[svn] Committed memory debugging stuff.

Published in <sxs1yw34pt4.fsf@florida.arsdigita.de>.

src/ChangeLog

@@ -1,3 +1,12 @@
+2000-11-22  Hrvoje Niksic  <hniksic@arsdigita.com>
+
+	* ftp.c (getftp): ftp_getaddress() returns a malloc'ed copy of the
+	string; no need to strdup() it.
+	(getftp): Make pwd_len a local variable.
+	(ftp_loop): Free PWD before returning.
+
+	* init.c (cleanup): Free opt.ftp_pass only if it's non-NULL.
+
 2000-11-22  Hrvoje Niksic  <hniksic@arsdigita.com>
 
 	* all: Use xfree() instead of free.

src/ftp.c

@@ -57,9 +57,11 @@ extern int h_errno;
 
 extern char ftp_last_respline[];
 
+/* #### Global variables??  These two should be members of struct
+   ccon!  */
+
 static enum stype host_type=ST_UNIX;
 static char *pwd;
-static int  pwd_len;
 
 /* Look for regexp "( *[0-9]+ *byte" (literal parenthesis) anywhere in
    the string S, and return the number converted to long, if found, 0
@@ -133,7 +135,7 @@ getftp (const struct urlinfo *u, long *len, long restval, ccon *con)
   search_netrc (u->host, (const char **)&user, (const char **)&passwd, 1);
   user = user ? user : opt.ftp_acc;
   if (!opt.ftp_pass)
-    opt.ftp_pass = xstrdup (ftp_getaddress ());
+    opt.ftp_pass = ftp_getaddress ();
   passwd = passwd ? passwd : opt.ftp_pass;
   assert (user && passwd);
 
@@ -279,7 +281,6 @@ Error in server response, closing control connection.\n"));
       if (!opt.server_response)
 	logprintf (LOG_VERBOSE, "==> PWD ... ");
       err = ftp_pwd(&con->rbuf, &pwd);
-      pwd_len = strlen(pwd);
       /* FTPRERR */
       switch (err)
       {
@@ -354,6 +355,7 @@ Error in server response, closing control connection.\n"));
              it to VMS style as VMS does not like leading slashes */
 	  if (*(u->dir) == '/')
 	    {
+	      int pwd_len = strlen (pwd);
 	      char *result = (char *)alloca (strlen (u->dir) + pwd_len + 10);
 	      *result = '\0';
 	      switch (host_type)
@@ -1634,6 +1636,8 @@ ftp_loop (struct urlinfo *u, int *dt)
   /* If a connection was left, quench it.  */
   if (rbuf_initialized_p (&con.rbuf))
     CLOSE (RBUF_FD (&con.rbuf));
+  FREE_MAYBE (pwd);
+  pwd = NULL;
   return res;
 }
 

src/host.c

@@ -372,6 +372,9 @@ sufmatch (const char **list, const char *what)
    function returns a short `username@' form, accepted by most
    anonymous servers.
 
+   The returned string is generated by malloc() and should be freed
+   using free().
+
    If not even the username cannot be divined, it means things are
    seriously fucked up, and Wget exits.  */
 char *

src/html-url.c

@@ -566,3 +566,10 @@ get_urls_html (const char *file, const char *this_url, int dash_p_leaf_HTML,
   read_file_free (fm);
   return closure.head;
 }
+
+void
+cleanup_html_url (void)
+{
+  FREE_MAYBE (interesting_tags);
+  FREE_MAYBE (interesting_attributes);
+}

src/init.c

@@ -981,6 +981,10 @@ check_user_specified_header (const char *s)
   return 1;
 }
 
+void cleanup_html_url PARAMS ((void));
+void downloaded_files_free PARAMS ((void));
+
+
 /* Free the memory allocated by global variables.  */
 void
 cleanup (void)
@@ -992,6 +996,8 @@ cleanup (void)
   free_netrc (netrc_list);
   if (opt.dfp)
     fclose (opt.dfp);
+  cleanup_html_url ();
+  downloaded_files_free ();
   FREE_MAYBE (opt.lfilename);
   xfree (opt.dir_prefix);
   FREE_MAYBE (opt.input_filename);
@@ -1004,7 +1010,7 @@ cleanup (void)
   free_vec (opt.follow_tags);
   free_vec (opt.ignore_tags);
   xfree (opt.ftp_acc);
-  xfree (opt.ftp_pass);
+  FREE_MAYBE (opt.ftp_pass);
   FREE_MAYBE (opt.ftp_proxy);
   FREE_MAYBE (opt.http_proxy);
   free_vec (opt.no_proxy);

src/main.c

@@ -809,6 +809,9 @@ Can't timestamp and not clobber old files at the same time.\n"));
     }
   log_close ();
   cleanup ();
+#ifdef DEBUG_MALLOC
+  print_malloc_debug_stats ();
+#endif
   if (status == RETROK)
     return 0;
   else

src/url.c

@@ -1577,6 +1577,14 @@ write_backup_file (const char *file, downloaded_file_t downloaded_file_return)
     }
 }
 
