On Sat, 2016-02-13 at 00:02 +0100, Sebastian Andrzej Siewior wrote:
> From: Thomas Gleixner <[email protected]>
> 
> We currently disable migration across lock acquisition. That includes the part
> where we block on the lock and schedule out. We cannot disable migration after
> taking the lock as that would cause a possible lock inversion.
> 
> But we can be smart and enable migration when we block and schedule out. That
> allows the scheduler to place the task freely at least if this is the first
> migrate disable level. For nested locking this does not help at all.

I met a problem while testing shiny new hotplug machinery.

rt/locking: Fix rt_spin_lock_slowlock() vs hotplug migrate_disable() bug

migrate_disable() -> pin_current_cpu() -> hotplug_lock() leads to..
        BUG_ON(rt_mutex_real_waiter(task->pi_blocked_on));
..so let's call migrate_disable() after we acquire the lock instead.

Fixes: e24b142cfb4a rt/locking: Reenable migration accross schedule
Signed-off-by: Mike Galbraith <[email protected]>
---
 kernel/locking/rtmutex.c |   15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

--- a/kernel/locking/rtmutex.c
+++ b/kernel/locking/rtmutex.c
@@ -1011,7 +1011,7 @@ static void  noinline __sched rt_spin_lo
        struct task_struct *lock_owner, *self = current;
        struct rt_mutex_waiter waiter, *top_waiter;
        unsigned long flags;
-       int ret;
+       bool mg_disable = false;
 
        rt_mutex_init_waiter(&waiter, true);
 
@@ -1035,8 +1035,7 @@ static void  noinline __sched rt_spin_lo
        __set_current_state_no_track(TASK_UNINTERRUPTIBLE);
        raw_spin_unlock(&self->pi_lock);
 
-       ret = task_blocks_on_rt_mutex(lock, &waiter, self, 
RT_MUTEX_MIN_CHAINWALK);
-       BUG_ON(ret);
+       BUG_ON(task_blocks_on_rt_mutex(lock, &waiter, self, 
RT_MUTEX_MIN_CHAINWALK));
 
        for (;;) {
                /* Try to acquire the lock again. */
@@ -1051,11 +1050,12 @@ static void  noinline __sched rt_spin_lo
                debug_rt_mutex_print_deadlock(&waiter);
 
                if (top_waiter != &waiter || adaptive_wait(lock, lock_owner)) {
-                       if (mg_off)
+                       if (mg_off && self->migrate_disable == 1) {
+                               mg_off = false;
+                               mg_disable = true;
                                migrate_enable();
+                       }
                        schedule();
-                       if (mg_off)
-                               migrate_disable();
                }
 
                raw_spin_lock_irqsave(&lock->wait_lock, flags);
@@ -1088,6 +1088,9 @@ static void  noinline __sched rt_spin_lo
 
        raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
 
+       if (mg_disable)
+               migrate_disable();
+
        debug_rt_mutex_free_waiter(&waiter);
 }
 

Reply via email to