sched/numa: Fix use of spin_{un}lock_irq() when interrupts are disabled
Steven Rostedt
reviewerPATCH v2 1/1 sched/numa: Fix use of spin_{un}lock_irq() when interrupts are disabled
As Peter Zijlstra told me, we have the following path:
do_exit()
exit_itimers()
itimer_delete()
spin_lock_irqsave(&timer->it_lock, &flags);
timer_delete_hook(timer);
kc->timer_del(timer) := posix_cpu_timer_del()
put_task_struct()
__put_task_struct()
task_numa_free()
spin_lock(&grp->lock);Which means that task_numa_free() can be called with interrupts disabled, which means that we should not be using spin_lock_irq() but spin_lock_irqsave() instead. Otherwise we are enabling interrupts while holding an interrupt unsafe lock!
---
v2: let's see if claws mail screws up the format on this on :-p
kernel/sched/fair.c +4 -3
@@ -1707,18 +1707,19 @@ no_join:void task_numa_free (struct task_struct *p) {struct numa_group *grp = p->numa_group; -int i;void *numa_faults = p->numa_faults_memory; +unsigned long flags; +int i;if (grp) { -spin_lock_irq (&grp->lock); +spin_lock_irqsave (&grp->lock, flags);for (i =0 ; i <NR_NUMA_HINT_FAULT_STATS * nr_node_ids; i++) grp->faults[i] -= p->numa_faults_memory[i]; grp->total_faults -= p->total_numa_faults;list_del (&p->numa_entry); grp->nr_tasks--; -spin_unlock_irq (&grp->lock); +spin_unlock_irqrestore (&grp->lock, flags);rcu_assign_pointer (p->numa_group,NULL );put_numa_group (grp); }
Peter Zijlstra
maintainerQuoted 14 lines · Steven Rostedt
As Peter Zijlstra told me, we have the following path:
do_exit()
exit_itimers()itimer_delete() spin_lock_irqsave(&timer->it_lock, &flags); timer_delete_hook(timer); kc->timer_del(timer) := posix_cpu_timer_del() put_task_struct() __put_task_struct() task_numa_free() spin_lock(&grp->lock);
Quoted 5 lines
Which means that
task_numa_free()can be called with interrupts disabled, which means that we should not be usingspin_lock_irq()butspin_lock_irqsave()instead. Otherwise we are enabling interrupts while holding an interrupt unsafe lock!
Duh, yes. Thanks Steve!