Fix crash caused by concurrent ObjectSpace.dump_all calls

Since the callback defined in the objspace module might give up the GVL,
we need to make sure the right cr->mfd value is set back after the GVL
is re-obtained.

1 files changed, 5 insertions, 1 deletions

diff --git a/gc.c b/gc.c
index 510b83c817..2c3d3e1100 100644
--- a/gc.c
+++ b/gc.c
@@ -11935,7 +11935,11 @@ static void
 reachable_objects_from_callback(VALUE obj)
 {
     rb_ractor_t *cr = GET_RACTOR();
-    cr->mfd->mark_func(obj, cr->mfd->data);
+    struct gc_mark_func_data_struct *cur_mfd = cr->mfd;
+    cur_mfd->mark_func(obj, cr->mfd->data);
+    /* mark_func might give up the GVL, in which time some other thread might set
+       mfd. In that case, set it back to the right value for this thread. */
+    cr->mfd = cur_mfd;
 }
 
 void