On 04/11/2017 03:18 AM, Jonas Bonn wrote:
For LTE, a context is created automatically when the modem registers
to the network. The read_settings function is called for these
automatic contexts to get their configuration.
---
drivers/qmimodem/gprs-context.c | 26 ++++++++++++++++++++++++++
1 file changed, 26 insertions(+)
I went ahead and applied this patch, but see below:
diff --git a/drivers/qmimodem/gprs-context.c b/drivers/qmimodem/gprs-context.c
index 1ae2a5a..3841c7e 100644
--- a/drivers/qmimodem/gprs-context.c
+++ b/drivers/qmimodem/gprs-context.c
@@ -153,6 +153,31 @@ done:
g_free(cbd);
}
+static void qmi_gprs_read_settings(struct ofono_gprs_context* gc,
+ unsigned int cid,
+ ofono_gprs_context_cb_t cb,
+ void *user_data)
+{
+ struct cb_data *cbd = cb_data_new(cb, user_data);
+ struct gprs_context_data *data = ofono_gprs_context_get_data(gc);
+
+ DBG("cid %u", cid);
+
+ data->active_context = cid;
+
+ cbd->user = gc;
+
+ if (qmi_service_send(data->wds, QMI_WDS_GET_SETTINGS, NULL,
+ get_settings_cb, cbd, NULL) > 0)
You really should be setting the qmi_destroy_func_t for
qmi_service_send. This saves you a g_free call inside get_settings_cb,
but more importantly it avoids a memory leak in case the
qmi_service/qmi_device is destroyed before the callback has been called.
I went ahead and fixed this in the upstream code in commits:
f29a316c918bfd67072bdb4f46ca90a3b8954108
7a2e198fd7430c468f2a15ff82467c18c649ddf3
But there may be other instances of this bug lurking around.
+ return;
+
+ data->active_context = 0;
+
+ CALLBACK_WITH_FAILURE(cb, cbd->data);
+
+ g_free(cbd);
+}
+
static void start_net_cb(struct qmi_result *result, void *user_data)
{
struct cb_data *cbd = user_data;
@@ -450,6 +475,7 @@ static struct ofono_gprs_context_driver driver = {
.remove = qmi_gprs_context_remove,
.activate_primary = qmi_activate_primary,
.deactivate_primary = qmi_deactivate_primary,
+ .read_settings = qmi_gprs_read_settings,
};
void qmi_gprs_context_init(void)
Regards,
-Denis
_______________________________________________
ofono mailing list
[email protected]
https://lists.ofono.org/mailman/listinfo/ofono
