GTK troubles...again...

Hello,
this is my second post today, and its the second question.

I've got a programm, which creates a seg fault after some days. I dont know, where this fault is generated.
Now i tried to analyse the programm. Therefore i used purify. And now the problem beginns.
With purify i get thousands and thousands of "Uninitialized memory read" errors.
But those errors seem to be in the gtk library.
I will post here a function which creates such errors, maybe somebody had the same problem and knows, how to solve it.

Code:

---

static GtkWidget *get_search_area () {
        GtkWidget *input_field;
        GtkWidget *search_button;
        GtkWidget *hboxSearch;
        GList *glist = NULL;
       
        // input field
        input_field = gtk_entry_new_with_max_length(5);
        gtk_entry_set_width_chars(GTK_ENTRY(input_field), 5);

        //combobox
        combo_box = gtk_combo_new();
       
        glist = g_list_append (glist, "All");
            glist = g_list_append (glist, "K15-U");

        search_button = gtk_button_new_with_label ("Update Users");
        g_signal_connect (G_OBJECT (search_button), "clicked",
        G_CALLBACK (update_button), input_field);
        gtk_widget_set_size_request (GTK_WIDGET (search_button), 80, 20);
       
        gtk_widget_show (input_field);
        gtk_widget_show (combo_box);
        gtk_widget_show (search_button);

        hboxSearch = gtk_hbox_new (FALSE, 20);
        gtk_box_pack_start (GTK_BOX (hboxSearch), input_field, FALSE, TRUE, 5); //This line is reported by purify
        gtk_box_pack_start (GTK_BOX (hboxSearch), combo_box, FALSE, TRUE, 5);
        gtk_box_pack_start (GTK_BOX (hboxSearch), search_button, FALSE, TRUE, 0);
       
        gtk_widget_show(hboxSearch);

        return hboxSearch;
}

---

Here is the purify output of this function:

Code:

---

      This is occurring while in:
            gtk_hbox_size_allocate [libgtk-x11-2.0.so.0]
            g_closure_invoke [libgobject-2.0.so.0]
            signal_emit_unlocked_R [libgobject-2.0.so.0]
            g_signal_emit_valist [libgobject-2.0.so.0]
            gtk_signal_emit [libgtk-x11-2.0.so.0]
            gtk_widget_size_allocate [libgtk-x11-2.0.so.0]
      Reading 4 bytes from 0x11ec84 in the heap (1 byte at 0x11ec87 uninit).
      Address 0x11ec84 is 4 bytes into a malloc'd block at 0x11ec80 of 8 bytes.
      This block was allocated from:
            malloc        [rtlib.o]
            g_malloc      [libglib-2.0.so.0]
            gtk_box_pack_start [libgtk-x11-2.0.so.0]
            get_search_area [MainWindow.c:256]
                      gtk_widget_show (search_button);
                      // hbox to put the search button and input field in
                      hboxSearch = gtk_hbox_new (FALSE, 20);
            =>        gtk_box_pack_start (GTK_BOX (hboxSearch), input_field, FALSE, TRUE, 5);
                      gtk_box_pack_start (GTK_BOX (hboxSearch), combo_box, FALSE, TRUE, 5);
                      gtk_box_pack_start (GTK_BOX (hboxSearch), search_button, FALSE, TRUE, 0);

---

I hope anybody knows this behaviour of gtk or my code, because i haven't got a clue.
Thank you

First thing to do with a new tool like that is start with your "hello world" equivalent GTK program and observe its behaviour on something which is "obviously" correct.

> With purify i get thousands and thousands of "Uninitialized memory read" errors.
> But those errors seem to be in the gtk library.
If you pass a pointer to uninitialised memory to GTK, then it might appear that the library is reading uninitialised memory.

Does your malloc have an option to fill allocated memory with a pattern (rather than either zeros or whatever was there from last time). Ditto for free, fill deallocated memory with a pattern to trap sneaky use after free problems.

Valgrind didn't found anything wrong with that code, but from your other post I assume that you're compiling it on Solaris.
You _coud_ try not to use deprecated widgets and functions... Migrating from GtkCombo to GtkComboBoxEntry and gtk_entry_new_with_max_length() (read the "Note").

I have not read the entire code closely--I will do so later--but I notice that you are using deprecated functions. For example, gtk_entry_new_with_max_length() is a deprecated function and should not be used in newly written code.

Another function that might be a source of problems is gtk_combo_new(). My GTK+ documentation does not even list it. I have only ever used gtk_combo_box_new_text() and I have never experienced a problem, so if the entries are only going to be text entries, you might want to look into this latter function instead.

Like I said, I have not read the code very closely. I will come back to it later.

Ok, you're right, those functions are deprecated. But GtkComboBoxEntry for example exists since GTK 2.4. The problem now is, that i have to use GTK 2.0. This is the actual stable Solaris version, as far as i know. If somebody can correct me here, please do so.

I also testet a simple helloWorld program, same behaviour. Only by creating the window, there are several Memory Errors.
By resizing the window, many more errors are created.