+typedef struct _downloaded_file_list {
+  char*                          file;
+  downloaded_file_t              download_type;
+  struct _downloaded_file_list*  next;
+} downloaded_file_list;
+
+static downloaded_file_list *downloaded_files;
+
 /* Remembers which files have been downloaded.  In the standard case, should be
    called with mode == FILE_DOWNLOADED_NORMALLY for each file we actually
    download successfully (i.e. not for ones we have failures on or that we skip
@@ -1592,15 +1600,7 @@ write_backup_file (const char *file, downloaded_file_t downloaded_file_return)
 downloaded_file_t
 downloaded_file (downloaded_file_t  mode, const char*  file)
 {
-  typedef struct _downloaded_file_list
-  {
-    char*                          file;
-    downloaded_file_t              download_type;
-    struct _downloaded_file_list*  next;
-  } downloaded_file_list;
-  
   boolean                       found_file = FALSE;
-  static downloaded_file_list*  downloaded_files = NULL;
   downloaded_file_list*         rover = downloaded_files;
 
   while (rover != NULL)
@@ -1628,6 +1628,19 @@ downloaded_file (downloaded_file_t  mode, const char*  file)
       return FILE_NOT_ALREADY_DOWNLOADED;
     }
 }
+
+void
+downloaded_files_free (void)
+{
+  downloaded_file_list*         rover = downloaded_files;
+  while (rover)
+    {
+      downloaded_file_list *next = rover->next;
+      xfree (rover->file);
+      xfree (rover);
+      rover = next;
+    }
+}
 
 /* Initialization of static stuff. */
 void

src/utils.c

@@ -60,80 +60,249 @@ Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.  */
 extern int errno;
 #endif
 
+/* This section implements several wrappers around the basic
+   allocation routines.  This is done for two reasons: first, so that
+   the callers of these functions need not consistently check for
+   errors.  If there is not enough virtual memory for running Wget,
+   something is seriously wrong, and Wget exits with an appropriate
+   error message.
+
+   The second reason why these are useful is that, if DEBUG_MALLOC is
+   defined, they also provide a handy (if crude) malloc debugging
+   interface that checks memory leaks.  */
 
 /* Croak the fatal memory error and bail out with non-zero exit
    status.  */
 static void
-memfatal (const char *s)
+memfatal (const char *what)
 {
   /* HACK: expose save_log_p from log.c, so we can turn it off in
      order to prevent saving the log.  Saving the log is dangerous
      because logprintf() and logputs() can call malloc(), so this
      could infloop.  When logging is turned off, infloop can no longer
-     happen.  */
+     happen.
+
+     #### This is no longer really necessary because the new routines
+     in log.c cons only if the line exceeds eighty characters.  But
+     this can come at the end of a line, so it's OK to be careful.
+
+     On a more serious note, it would be good to have a
+     log_forced_shutdown() routine that exposes this cleanly.  */
   extern int save_log_p;
 
   save_log_p = 0;
-  logprintf (LOG_ALWAYS, _("%s: %s: Not enough memory.\n"), exec_name, s);
+  logprintf (LOG_ALWAYS, _("%s: %s: Not enough memory.\n"), exec_name, what);
   exit (1);
 }
 
-/* xmalloc, xrealloc and xstrdup exit the program if there is not
-   enough memory.  xstrdup also implements strdup on systems that do
-   not have it.  xfree is provided to make leak-tracking easier.
-   Currently it's a no-op.  */
+/* These functions end with _real because they need to be
+   distinguished from the debugging functions, and from the macros.
+   Explanation follows:
+
+   If memory debugging is not turned on, wget.h defines these:
+
+     #define xmalloc xmalloc_real
+     #define xfree xfree_real
+     #define xrealloc xrealloc_real
+     #define xstrdup xstrdup_real
+
+   In case of memory debugging, the definitions are a bit more
+   complex, because we want to provide more information, *and* we want
+   to call the debugging code.  (The former is the reason why xmalloc
+   and friends need to be macros in the first place.)  Then it looks
+   like this:
+
+     #define xmalloc(a) xmalloc_debug (a, __FILE__, __LINE__)
+     #define xfree(a)   xfree_debug (a, __FILE__, __LINE__)
+     #define xrealloc(a, b) xrealloc_debug (a, b, __FILE__, __LINE__)
+     #define xstrdup(a) xstrdup_debug (a, __FILE__, __LINE__)
+
+   Each of the *_debug function does its magic and calls the real one.  */
 
