* cp-demangle.h: Add comment explaining what to do to avoid

	overrunning string.
	(d_check_char): New.
	(d_next_char): Don't advance past trailing '\0'.
	* cp-demangle.c (cplus_demangle_mangled_name): Use d_check_char.
	(d_nested_name): Likewise.
	(d_special_name): Likewise.
	(d_call_offset): Likewise.
	(d_function_type): Likewise.
	(d_array_type): Likewise.
	(d_pointer_to_member_type): Likewise.
	(d_template_param): Likewise.
	(d_template_args): Likewise.
	(d_template_arg): Likewise.
	(d_expr_primary): Likewise.
	(d_local_name): Likewise.
	(d_substitution): Likewise.
	(d_ctor_dtor_name): Use d_advance rather than d_next_char.
	* testsuite/test-demangle.c: Include sys/mman.h.
	(MAP_ANONYMOUS): Define.
	(protect_end): New.
	(main): Use protect_end.
	* testsuite/demangle-expected: Add testcases for overrunning
	the end of the string.

5 files changed, 136 insertions, 34 deletions

diff --git a/libiberty/ChangeLog b/libiberty/ChangeLog
index 4b27eb0..e72f845 100644
--- a/libiberty/ChangeLog
+++ b/libiberty/ChangeLog
@@ -1,3 +1,30 @@
+2006-12-20  Geoffrey Keating  <geoffk@apple.com>
+
+	* cp-demangle.h: Add comment explaining what to do to avoid
+	overrunning string.
+	(d_check_char): New.
+	(d_next_char): Don't advance past trailing '\0'.
+	* cp-demangle.c (cplus_demangle_mangled_name): Use d_check_char.
+	(d_nested_name): Likewise.
+	(d_special_name): Likewise.
+	(d_call_offset): Likewise.
+	(d_function_type): Likewise.
+	(d_array_type): Likewise.
+	(d_pointer_to_member_type): Likewise.
+	(d_template_param): Likewise.
+	(d_template_args): Likewise.
+	(d_template_arg): Likewise.
+	(d_expr_primary): Likewise.
+	(d_local_name): Likewise.
+	(d_substitution): Likewise.
+	(d_ctor_dtor_name): Use d_advance rather than d_next_char.
+	* testsuite/test-demangle.c: Include sys/mman.h.
+	(MAP_ANONYMOUS): Define.
+	(protect_end): New.
+	(main): Use protect_end.
+	* testsuite/demangle-expected: Add testcases for overrunning
+	the end of the string.
+
 2006-11-30  Andrew Stubbs  <andrew.stubbs@st.com>
             J"orn Rennecke <joern.rennecke@st.com>
 
diff --git a/libiberty/cp-demangle.c b/libiberty/cp-demangle.c
index 2fa5978..ac1dfe5 100644
--- a/libiberty/cp-demangle.c
+++ b/libiberty/cp-demangle.c
@@ -913,9 +913,9 @@ CP_STATIC_IF_GLIBCPP_V3
 struct demangle_component *
 cplus_demangle_mangled_name (struct d_info *di, int top_level)
 {
-  if (d_next_char (di) != '_')
+  if (! d_check_char (di, '_'))
     return NULL;
-  if (d_next_char (di) != 'Z')
+  if (! d_check_char (di, 'Z'))
     return NULL;
   return d_encoding (di, top_level);
 }
@@ -1123,7 +1123,7 @@ d_nested_name (struct d_info *di)
   struct demangle_component *ret;
   struct demangle_component **pret;
 
-  if (d_next_char (di) != 'N')
+  if (! d_check_char (di, 'N'))
     return NULL;
 
   pret = d_cv_qualifiers (di, &ret, 1);
@@ -1134,7 +1134,7 @@ d_nested_name (struct d_info *di)
   if (*pret == NULL)
     return NULL;
 
