diff --git a/src/gc.c b/src/gc.c
index 7089bb12fc347..7fac0f130d4f4 100644
--- a/src/gc.c
+++ b/src/gc.c
@@ -2969,7 +2969,7 @@ size_t gc_count_work_in_queue(jl_ptls_t ptls) JL_NOTSAFEPOINT
  * have tried to steal from the queue which still has a work item left, but failed to do so,
  * which violates the semantics of Chase-Lev's work-stealing queue.
  *
- * - Let E1 be the event "master thread writes -1 to gc_master_tid" and E2 be the even
+ * - Let E1 be the event "master thread writes -1 to gc_master_tid" and E2 be the event
  * "master thread observes that `gc_n_threads_marking` is zero". Since we're using
  * sequentially consistent atomics, E1 => E2. Now suppose one thread which is spinning in
  * `gc_should_mark` tries to enter the mark-loop after E2. In order to do so, it must
diff --git a/test/gc.jl b/test/gc.jl
index e085c1d8658e5..99b96a5ae1fd2 100644
--- a/test/gc.jl
+++ b/test/gc.jl
@@ -15,6 +15,13 @@ function run_gctest(file)
     end
 end
 
+function run_nonzero_page_utilization_test()
+    GC.gc()
+    page_utilization = Base.gc_page_utilization_data()
+    # at least one of the pools should have nonzero page_utilization
+    @test any(page_utilization .> 0)
+end
+
 # !!! note:
 #     Since we run our tests on 32bit OS as well we confine ourselves
 #     to parameters that allocate about 512MB of objects. Max RSS is lower
@@ -24,4 +31,5 @@ end
     run_gctest("gc/linkedlist.jl")
     run_gctest("gc/objarray.jl")
     run_gctest("gc/chunks.jl")
+    run_nonzero_page_utilization_test()
 end