Bugfix: synchronize with all other CPUs before switching plugin

The CPU triggering the plugin switch should wait until all other CPUs
are in a proper state (synch_on_plugin_switch()) before performing the
actual switch.

Based on the original patch from Jeremy Erickson <jerickso@cs.unc.edu>.

This should solve (for most practical cases) the C-EDF-related
plugin-switch problem reported on the ML.

1 files changed, 6 insertions, 1 deletions

diff --git a/litmus/litmus.c b/litmus/litmus.c
index 0756d0156f8f..8efd3f9ef7ee 100644
--- a/litmus/litmus.c
+++ b/litmus/litmus.c
@@ -382,7 +382,8 @@ void litmus_exit_task(struct task_struct* tsk)
 /* IPI callback to synchronize plugin switching */
 static void synch_on_plugin_switch(void* info)
 {
-        while (atomic_read(&cannot_use_plugin))
+        atomic_inc(&cannot_use_plugin);
+        while (atomic_read(&cannot_use_plugin) > 0)
                 cpu_relax();
 }
 
@@ -403,6 +404,10 @@ int switch_sched_plugin(struct sched_plugin* plugin)
         /* send IPI to force other CPUs to synch with us */
         smp_call_function(synch_on_plugin_switch, NULL, 0);
 
+        /* wait until all other CPUs have started synch */
+        while (atomic_read(&cannot_use_plugin) < num_online_cpus())
+                cpu_relax();
+
         /* stop task transitions */
         raw_spin_lock_irqsave(&task_transition_lock, flags);