Re: [MM] [PATCH v2] iface-modem, novatel-lte: disable network scan in LTE mode

[Ben Chan <benchan chromium org>]

Thanks for the notes.  Submitted patch v3.

- Ben

On Thu, Jan 3, 2013 at 3:58 PM, Aleksander Morgado <aleksander lanedo com> wrote:

> +static GList *
> +scan_networks_finish (MMIfaceModem3gpp *self,
> +                      GAsyncResult *res,
> +                      GError **error)
> +{
> +    const gchar *result;
> +
> +    result = mm_base_modem_at_command_finish (MM_BASE_MODEM (self), res, error);
> +    if (!result)
> +        return NULL;
> +
> +    return mm_3gpp_parse_cops_test_response (result, error);
> +}
> +
> +static void
> +scan_networks (MMIfaceModem3gpp *self,
> +               GAsyncReadyCallback callback,
> +               gpointer user_data)
> +{
> +    MMModemAccessTechnology access_tech;
> +
> +    mm_dbg ("scanning for networks (Novatel LTE)...");
> +
> +    access_tech = mm_iface_modem_get_access_technologies (MM_IFACE_MODEM (self));
> +    /* The Novatel LTE modem does not properly support AT+COPS=? in LTE mode.
> +     * Thus, do not try to scan networks when the current access technologies
> +     * include LTE.
> +     */
> +    if (access_tech & MM_MODEM_ACCESS_TECHNOLOGY_LTE) {
> +        GSimpleAsyncResult *simple;
> +        gchar *access_tech_string;
> +
> +        access_tech_string = mm_modem_access_technology_build_string_from_mask (access_tech);
> +        mm_warn ("Couldn't scan for networks with access technologies: %s", access_tech_string);
> +        simple = g_simple_async_result_new (G_OBJECT (self),
> +                                            callback,
> +                                            user_data,
> +                                            scan_networks);
> +        g_simple_async_result_set_error (simple,
> +                                         MM_CORE_ERROR,
> +                                         MM_CORE_ERROR_UNSUPPORTED,
> +                                         "Couldn't scan for networks with access technologies: %s",
> +                                         access_tech_string);
> +        g_simple_async_result_complete_in_idle (simple);
> +        g_object_unref (simple);
> +        g_free (access_tech_string);
> +        return;
> +    }
> +
> +    mm_base_modem_at_command (MM_BASE_MODEM (self),
> +                              "+COPS=?",
> +                              120,
> +                              FALSE,
> +                              callback,
> +                              user_data);
> +}

You cannot mix in the same async method a code execution path using
mm_base_modem_at_command() with another one using GSimpleAsyncResult and
completion in idle.  When you use mm_base_modem_at_command(), in
finish() you're expected to use mm_base_modem_at_command_finish(), like
you did. But when you use GSimpleAsyncResult and completion in idle you
should use g_simple_async_result_propagate_error() in finish(). The fact
that it may work as expected is due to how at_command_finish() is
implemented; but you shouldn't rely on that.

So you'll need to use GSimpleAsyncResult for both cases, i.e. provide a
_ready() GAsyncReadyCallback in mm_base_modem_at_command() to which you
pass 'simple' as user_data; and then complete the 'simple' from within
the _ready() method. In this way, you can safely call
g_simple_async_result_propagate_error() in finish().

See for example:
http://cgit.freedesktop.org/ModemManager/ModemManager/tree/plugins/simtech/mm-broadband-modem-simtech.c#n416

And btw, a hint for when you just always need to report an error in idle
in the async method (avoids the need to create a GSimpleAsyncResult
yourself):
http://developer.gnome.org/gio/unstable/GSimpleAsyncResult.html#g-simple-async-report-error-in-idle

--
Aleksander