-  if (d_next_char (di) != 'E')
+  if (! d_check_char (di, 'E'))
     return NULL;
 
   return ret;
@@ -1449,11 +1449,8 @@ d_operator_name (struct d_info *di)
 static struct demangle_component *
 d_special_name (struct d_info *di)
 {
-  char c;
-
   di->expansion += 20;
-  c = d_next_char (di);
-  if (c == 'T')
+  if (d_check_char (di, 'T'))
     {
       switch (d_next_char (di))
 	{
@@ -1502,7 +1499,7 @@ d_special_name (struct d_info *di)
 	    offset = d_number (di);
 	    if (offset < 0)
 	      return NULL;
-	    if (d_next_char (di) != '_')
+	    if (! d_check_char (di, '_'))
 	      return NULL;
 	    base_type = cplus_demangle_type (di);
 	    /* We don't display the offset.  FIXME: We should display
@@ -1523,7 +1520,7 @@ d_special_name (struct d_info *di)
 	  return NULL;
 	}
     }
-  else if (c == 'G')
+  else if (d_check_char (di, 'G'))
     {
       switch (d_next_char (di))
 	{
@@ -1570,14 +1567,14 @@ d_call_offset (struct d_info *di, int c)
   else if (c == 'v')
     {
       d_number (di);
-      if (d_next_char (di) != '_')
+      if (! d_check_char (di, '_'))
 	return 0;
       d_number (di);
     }
   else
     return 0;
 
-  if (d_next_char (di) != '_')
+  if (! d_check_char (di, '_'))
     return 0;
 
   return 1;
@@ -1601,13 +1598,13 @@ d_ctor_dtor_name (struct d_info *di)
       else if (di->last_name->type == DEMANGLE_COMPONENT_SUB_STD)
 	di->expansion += di->last_name->u.s_string.len;
     }
-  switch (d_next_char (di))
+  switch (d_peek_char (di))
     {
     case 'C':
       {
 	enum gnu_v3_ctor_kinds kind;
 
-	switch (d_next_char (di))
+	switch (d_peek_next_char (di))
 	  {
 	  case '1':
 	    kind = gnu_v3_complete_object_ctor;
@@ -1621,6 +1618,7 @@ d_ctor_dtor_name (struct d_info *di)
 	  default:
 	    return NULL;
 	  }
+	d_advance (di, 2);
 	return d_make_ctor (di, kind, di->last_name);
       }
 
@@ -1628,7 +1626,7 @@ d_ctor_dtor_name (struct d_info *di)
       {
 	enum gnu_v3_dtor_kinds kind;
 
-	switch (d_next_char (di))
+	switch (d_peek_next_char (di))
 	  {
 	  case '0':
 	    kind = gnu_v3_deleting_dtor;
@@ -1642,6 +1640,7 @@ d_ctor_dtor_name (struct d_info *di)
 	  default:
 	    return NULL;
 	  }
+	d_advance (di, 2);
 	return d_make_dtor (di, kind, di->last_name);
       }
 
@@ -1925,7 +1924,7 @@ d_function_type (struct d_info *di)
 {
   struct demangle_component *ret;
 
-  if (d_next_char (di) != 'F')
+  if (! d_check_char (di, 'F'))
     return NULL;
   if (d_peek_char (di) == 'Y')
     {
@@ -1934,7 +1933,7 @@ d_function_type (struct d_info *di)
       d_advance (di, 1);
     }
   ret = d_bare_function_type (di, 1);
-  if (d_next_char (di) != 'E')
+  if (! d_check_char (di, 'E'))
     return NULL;
   return ret;
 }
@@ -2021,7 +2020,7 @@ d_array_type (struct d_info *di)
   char peek;
   struct demangle_component *dim;
 
-  if (d_next_char (di) != 'A')
+  if (! d_check_char (di, 'A'))
     return NULL;
 
   peek = d_peek_char (di);
@@ -2049,7 +2048,7 @@ d_array_type (struct d_info *di)
 	return NULL;
     }
 
-  if (d_next_char (di) != '_')
+  if (! d_check_char (di, '_'))
     return NULL;
 
   return d_make_comp (di, DEMANGLE_COMPONENT_ARRAY_TYPE, dim,
@@ -2065,7 +2064,7 @@ d_pointer_to_member_type (struct d_info *di)
   struct demangle_component *mem;
   struct demangle_component **pmem;
 
-  if (d_next_char (di) != 'M')
+  if (! d_check_char (di, 'M'))
     return NULL;
 
   cl = cplus_demangle_type (di);
@@ -2109,7 +2108,7 @@ d_template_param (struct d_info *di)
 {
   long param;
 
