Fix another setting of a flag out of lock protection

In this case is dangerous because in split() we reset the flag to
false, but if it was set due to a cut-off higher in the tree we
completely miss that and go on with the full search.

No functional change.

Signed-off-by: Marco Costalba <mcostalba@gmail.com>

diff --git a/src/search.cpp b/src/search.cpp
index f0bad5edc771c135b61d96357baecd18aa272ef7..dece7fd4958217523da01dbdae0dc0e5196fc971 100644 (file)
--- a/src/search.cpp
+++ b/src/search.cpp
@@ -2904,10 +2904,16 @@ namespace {
     for (int i = 0; i < ActiveThreads; i++)
         splitPoint->slaves[i] = 0;
 
-    threads[master].idle = false;
-    threads[master].stopRequest = false;
     threads[master].splitPoint = splitPoint;
 
+    // If we are here it means we are not idle
+    assert(!threads[master].idle);
+
+    // Following assert could fail because we could be slave of a master
+    // thread that has just raised a stop request. Note that stopRequest
+    // can be changed with only splitPoint::lock held, not with MPLock.
+    /* assert(!threads[master].stopRequest); */
+
     // Allocate available threads setting idle flag to false
     for (int i = 0; i < ActiveThreads && splitPoint->cpus < MaxThreadsPerSplitPoint; i++)
         if (thread_is_available(i, master))