[PATCH/RFC] analyzer: workaround for nested pp_printf

Message ID 20200206193923.10530-1-dmalcolm@redhat.com
State New
Headers show
Series
  • [PATCH/RFC] analyzer: workaround for nested pp_printf
Related show

Commit Message

David Malcolm Feb. 6, 2020, 7:39 p.m.
The dumps from the analyzer sometimes contain garbled output.

The root cause is due to nesting of calls to pp_printf: I'm using
pp_printf with %qT to print types with a PP using default_tree_printer.

default_tree_printer handles 'T' (and various other codes) via
  dump_generic_node (pp, t, 0, TDF_SLIM, 0);
and dump_generic_node can call pp_printf in various ways, leading
to a pp_printf within a pp_printf, and garbled output.

I don't think it's feasible to fix pp_printf to be reentrant, in
stage 4, at least, so for the moment this patch works around it
in the analyzer.

Successfully bootstrapped & regrtested on x86_64-pc-linux-gnu.

OK for master?  Or am I missing something here?  (in that I'm
surprised that this problem isn't more widespread)

Thanks
Dave

gcc/analyzer/ChangeLog:
	* region-model.cc (print_quoted_type): New function.
	(svalue::print): Use it to replace %qT.
	(region::dump_to_pp): Likewise.
	(region::dump_child_label): Likewise.
	(region::print_fields): Likewise.
---
 gcc/analyzer/region-model.cc | 35 +++++++++++++++++++++++++++++++----
 1 file changed, 31 insertions(+), 4 deletions(-)

-- 
2.21.0

Patch

diff --git a/gcc/analyzer/region-model.cc b/gcc/analyzer/region-model.cc
index c837ec6ed3b..571ae6e4d56 100644
--- a/gcc/analyzer/region-model.cc
+++ b/gcc/analyzer/region-model.cc
@@ -73,6 +73,25 @@  dump_tree (pretty_printer *pp, tree t)
   dump_generic_node (pp, t, 0, TDF_SLIM, 0);
 }
 
+/* Equivalent to pp_printf (pp, "%qT", t), to avoid nesting pp_printf
+   calls within other pp_printf calls.
+
+   default_tree_printer handles 'T' and some other codes by calling
+     dump_generic_node (pp, t, 0, TDF_SLIM, 0);
+   dump_generic_node calls pp_printf in various places, leading to
+   garbled output.
+
+   Ideally pp_printf could be made to be reentrant, but in the meantime
+   this function provides a workaround.  */
+
+static void
+print_quoted_type (pretty_printer *pp, tree t)
+{
+  pp_begin_quote (pp, pp_show_color (pp));
+  dump_generic_node (pp, t, 0, TDF_SLIM, 0);
+  pp_end_quote (pp, pp_show_color (pp));
+}
+
 /* Dump this path_var to PP (which must support %E for trees).
 
    Express the stack depth using an "@DEPTH" suffix, so e.g. given
@@ -319,7 +338,9 @@  svalue::print (const region_model &model,
   if (m_type)
     {
       gcc_assert (TYPE_P (m_type));
-      pp_printf (pp, "type: %qT, ", m_type);
+      pp_string (pp, "type: ");
+      print_quoted_type (pp, m_type);
+      pp_string (pp, ", ");
     }
 
   /* vfunc.  */
@@ -1282,7 +1303,8 @@  region::dump_to_pp (const region_model &model,
     }
   if (m_type)
     {
-      pp_printf (pp, "%s type: %qT", field_prefix, m_type);
+      pp_printf (pp, "%s type: ", field_prefix);
+      print_quoted_type (pp, m_type);
       pp_newline (pp);
     }
 
@@ -1332,7 +1354,9 @@  region::dump_child_label (const region_model &model,
 	pp_string (pp, "active ");
       else
 	pp_string (pp, "inactive ");
-      pp_printf (pp, "view as %qT: ", child->get_type ());
+      pp_string (pp, "view as ");
+      print_quoted_type (pp, child->get_type ());
+      pp_string (pp, ": ");
     }
 }
 
@@ -1463,7 +1487,10 @@  region::print_fields (const region_model &model ATTRIBUTE_UNUSED,
   m_sval_id.print (pp);
 
   if (m_type)
-    pp_printf (pp, ", type: %qT", m_type);
+    {
+      pp_printf (pp, ", type: ");
+      print_quoted_type (pp, m_type);
+    }
 }
 
 /* Determine if a pointer to this region must be non-NULL.