libdep plugin: fix bugs in parser and drop escaping

PR ld/31906
  * libdep_plugin.c (str2vec): Fix bug where null byte was not copied on memmove during quote handling and escaping, causing repeat of the last character in the last argument.
  Fix buffer overflow in **res when arguments were separated by `\t` instead of ` `.
  Remove handling of the escape character `\`, as it made it impossible to specify paths containing `\`
  -- the implementation merely dropped `\`, and was affected by the memmove bug, so this should not be breaking; just single and double quotes are sufficient to deal with white space and quote characters, there is no need for escaping.
  Handle syntax errors   on unterminated quotes.
  Make the parser linear time instead of   quadratic.

ld/libdep_plugin.c

@@ -132,88 +132,68 @@ get_libdeps (int fd)
   return rc;
 }
 
-/* Turn a string into an argvec.  */
+/* Parse arguments in-place as contiguous C-strings
-static char **
+   and return the number of arguments.  */
-str2vec (char *in)
+
+static int
+parse_libdep (char *str)
 {
-  char **res;
+  char *src, *dst;
-  char *s, *first, *end;
+  char quote;
-  char *sq, *dq;
+  int narg;
-  int i;
 
-  end = in + strlen (in);
+  src = dst = str;
-  s = in;
-  while (isspace ((unsigned char) *s)) s++;
-  first = s;
 
-  i = 1;
+  for (; isspace ((unsigned char) *src); ++src)
-  while ((s = strchr (s, ' ')))
+    ;
+
+  if (*src == '\0')
+    return 0;
+
+  narg = 1;
+  quote = 0;
+
+  while (*src)
     {
-      s++;
+      if (*src == '\'' || *src == '\"')
-      i++;
-    }
-  res = (char **)malloc ((i+1) * sizeof (char *));
-  if (!res)
-    return res;
-
-  i = 0;
-  sq = NULL;
-  dq = NULL;
-  res[0] = first;
-  for (s = first; *s; s++)
-    {
-      if (*s == '\\')
 	{
-	  memmove (s, s+1, end-s-1);
+	  if (!quote)
-	  end--;
+	    quote = *src++;
-	}
+	  else if (*src == quote)
-      if (isspace ((unsigned char) *s))
-	{
-	  if (sq || dq)
-	    continue;
-	  *s++ = '\0';
-	  while (isspace ((unsigned char) *s)) s++;
-	  if (*s)
-	    res[++i] = s;
-	}
-      if (*s == '\'' && !dq)
-	{
-	  if (sq)
 	    {
-	      memmove (sq, sq+1, s-sq-1);
+	      ++src;
-	      memmove (s-2, s+1, end-s-1);
+	      quote = 0;
-	      end -= 2;
-	      s--;
-	      sq = NULL;
 	    }
 	  else
-	    {
+	    *dst++ = *src++;
-	      sq = s;
-	    }
 	}
-      if (*s == '"' && !sq)
+      else if (!quote && isspace ((unsigned char) *src))
 	{
-	  if (dq)
+	  ++narg;
-	    {
+	  ++src;
-	      memmove (dq, dq+1, s-dq-1);
+	  *dst++ = '\0';
-	      memmove (s-2, s+1, end-s-1);
+	  for (; isspace ((unsigned char) *src); ++src);
-	      end -= 2;
-	      s--;
-	      dq = NULL;
-	    }
-	  else
-	    {
-	      dq = s;
-	    }
 	}
+      else
+	*dst++ = *src++;
     }
-  res[++i] = NULL;
+
-  return res;
+  *dst = '\0';
+
+  if (quote)
+    {
+      TV_MESSAGE (LDPL_WARNING,
+		  "libdep syntax error: unterminated quoted string");
+      return 0;
+    }
+
+  return narg;
 }
 
 static char *prevfile;
 
 /* Standard plugin API registerable hook.  */
+
 static enum ld_plugin_status
 onclaim_file (const struct ld_plugin_input_file *file, int *claimed)
 {
@@ -237,8 +217,8 @@ onclaim_file (const struct ld_plugin_input_file *file, int *claimed)
     return LDPS_ERR;
 
   /* This hook only gets called on actual object files.
-   * We have to examine the archive ourselves, to find
+     We have to examine the archive ourselves, to find
-   * our LIBDEPS member.  */
+     our LIBDEPS member.  */
   rv = get_libdeps (file->fd);
   if (rv == LDPS_ERR)
     return rv;
@@ -256,51 +236,51 @@ onclaim_file (const struct ld_plugin_input_file *file, int *claimed)
 }
 
 /* Standard plugin API registerable hook.  */
+
 static enum ld_plugin_status
 onall_symbols_read (void)
 {
   linerec *lr;
-  char **vec;
+  int nargs;
+  char const *arg;
   enum ld_plugin_status rv = LDPS_OK;
 
   while ((lr = line_head))
     {
       line_head = lr->next;
-      vec = str2vec (lr->line);
+      nargs = parse_libdep (lr->line);
-      if (vec)
+      arg = lr->line;
+
+      int i;
+      for (i = 0; i < nargs; i++, arg = strchr (arg, '\0') + 1)
 	{
-	  int i;
+	  if (arg[0] != '-')
-	  for (i = 0; vec[i]; i++)
 	    {
-	      if (vec[i][0] != '-')
+	      TV_MESSAGE (LDPL_WARNING, "ignoring libdep argument %s", arg);
-		{
+	      fflush (NULL);
-		  TV_MESSAGE (LDPL_WARNING, "ignoring libdep argument %s",
+	      continue;
-			      vec[i]);
-		  fflush (NULL);
-		  continue;
-		}
-	      if (vec[i][1] == 'l')
-		rv = tv_add_input_library (vec[i]+2);
-	      else if (vec[i][1] == 'L')
-		rv = tv_set_extra_library_path (vec[i]+2);
-	      else
-		{
-		  TV_MESSAGE (LDPL_WARNING, "ignoring libdep argument %s",
-			      vec[i]);
-		  fflush (NULL);
-		}
-	      if (rv != LDPS_OK)
-		break;
 	    }
-	  free (vec);
+	  if (arg[1] == 'l')
+	    rv = tv_add_input_library (arg + 2);
+	  else if (arg[1] == 'L')
+	    rv = tv_set_extra_library_path (arg + 2);
+	  else
+	    {
+	      TV_MESSAGE (LDPL_WARNING, "ignoring libdep argument %s", arg);
+	      fflush (NULL);
+	    }
+	  if (rv != LDPS_OK)
+	    break;
 	}
       free (lr);
     }
+
   line_tail = NULL;
   return rv;
 }
 
 /* Standard plugin API registerable hook.  */
+
 static enum ld_plugin_status
 oncleanup (void)
 {
@@ -309,20 +289,25 @@ oncleanup (void)
       free (prevfile);
       prevfile = NULL;
     }
+
   if (line_head)
     {
       linerec *lr;
+
       while ((lr = line_head))
 	{
 	  line_head = lr->next;
 	  free (lr);
 	}
+
       line_tail = NULL;
     }
+
   return LDPS_OK;
 }
 
 /* Standard plugin API entry point.  */
+
 enum ld_plugin_status
 onload (struct ld_plugin_tv *tv)
 {
@@ -351,6 +336,7 @@ onload (struct ld_plugin_tv *tv)
       (*tv_register_all_symbols_read) (onall_symbols_read);
       (*tv_register_cleanup) (oncleanup);
     }
+
   fflush (NULL);
   return LDPS_OK;
 }