 void *
-xmalloc (size_t size)
+xmalloc_real (size_t size)
 {
-  void *res;
-
-  res = malloc (size);
-  if (!res)
+  void *ptr = malloc (size);
+  if (!ptr)
     memfatal ("malloc");
-  return res;
+  return ptr;
 }
 
 void
-xfree (void *ptr)
+xfree_real (void *ptr)
 {
   free (ptr);
 }
 
 void *
-xrealloc (void *obj, size_t size)
+xrealloc_real (void *ptr, size_t newsize)
 {
-  void *res;
+  void *newptr;
 
   /* Not all Un*xes have the feature of realloc() that calling it with
      a NULL-pointer is the same as malloc(), but it is easy to
      simulate.  */
-  if (obj)
-    res = realloc (obj, size);
+  if (ptr)
+    newptr = realloc (ptr, newsize);
   else
-    res = malloc (size);
-  if (!res)
+    newptr = malloc (newsize);
+  if (!newptr)
     memfatal ("realloc");
-  return res;
+  return newptr;
 }
 
 char *
-xstrdup (const char *s)
+xstrdup_real (const char *s)
 {
+  char *copy;
+
 #ifndef HAVE_STRDUP
   int l = strlen (s);
-  char *s1 = malloc (l + 1);
-  if (!s1)
+  copy = malloc (l + 1);
+  if (!copy)
     memfatal ("strdup");
-  memcpy (s1, s, l + 1);
-  return s1;
+  memcpy (copy, s, l + 1);
 #else  /* HAVE_STRDUP */
-  char *s1 = strdup (s);
-  if (!s1)
+  copy = strdup (s);
+  if (!copy)
     memfatal ("strdup");
-  return s1;
 #endif /* HAVE_STRDUP */
+
+  return copy;
 }
+
+#ifdef DEBUG_MALLOC
+
+/* Crude home-grown routines for debugging some malloc-related
+   problems.  Featured:
+
+   * Counting the number of malloc and free invocations, and reporting
+     the "balance", i.e. how many times more malloc was called than it
+     was the case with free.
+
+   * Making malloc store its entry into a simple array and free remove
+     stuff from that array.  At the end, print the pointers which have
+     not been freed, along with the source file and the line number.
+     This also has the side-effect of detecting freeing memory that
+     was never allocated.
+
+   Note that this kind of memory leak checking strongly depends on
+   every malloc() being followed by a free(), even if the program is
+   about to finish.  Wget is careful to free the data structure it
+   allocated in init.c.  */
+
+static int malloc_count, free_count;
+
+static struct {
+  char *ptr;
+  const char *file;
+  int line;
+} malloc_debug[100000];
+
+/* Both register_ptr and unregister_ptr take O(n) operations to run,
+   which can be a real problem.  It would be nice to use a hash table
+   for malloc_debug, but the functions in hash.c are not suitable
+   because they can call malloc() themselves.  Maybe it would work if
+   the hash table were preallocated to a huge size, and if we set the
+   rehash threshold to 1.0.  */
+
+/* Register PTR in malloc_debug.  Abort if this is not possible
+   (presumably due to the number of current allocations exceeding the
+   size of malloc_debug.)  */
+
+static void
+register_ptr (void *ptr, const char *file, int line)
+{
+  int i;
+  for (i = 0; i < ARRAY_SIZE (malloc_debug); i++)
+    if (malloc_debug[i].ptr == NULL)
+      {
+	malloc_debug[i].ptr = ptr;
+	malloc_debug[i].file = file;
+	malloc_debug[i].line = line;
+	return;
+      }
+  abort ();
+}
+
+/* Unregister PTR from malloc_debug.  Abort if PTR is not present in
+   malloc_debug.  (This catches calling free() with a bogus pointer.)  */
+
+static void
+unregister_ptr (void *ptr)
+{
+  int i;
+  for (i = 0; i < ARRAY_SIZE (malloc_debug); i++)
+    if (malloc_debug[i].ptr == ptr)
+      {
+	malloc_debug[i].ptr = NULL;
+	return;
+      }
+  abort ();
+}
+
+/* Print the malloc debug stats that can be gathered from the above
+   information.  Currently this is the count of mallocs, frees, the
+   difference between the two, and the dump of the contents of
+   malloc_debug.  The last part are the memory leaks.  */
+
+void
+print_malloc_debug_stats (void)
+{
+  int i;
+  printf ("\nMalloc:  %d\nFree:    %d\nBalance: %d\n\n",
+	  malloc_count, free_count, malloc_count - free_count);
+  for (i = 0; i < ARRAY_SIZE (malloc_debug); i++)
+    if (malloc_debug[i].ptr != NULL)
+      printf ("0x%08ld: %s:%d\n", (long)malloc_debug[i].ptr,
+	      malloc_debug[i].file, malloc_debug[i].line);
+}
+
+void *
+xmalloc_debug (size_t size, const char *source_file, int source_line)
+{
+  void *ptr = xmalloc_real (size);
+  ++malloc_count;
+  register_ptr (ptr, source_file, source_line);
+  return ptr;
+}
+
+void
+xfree_debug (void *ptr, const char *source_file, int source_line)
+{
+  assert (ptr != NULL);
+  ++free_count;
+  unregister_ptr (ptr);
+  xfree_real (ptr);
+}
+
+void *
+xrealloc_debug (void *ptr, size_t newsize, const char *source_file, int source_line)
+{
+  void *newptr = xrealloc_real (ptr, newsize);
+  if (!ptr)
+    {
+      ++malloc_count;
+      register_ptr (newptr, source_file, source_line);
+    }
+  else
+    {
+      unregister_ptr (ptr);
+      register_ptr (newptr, source_file, source_line);
+    }
+  return newptr;
+}
+
+char *
+xstrdup_debug (const char *s, const char *source_file, int source_line)
+{
+  char *copy = xstrdup_real (s);
+  ++malloc_count;
+  register_ptr (copy, source_file, source_line);
+  return copy;
+}
+
+#endif /* DEBUG_MALLOC */
 
