Hi Richard,
> + gprs->netreg_status = -1;
> gprs->used_pids = l_uintset_new(MAX_CONTEXTS);
>
Wouldn't changes to gprs.c be sufficient, without touching netreg.c?
No unfortunately not. gprs reads the netreg status when it attaches to
the atom. Which would be unknown if we do not touch netreg.c.
.. and unknown is a valid status which can not be distinguished from
"not yet updated".
Hmm, okay. But then, can't you set gprs->netreg_status to -1 inside
netreg_watch and just wait until netreg_status_changed to get it set
properly?
That seems to be your real culprit, the gprs->netreg_status assignment
from ofono_netreg_get_status()
> return gprs;
> diff --git a/src/network.c b/src/network.c
> index 2824eae6..1db01bcf 100644
> --- a/src/network.c
> +++ b/src/network.c
> @@ -1874,7 +1874,7 @@ struct ofono_netreg
*ofono_netreg_create(struct ofono_modem *modem,
> if (netreg == NULL)
> return NULL;
>
> - netreg->status = NETWORK_REGISTRATION_STATUS_UNKNOWN;
> + netreg->status = -1;
Okay, but a minor problem with this is that there now exists a brief
moment in type where the Status property has a weird value...
I looked at this, and most statuses, even some defined results as
unknown. Any unhandled values as negative would do that too.
If you are refering to the D-Bus property?
Yes. Most of the time -1 simply means the property is omitted. In this
case that is not actually what happens. Refer to src/common.c
registration_status_to_string()
Regards,
-Denis
_______________________________________________
ofono mailing list
[email protected]
https://lists.ofono.org/mailman/listinfo/ofono