-  if (d_next_char (di) != 'T')
+  if (! d_check_char (di, 'T'))
     return NULL;
 
   if (d_peek_char (di) == '_')
@@ -2122,7 +2121,7 @@ d_template_param (struct d_info *di)
       param += 1;
     }
 
-  if (d_next_char (di) != '_')
+  if (! d_check_char (di, '_'))
     return NULL;
 
   ++di->did_subs;
@@ -2144,7 +2143,7 @@ d_template_args (struct d_info *di)
      constructor or destructor.  */
   hold_last_name = di->last_name;
 
-  if (d_next_char (di) != 'I')
+  if (! d_check_char (di, 'I'))
     return NULL;
 
   al = NULL;
@@ -2189,7 +2188,7 @@ d_template_arg (struct d_info *di)
     case 'X':
       d_advance (di, 1);
       ret = d_expression (di);
-      if (d_next_char (di) != 'E')
+      if (! d_check_char (di, 'E'))
 	return NULL;
       return ret;
 
@@ -2316,7 +2315,7 @@ d_expr_primary (struct d_info *di)
 {
   struct demangle_component *ret;
 
-  if (d_next_char (di) != 'L')
+  if (! d_check_char (di, 'L'))
     return NULL;
   if (d_peek_char (di) == '_')
     ret = cplus_demangle_mangled_name (di, 0);
@@ -2362,7 +2361,7 @@ d_expr_primary (struct d_info *di)
 	}
       ret = d_make_comp (di, t, type, d_make_name (di, s, d_str (di) - s));
     }
-  if (d_next_char (di) != 'E')
+  if (! d_check_char (di, 'E'))
     return NULL;
   return ret;
 }
@@ -2376,12 +2375,12 @@ d_local_name (struct d_info *di)
 {
   struct demangle_component *function;
 
-  if (d_next_char (di) != 'Z')
+  if (! d_check_char (di, 'Z'))
     return NULL;
 
   function = d_encoding (di, 0);
 
-  if (d_next_char (di) != 'E')
+  if (! d_check_char (di, 'E'))
     return NULL;
 
   if (d_peek_char (di) == 's')
@@ -2486,7 +2485,7 @@ d_substitution (struct d_info *di, int prefix)
 {
   char c;
 
-  if (d_next_char (di) != 'S')
+  if (! d_check_char (di, 'S'))
     return NULL;
 
   c = d_next_char (di);
diff --git a/libiberty/cp-demangle.h b/libiberty/cp-demangle.h
index 2517a57e..920ca47 100644
--- a/libiberty/cp-demangle.h
+++ b/libiberty/cp-demangle.h
@@ -123,10 +123,16 @@ struct d_info
   int expansion;
 };
 
+/* To avoid running past the ending '\0', don't:
+   - call d_peek_next_char if d_peek_char returned '\0'
+   - call d_advance with an 'i' that is too large
+   - call d_check_char(di, '\0')
+   Everything else is safe.  */
 #define d_peek_char(di) (*((di)->n))
 #define d_peek_next_char(di) ((di)->n[1])
 #define d_advance(di, i) ((di)->n += (i))
-#define d_next_char(di) (*((di)->n++))
+#define d_check_char(di, c) (d_peek_char(di) == c ? ((di)->n++, 1) : 0)
+#define d_next_char(di) (d_peek_char(di) == '\0' ? '\0' : *((di)->n++))
 #define d_str(di) ((di)->n)
 
