From: Mathieu Desnoyers Date: Wed, 2 Apr 2008 20:04:51 +0000 (-0700) Subject: markers: use synchronize_sched() X-Git-Tag: v2.6.25-rc9~82 X-Git-Url: http://pilppa.com/gitweb/?a=commitdiff_plain;h=6496968e6cc3f01faafa63a5a28549a708539ac0;p=linux-2.6-omap-h63xx.git markers: use synchronize_sched() Markers do not mix well with CONFIG_PREEMPT_RCU because it uses preempt_disable/enable() and not rcu_read_lock/unlock for minimal intrusiveness. We would need call_sched and sched_barrier primitives. Currently, the modification (connection and disconnection) of probes from markers requires changes to the data structure done in RCU-style : a new data structure is created, the pointer is changed atomically, a quiescent state is reached and then the old data structure is freed. The quiescent state is reached once all the currently running preempt_disable regions are done running. We use the call_rcu mechanism to execute kfree() after such quiescent state has been reached. However, the new CONFIG_PREEMPT_RCU version of call_rcu and rcu_barrier does not guarantee that all preempt_disable code regions have finished, hence the race. The "proper" way to do this is to use rcu_read_lock/unlock, but we don't want to use it to minimize intrusiveness on the traced system. (we do not want the marker code to call into much of the OS code, because it would quickly restrict what can and cannot be instrumented, such as the scheduler). The temporary fix, until we get call_rcu_sched and rcu_barrier_sched in mainline, is to use synchronize_sched before each call_rcu calls, so we wait for the quiescent state in the system call code path. It will slow down batch marker enable/disable, but will make sure the race is gone. Signed-off-by: Mathieu Desnoyers Acked-by: Paul E. McKenney Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- diff --git a/kernel/marker.c b/kernel/marker.c index 041c33e3e95..005b9595459 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -671,6 +671,9 @@ int marker_probe_register(const char *name, const char *format, entry->rcu_pending = 1; /* write rcu_pending before calling the RCU callback */ smp_wmb(); +#ifdef CONFIG_PREEMPT_RCU + synchronize_sched(); /* Until we have the call_rcu_sched() */ +#endif call_rcu(&entry->rcu, free_old_closure); end: mutex_unlock(&markers_mutex); @@ -714,6 +717,9 @@ int marker_probe_unregister(const char *name, entry->rcu_pending = 1; /* write rcu_pending before calling the RCU callback */ smp_wmb(); +#ifdef CONFIG_PREEMPT_RCU + synchronize_sched(); /* Until we have the call_rcu_sched() */ +#endif call_rcu(&entry->rcu, free_old_closure); remove_marker(name); /* Ignore busy error message */ ret = 0; @@ -792,6 +798,9 @@ int marker_probe_unregister_private_data(marker_probe_func *probe, entry->rcu_pending = 1; /* write rcu_pending before calling the RCU callback */ smp_wmb(); +#ifdef CONFIG_PREEMPT_RCU + synchronize_sched(); /* Until we have the call_rcu_sched() */ +#endif call_rcu(&entry->rcu, free_old_closure); remove_marker(entry->name); /* Ignore busy error message */ end: