Improve the default parallel GC settings, and sanitise the flags (#3340)

[ghc-hetmet.git] / rts / sm / GC.c

diff --git a/rts/sm/GC.c b/rts/sm/GC.c
index 02fd6d9..6f15a47 100644 (file)
--- a/rts/sm/GC.c
+++ b/rts/sm/GC.c
@@ -33,6 +33,7 @@
 #endif
 #include "Trace.h"
 #include "RetainerProfile.h"
+#include "LdvProfile.h"
 #include "RaiseAsync.h"
 #include "Papi.h"
 #include "Stable.h"
@@ -236,7 +237,8 @@ GarbageCollect (rtsBool force_major_gc,
   n = initialise_N(force_major_gc);
 
 #if defined(THREADED_RTS)
-  work_stealing = RtsFlags.ParFlags.parGcLoadBalancing;
+  work_stealing = RtsFlags.ParFlags.parGcLoadBalancingEnabled &&
+                  N >= RtsFlags.ParFlags.parGcLoadBalancingGen;
       // It's not always a good idea to do load balancing in parallel
       // GC.  In particular, for a parallel program we don't want to
       // lose locality by moving cached data into another CPU's cache
@@ -1065,6 +1067,11 @@ loop:
 void
 gcWorkerThread (Capability *cap)
 {
+    gc_thread *saved_gct;
+
+    // necessary if we stole a callee-saves register for gct:
+    saved_gct = gct;
+
     cap->in_gc = rtsTrue;
 
     gct = gc_threads[cap->no];
@@ -1104,6 +1111,8 @@ gcWorkerThread (Capability *cap)
                gct->thread_index);
     ACQUIRE_SPIN_LOCK(&gct->mut_spin);
     debugTrace(DEBUG_gc, "GC thread %d on my way...", gct->thread_index);
+
+    SET_GCT(saved_gct);
 }
 
 #endif