks) @@ -593,8 +596,6 @@ static int rcu_preempt_offline_tasks(struct rcu_state *rsp, raw_spin_unlock(&rnp_root->lock); /* irqs still disabled */ #endif /* #ifdef CONFIG_RCU_BOOST */ - rnp->gp_tasks = NULL; - rnp->exp_tasks = NULL; return retval; }