Re: [PATCH 2/2] atmodem: timeout for gprs context deactivation
Hi Kalle,
Huawei modems have an annoying feature to hangup the chat line
after ppp shutdown. Basically what happens is this:
14:15:08 ofonod[11768]: > AT+CGACT=0,1\r
14:15:09 ofonod[11768]: EventChannel: < \r\n^BOOT:38645652,0,0,0,87\r\n
14:15:10 ofonod[11768]: EventChannel: < \r\n^RSSI:22\r\n
we might wanna at least decode the ^BOOT messages as extra debug
information.
ofono sends AT+CGACT to the modem but does not get any reply and
will
wait forever.
Workaround the issue by adding a five second timeout for the CGACT command.
This makes it possible to properly use huawei modems with connman. Without
this patch you can connect only once from connman and then you have to
restart connman.
We don't need to re-open the TTY or anything alike. Just waiting for the
timeout is enough? After that the TTY starts responding to commands
again?
ofono_gprs_context_up_cb_t up_cb; /* Up callback */
};
void *cb_data; /* Callback data */
+ gint cgact_timeout;
};
The timeouts are GSources and so normally guint actually.
static void at_cgact_down_cb(gboolean ok, GAtResult *result,
gpointer user_data)
{
struct cb_data *cbd = user_data;
@@ -73,6 +96,13 @@ static void at_cgact_down_cb(gboolean ok, GAtResult *result, gpointer
user_data)
struct gprs_context_data *gcd = ofono_gprs_context_get_data(gc);
struct ofono_error error;
+ if (gcd->cgact_timeout == 0)
+ /* already timed out */
+ return;
+
In theory you are correct with no { } since it is a single statement,
but if you put a comment on top of it, then please use the { } to make
it easier for the reader.
+ g_source_remove(gcd->cgact_timeout);
+ gcd->cgact_timeout = 0;
+
if (ok)
gcd->active_context = 0;
@@ -136,6 +166,11 @@ static void ppp_disconnect(GAtPPPDisconnectReason reason, gpointer
user_data)
g_free(cbd);
goto out;
}
+
+ /* workaround for huawei hangup problem */
+ gcd->cgact_timeout = g_timeout_add_seconds(5, cgact_timeout,
+ cbd);
+
I think we should include this into a Huawei specific quirk.
Regards
Marcel