diff --git a/data/local.mk b/data/local.mk
index 50efda39..21106451 100644
--- a/data/local.mk
+++ b/data/local.mk
@@ -34,6 +34,7 @@ dist_skeletons_DATA =                           \
   data/skeletons/lalr1.java                     \
   data/skeletons/location.cc                    \
   data/skeletons/stack.hh                       \
+  data/skeletons/traceon.m4                     \
   data/skeletons/variant.hh                     \
   data/skeletons/yacc.c
 
diff --git a/data/skeletons/traceon.m4 b/data/skeletons/traceon.m4
new file mode 100644
index 00000000..344d7d1c
--- /dev/null
+++ b/data/skeletons/traceon.m4
@@ -0,0 +1,2 @@
+dnl GNU M4 treats -dV in a position-independent manner.
+m4_debugmode(V)m4_traceon()dnl
diff --git a/src/getargs.c b/src/getargs.c
index 6f19a4ea..55884d34 100644
--- a/src/getargs.c
+++ b/src/getargs.c
@@ -257,7 +257,8 @@ static const argmatch_trace_doc argmatch_trace_docs[] =
   { "sets",       "grammar sets: firsts, nullable etc." },
   { "muscles",    "m4 definitions passed to the skeleton" },
   { "tools",      "m4 invocation" },
-  { "m4",         "m4 traces" },
+  { "m4-early",   "m4 traces starting from the start" },
+  { "m4",         "m4 traces starting from the skeleton evaluation" },
   { "skeleton",   "skeleton postprocessing" },
   { "time",       "time consumption" },
   { "ielr",       "IELR conversion" },
@@ -279,6 +280,7 @@ static const argmatch_trace_arg argmatch_trace_args[] =
   { "sets",      trace_sets },
   { "muscles",   trace_muscles },
   { "tools",     trace_tools },
+  { "m4-early",  trace_m4_early },
   { "m4",        trace_m4 },
   { "skeleton",  trace_skeleton },
   { "time",      trace_time },
diff --git a/src/getargs.h b/src/getargs.h
index 39741926..c1629ab5 100644
--- a/src/getargs.h
+++ b/src/getargs.h
@@ -100,11 +100,12 @@ enum trace
     trace_grammar   = 1 << 7,  /**< Reading, reducing the grammar. */
     trace_time      = 1 << 8,  /**< Time consumption. */
     trace_skeleton  = 1 << 9,  /**< Skeleton postprocessing. */
-    trace_m4        = 1 << 10, /**< M4 traces. */
-    trace_muscles   = 1 << 11, /**< M4 definitions of the muscles. */
-    trace_ielr      = 1 << 12, /**< IELR conversion. */
-    trace_closure   = 1 << 13, /**< Input/output of closure(). */
-    trace_locations = 1 << 14, /**< Full display of locations. */
+    trace_m4_early  = 1 << 10, /**< M4 early traces. */
+    trace_m4        = 1 << 11, /**< M4 traces. */
+    trace_muscles   = 1 << 12, /**< M4 definitions of the muscles. */
+    trace_ielr      = 1 << 13, /**< IELR conversion. */
+    trace_closure   = 1 << 14, /**< Input/output of closure(). */
+    trace_locations = 1 << 15, /**< Full display of locations. */
     trace_all       = ~0       /**< All of the above.  */
   };
 /** What debug items bison displays during its run.  */
diff --git a/src/output.c b/src/output.c
index 5bee092d..e63c3254 100644
--- a/src/output.c
+++ b/src/output.c
@@ -687,6 +687,7 @@ output_skeleton (void)
   char *skeldir = xpath_join (datadir, "skeletons");
   char *m4sugar = xpath_join (datadir, "m4sugar/m4sugar.m4");
   char *m4bison = xpath_join (skeldir, "bison.m4");
+  char *traceon = xpath_join (skeldir, "traceon.m4");
   char *skel = (IS_PATH_WITH_DIR (skeleton)
                 ? xstrdup (skeleton)
                 : xpath_join (skeldir, skeleton));
@@ -701,7 +702,7 @@ output_skeleton (void)
   int filter_fd[2];
   pid_t pid;
   {
-    char const *argv[10];
+    char const *argv[11];
     int i = 0;
     argv[i++] = m4;
 
@@ -720,17 +721,16 @@ output_skeleton (void)
     argv[i++] = "-I";
     argv[i++] = datadir;
     /* Some future version of GNU M4 (most likely 1.6) may treat the
-       -dV in a position-dependent manner.  Keep it as the first
-       argument so that all files are traced.
-
-       See the thread starting at
+       -dV in a position-dependent manner.  See the thread starting at
        <http://lists.gnu.org/archive/html/bug-bison/2008-07/msg00000.html>
        for details.  */
-    if (trace_flag & trace_m4)
+    if (trace_flag & trace_m4_early)
       argv[i++] = "-dV";
     argv[i++] = m4sugar;
     argv[i++] = "-";
     argv[i++] = m4bison;
+    if (trace_flag & trace_m4)
+      argv[i++] = traceon;
     argv[i++] = skel;
     argv[i++] = NULL;
     aver (i <= ARRAY_CARDINALITY (argv));
@@ -751,6 +751,7 @@ output_skeleton (void)
   free (skeldir);
   free (m4sugar);
   free (m4bison);
+  free (traceon);
   free (skel);
 
   if (trace_flag & trace_muscles)