Re: [Rhythmbox-devel] Segmentation fault during property read

["Jonathan Matthew" <notverysmart gmail com>]

On Tue, Mar 4, 2008 at 7:16 AM, Bartosz Wiklak <bwiklak gmail com> wrote:
> Hi, I'm doing something like that:
>
>         RBSource *source;
>         RhythmDBQueryModel *model;
>         RhythmDBEntry *entry1;
>         GtkTreeIter iter;
>         gboolean valid;
>         gchar buff[1024];
>
>     if(plugin->shell==NULL) g_print("NIC\n");
>
>         g_object_get (plugin->shell, "selected-source", &source, NULL);
>     g_object_get (source, "query-model", &model, NULL);

You need to unref objects you get via g_object_set when you're done with them

>         allSongsList = (SongInfo*)g_malloc( covers_count*sizeof(SongInfo) );

What is covers_count, and where is it being set to a useful value?  It
looks like you're allocating a zero length array here, so it'll crash
the first time you try to write to it...

>
>         valid = gtk_tree_model_get_iter_first (model, &iter);
>         covers_count=0;
>         while (valid)
>         {
>                 gchar *str_album = NULL;
>                 gchar *str_artist = NULL;
>
>
>         gtk_tree_model_get (model, &iter,  0, &entry1,-1);
>         str_album = rhythmdb_entry_get_string(entry1,RHYTHMDB_PROP_ALBUM);
>                 str_artist = rhythmdb_entry_get_string(entry1,RHYTHMDB_PROP_ARTIST);
>
>         if(str_artist){
>
>
>
>                        allSongsList[covers_count].songPath =
>  rhythmdb_entry_get_string(entry1,RHYTHMDB_PROP_LOCATION);

.. which is here.

>
>
>                         allSongsList[covers_count].coverPath= (gchar*)g_malloc( strlen(buff)+1 );
>                         strcpy(allSongsList[covers_count].coverPath,buff);
>                         g_print(allSongsList[covers_count].songPath);
>
>
>                 }
>                 valid = gtk_tree_model_iter_next (model, &iter);
>                 covers_count++;
>         }
>
>  When I'm tying to get file location (RHYTHMDB_PROP_LOCATION) I'm
>  getting segmentation fault.
>  What is wrong?

I'd suggest that gdb makes a better debugger than this mailing list.