Re: [RFC 3/3] gprs: Release context in driver and core when network is lost

[Guillaume Zajac <guillaume.zajac@linux.intel.com>]

[-- Attachment #1: Type: text/plain, Size: 4703 bytes --]

Hi Denis,

On 01/06/2012 00:45, Denis Kenzior wrote:
> Hi Guillaume,
>
> On 05/31/2012 04:59 AM, Guillaume Zajac wrote:
>> ---
>>   src/gprs.c |   84 
>> +++++++++++++++++++++++++++++++++++++++++++++---------------
>>   1 files changed, 63 insertions(+), 21 deletions(-)
>>
>> diff --git a/src/gprs.c b/src/gprs.c
>> index 994607d..9ec0ca1 100644
>> --- a/src/gprs.c
>> +++ b/src/gprs.c
>> @@ -48,6 +48,7 @@
>>
>>   #define GPRS_FLAG_ATTACHING 0x1
>>   #define GPRS_FLAG_RECHECK 0x2
>> +#define GPRS_FLAG_ATTACHED_UPDATE 0x4
>>
>>   #define SETTINGS_STORE "gprs"
>>   #define SETTINGS_GROUP "Settings"
>> @@ -1440,6 +1441,44 @@ void ofono_gprs_resume_notify(struct 
>> ofono_gprs *gprs)
>>       update_suspended_property(gprs, FALSE);
>>   }
>>
>> +static gboolean check_active_contexts(struct ofono_gprs *gprs)
>
> Name this have_active_contexts

Ok

>
>> +{
>> +    GSList *l;
>> +    struct pri_context *ctx;
>> +
>> +    for (l = gprs->contexts; l; l = l->next) {
>> +        ctx = l->data;
>> +
>> +        if (ctx->active == TRUE)
>> +            return TRUE;
>> +    }
>> +
>> +    return FALSE;
>> +}
>> +
>> +static void release_active_contexts(struct ofono_gprs *gprs)
>> +{
>> +    GSList *l;
>> +    struct pri_context *ctx;
>> +
>> +    for (l = gprs->contexts; l; l = l->next) {
>> +        struct ofono_gprs_context *gc;
>> +
>> +        ctx = l->data;
>> +
>> +        if (ctx->active == FALSE)
>> +            continue;
>> +
>> +        /* This context is already being messed with */
>> +        if (ctx->pending)
>> +            continue;
>> +
>> +        gc = ctx->context_driver;
>> +
>> +        gc->driver->release_primary(gc, ctx->context.cid);
>
> You better check that this driver function is implemented

Yes

>
>
>> +    }
>> +}
>> +
>>   static void gprs_attached_update(struct ofono_gprs *gprs)
>>   {
>>       DBusConnection *conn = ofono_dbus_get_connection();
>> @@ -1454,30 +1493,23 @@ static void gprs_attached_update(struct 
>> ofono_gprs *gprs)
>>       if (attached == gprs->attached)
>>           return;
>>
>> -    gprs->attached = attached;
>> -
>> -    if (gprs->attached == FALSE) {
>> -        GSList *l;
>> -        struct pri_context *ctx;
>> -
>> -        for (l = gprs->contexts; l; l = l->next) {
>> -            ctx = l->data;
>> -
>> -            if (ctx->active == FALSE)
>> -                continue;
>> -
>> -            pri_reset_context_settings(ctx);
>> -            release_context(ctx);
>> -
>> -            value = FALSE;
>> -            ofono_dbus_signal_property_changed(conn, ctx->path,
>> -                    OFONO_CONNECTION_CONTEXT_INTERFACE,
>> -                    "Active", DBUS_TYPE_BOOLEAN,&value);
>> -        }
>> -
>> +    /*
>> +     * If an active context is found, a PPP session might be still 
>> active
>> +     * at driver level. "Attached" = TRUE property can't be 
>> signalled to
>> +     * the applications registered on GPRS properties.
>> +     * Active contexts have to be release at driver level.
>> +     */
>> +    if (attached == FALSE) {
>> +        release_active_contexts(gprs);
>>           gprs->bearer = -1;
>> +    } else if (check_active_contexts(gprs) == TRUE) {
>> +        gprs->flags |= GPRS_FLAG_ATTACHED_UPDATE;
>> +        release_active_contexts(gprs);
>
> Why do we run the release operation here, wasn't it sufficient to run 
> it once on the Attached goes Detached transition above?

Yes you are right, if context is still active it must be being shut down.

>
>> +        return;
>>       }
>>
>> +    gprs->attached = attached;
>> +
>>       path = __ofono_atom_get_path(gprs->atom);
>>       value = attached;
>>       ofono_dbus_signal_property_changed(conn, path,
>> @@ -2266,6 +2298,16 @@ void ofono_gprs_context_deactivated(struct 
>> ofono_gprs_context *gc,
>>                       OFONO_CONNECTION_CONTEXT_INTERFACE,
>>                       "Active", DBUS_TYPE_BOOLEAN,&value);
>>       }
>> +
>> +    /*
>> +     * If "Attached" property was about to be signalled as TRUE but 
>> there
>> +     * were still active contexts, try again to signal "Attached" 
>> property
>> +     * to registered applications after active contexts have been 
>> released.
>> +     */
>> +    if (gc->gprs->flags&  GPRS_FLAG_ATTACHED_UPDATE) {
>> +        gc->gprs->flags&= ~GPRS_FLAG_ATTACHED_UPDATE;
>> +        gprs_attached_update(gc->gprs);
>> +    }
>>   }
>>
>>   int ofono_gprs_context_driver_register(
>
> Overall I like this approach, but lets get this tested really well.
>
> Regards,
> -Denis
>

Kind regards,
Guillaume