 /* Functions and arrays in cp-demangle.c which are referenced by
diff --git a/libiberty/testsuite/demangle-expected b/libiberty/testsuite/demangle-expected
index 3f5622f..56d9f89 100644
--- a/libiberty/testsuite/demangle-expected
+++ b/libiberty/testsuite/demangle-expected
@@ -3816,3 +3816,25 @@ f
 SASDASDFASDF_sdfsdf
 SASDASDFASDF_sdfsdf
 SASDASDFASDF_sdfsdf
+# These are all cases of invalid manglings where the demangler would read
+# past the end of the string.
+# d_name wasn't honouring a NULL from d_substitution
+--format=gnu-v3
+_ZSA
+_ZSA
+# d_expr_primary wasn't honouring NULL from cplus_demangle_mangled_name
+--format=gnu-v3
+_ZN1fIL_
+_ZN1fIL_
+# d_operator_name was taking two characters in a row
+--format=gnu-v3
+_Za
+_Za
+# d_prefix wasn't honouring NULL from d_substitution
+--format=gnu-v3
+_ZNSA
+_ZNSA
+# d_prefix wasn't honouring NULL from d_template_param
+--format=gnu-v3
+_ZNT
+_ZNT
diff --git a/libiberty/testsuite/test-demangle.c b/libiberty/testsuite/test-demangle.c
index 9379399..12b07dd 100644
--- a/libiberty/testsuite/test-demangle.c
+++ b/libiberty/testsuite/test-demangle.c
@@ -86,6 +86,50 @@ getline(buf)
   buf->alloced = alloc;
 }
 
+/* If we have mmap() and mprotect(), copy the string S just before a
+   protected page, so that if the demangler runs over the end of the
+   string we'll get a fault, and return the address of the new string.
+   If no mmap, or it fails, or it looks too hard, just return S.  */
+
+#ifdef HAVE_SYS_MMAN_H
+#include <sys/mman.h>
+#endif
+#if defined(MAP_ANON) && ! defined (MAP_ANONYMOUS)
+#define MAP_ANONYMOUS MAP_ANON
+#endif
+
+static const char *
+protect_end (const char * s)
+{
+#if defined(HAVE_MMAP) && defined (MAP_ANONYMOUS)
+  size_t pagesize = getpagesize();
+  static char * buf;
+  size_t s_len = strlen (s);
+  char * result;
+  
+  /* Don't try if S is too long.  */
+  if (s_len >= pagesize)
+    return s;
+
+  /* Allocate one page of allocated space followed by an unmapped
+     page.  */
+  if (buf == NULL)
+    {
+      buf = mmap (NULL, pagesize * 2, PROT_READ | PROT_WRITE,
+		  MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+      if (! buf)
+	return s;
+      munmap (buf + pagesize, pagesize);
+    }
+  
+  result = buf + (pagesize - s_len - 1);
+  memcpy (result, s, s_len + 1);
+  return result;
+#else
+  return s;
+#endif
+}
+
 static void
 fail (lineno, opts, in, out, exp)
      int lineno;
@@ -150,6 +194,8 @@ main(argc, argv)
 
   for (;;)
     {
+      const char *inp;
+      
       getline (&format);
       if (feof (stdin))
 	break;
@@ -157,6 +203,8 @@ main(argc, argv)
       getline (&input);
       getline (&expect);
 
+      inp = protect_end (input.data);
+
       tests++;
 
       no_params = 0;
@@ -237,14 +285,14 @@ main(argc, argv)
 	    {
 	      enum gnu_v3_ctor_kinds kc;
 
-	      kc = is_gnu_v3_mangled_ctor (input.data);
+	      kc = is_gnu_v3_mangled_ctor (inp);
 	      sprintf (buf, "%d", (int) kc);
 	    }
 	  else
 	    {
 	      enum gnu_v3_dtor_kinds kd;
 
-	      kd = is_gnu_v3_mangled_dtor (input.data);
+	      kd = is_gnu_v3_mangled_dtor (inp);
 	      sprintf (buf, "%d", (int) kd);
 	    }
 
@@ -259,7 +307,7 @@ main(argc, argv)
 
       cplus_demangle_set_style (style);
 
-      result = cplus_demangle (input.data,
+      result = cplus_demangle (inp,
 			       DMGL_PARAMS|DMGL_ANSI|DMGL_TYPES
 			       |(ret_postfix ? DMGL_RET_POSTFIX : 0));
 
@@ -275,7 +323,7 @@ main(argc, argv)
       if (no_params)
 	{
 	  getline (&expect);
-	  result = cplus_demangle (input.data, DMGL_ANSI|DMGL_TYPES);
+	  result = cplus_demangle (inp, DMGL_ANSI|DMGL_TYPES);
 
 	  if (result
 	      ? strcmp (result, expect.data)