 /* Copy the string formed by two pointers (one on the beginning, other
    on the char after the last char) to a new, malloc-ed location.

src/utils.h

@@ -42,6 +42,10 @@ struct file_memory {
 char *time_str PARAMS ((time_t *));
 const char *uerrmsg PARAMS ((uerr_t));
 
+#ifdef DEBUG_MALLOC
+void print_malloc_debug_stats ();
+#endif
+
 char *strdupdelim PARAMS ((const char *, const char *));
 char **sepstring PARAMS ((const char *));
 int frontcmp PARAMS ((const char *, const char *));

src/wget.h

@@ -98,10 +98,31 @@ void debug_logprintf ();
 void logputs PARAMS ((enum log_options, const char *));
 
 /* Defined in `utils.c', but used literally everywhere.  */
-void *xmalloc PARAMS ((size_t));
-void xfree PARAMS ((void *));
-void *xrealloc PARAMS ((void *, size_t));
-char *xstrdup PARAMS ((const char *));
+#ifndef DEBUG_MALLOC
+
+#define xmalloc  xmalloc_real
+#define xfree    xfree_real
+#define xrealloc xrealloc_real
+#define xstrdup  xstrdup_real
+
+void *xmalloc_real PARAMS ((size_t));
+void xfree_real PARAMS ((void *));
+void *xrealloc_real PARAMS ((void *, size_t));
+char *xstrdup_real PARAMS ((const char *));
+
+#else  /* DEBUG_MALLOC */
+
+#define xmalloc(s)     xmalloc_debug (s, __FILE__, __LINE__)
+#define xfree(p)       xfree_debug (p, __FILE__, __LINE__)
+#define xrealloc(p, s) xrealloc_debug (p, s, __FILE__, __LINE__)
+#define xstrdup(p)     xstrdup_debug (p, __FILE__, __LINE__)
+
+void *xmalloc_debug PARAMS ((size_t, const char *, int));
+void xfree_debug PARAMS ((void *, const char *, int));
+void *xrealloc_debug PARAMS ((void *, size_t, const char *, int));
+char *xstrdup_debug PARAMS ((const char *, const char *, int));
+
+#endif /* DEBUG_MALLOC */
 
 /* #### Find a better place for this.  */
 /* The log file to which Wget writes to after HUP.  */