6:10 a.m.
From: Richard Röjfors <richard(a)puffinpack.se>
There was a race condition where a context might be
registered before the netreg status updates to LTE.
The code took for granted the context is activated after
the technology update. With this change, any order is
is accepted.
---
src/gprs.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/src/gprs.c b/src/gprs.c
index a8702958..a1e52572 100644
--- a/src/gprs.c
+++ b/src/gprs.c
@@ -1721,12 +1721,18 @@ static void gprs_netreg_update(struct ofono_gprs *gprs)
DBG("attach: %u, driver_attached: %u", attach, gprs->driver_attached);
- if (on_lte(gprs))
+ if (on_lte(gprs)) {
/*
* For LTE we set attached status only on successful
* context activation.
+ *
+ * The context could potentially be registered before the netreg
+ * update is received.
*/
- return;
+ gprs_attached_update(gprs);
+
+ return;
+ }
if (gprs->driver_attached == attach)
return;
--
2.20.1
6:14 a.m.
Hope this one applies ;-)
Here is a log showing when this issue happened, where a context was
registered before we got signaled
LTE tech...
Oct 1 09:43:47 info : Aux: > AT+COPS=3,0\r
Oct 1 09:43:47 info : Aux: < \r\nOK\r\n
Oct 1 09:43:47 info : Aux: > AT+COPS?\r
We are running UTRAN + HSPA
Oct 1 09:43:48 info : Aux: < \r\n+COPS: 0,0,"XXX",6\r\n
Oct 1 09:43:48 info : Aux: < \r\nOK\r\n
Oct 1 09:43:48 debug :
../git/drivers/atmodem/network-registration.c:cops_cb() cops_cb: XXX, 240
07 6
Oct 1 09:43:48 debug : ../git/src/network.c:current_operator_callback()
0x10d1218, 0x10d6fb8
Oct 1 09:43:50 info : Aux: < \r\n+CIEV: 2,4\r\n
We are transitioning to E-UTRAN because context 4 gets activated
Oct 1 09:43:51 info : Aux: < \r\n+CGEV: ME PDN ACT 4\r\n
Oct 1 09:43:51 info : Aux: > AT+CGDCONT?\r
Oct 1 09:43:51 info : Aux: < \r\n+CGDCONT:
1,"IP","XXX","",0,0,0,0,0,0\r\n
Oct 1 09:43:51 info : Aux: < +CGDCONT:
4,"IP","XXX.mnc003.mcc248.gprs","100.98.237.128",0,0,0,0,0,0\r\n
Oct 1 09:43:51 info : Aux: < \r\nOK\r\n
gprs status goes to unknown(normal on ublox when on E-UTRAN) which will
trigger a "detach"
Oct 1 09:43:51 info : Aux: < \r\n+CGREG: 4\r\n
Oct 1 09:43:51 info : Aux: < \r\n+UREG: 0\r\n
Oct 1 09:43:51 debug : ../git/drivers/atmodem/gprs.c:at_cgdcont_read_cb()
ok 1
Oct 1 09:43:51 debug : ../git/src/gprs.c:ofono_gprs_cid_activated() cid 4
Oct 1 09:43:51 debug :
../git/drivers/ubloxmodem/gprs-context.c:ublox_gprs_read_settings() cid 4
Oct 1 09:43:51 debug : ../git/src/modem.c:get_modem_property() modem
0x10ccaa0 property NetworkInterface
Oct 1 09:43:51 debug : ../git/src/gprs.c:pri_read_settings_callback()
0x10d6868
Oct 1 09:43:51 debug : ../git/src/gprs.c:ofono_gprs_status_notify()
/ublox_0 status unknown (4)
Oct 1 09:43:51 debug : ../git/src/gprs.c:ofono_gprs_detached_notify()
/ublox_0
Oct 1 09:43:51 info : Aux: > AT+CREG?\r
Oct 1 09:43:51 info : Aux: < \r\n+CIEV: 9,1\r\n\r\n+UREG: 7\r\n\r\n+CIEV:
9,2\r\n\r\n+CTZE: +08,1,"19/10/01,11:43:53"\r\n
Oct 1 09:43:51 debug :
../git/drivers/ubloxmodem/network-registration.c:ctze_notify() tz +08 dst 1
time 19/10/01,11:43:53
Oct 1 09:43:51 debug : ../git/src/network.c:ofono_netreg_time_notify() net
time 2019-10-01 11:43:53 utcoff 7200 dst 1
We are on E-UTRAN
Oct 1 09:43:51 info : Aux: < \r\n+CREG:
2,5,"2B02","01B57901",7\r\n
Oct 1 09:43:51 info : Aux: < \r\nOK\r\n
Oct 1 09:43:51 debug : ../git/src/network.c:ofono_netreg_status_notify()
/ublox_0 status 5 tech 7 lac 11010 ci 28670209
Oct 1 09:43:51 debug : ../git/src/gprs.c:netreg_status_changed() 5
(roaming)
gprs.c gets signaled about the change to E-UTRAN.
But since we now are on LTE, the code currently expect expect to change to
attached state first
when a new context is registered...But that happened above, so we will wait
for ever...
Oct 1 09:43:51 debug : ../git/src/gprs.c:gprs_netreg_update() attach: 1,
driver_attached: 0
Oct 1 09:43:51 info : Aux: > AT+COPS=3,2\r
Oct 1 09:43:51 info : Aux: < \r\nOK\r\n
Oct 1 09:43:51 info : Aux: > AT+COPS?\r
Den tis 1 okt. 2019 kl 22:11 skrev <richard.rojfors(a)gmail.com>:
From: Richard Röjfors <richard(a)puffinpack.se>
There was a race condition where a context might be
registered before the netreg status updates to LTE.
The code took for granted the context is activated after
the technology update. With this change, any order is
is accepted.
---
src/gprs.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/src/gprs.c b/src/gprs.c
index a8702958..a1e52572 100644
--- a/src/gprs.c
+++ b/src/gprs.c
@@ -1721,12 +1721,18 @@ static void gprs_netreg_update(struct ofono_gprs
*gprs)
DBG("attach: %u, driver_attached: %u", attach,
gprs->driver_attached);
- if (on_lte(gprs))
+ if (on_lte(gprs)) {
/*
* For LTE we set attached status only on successful
* context activation.
+ *
+ * The context could potentially be registered before the
netreg
+ * update is received.
*/
- return;
+ gprs_attached_update(gprs);
+
+ return;
+ }
if (gprs->driver_attached == attach)
return;
--
2.20.1
4:31 a.m.
Hi Richard,
On 10/1/19 3:10 PM, richard.rojfors(a)gmail.com wrote:
From: Richard Röjfors <richard(a)puffinpack.se>
There was a race condition where a context might be
registered before the netreg status updates to LTE.
The code took for granted the context is activated after
the technology update. With this change, any order is
is accepted.
---
src/gprs.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
Applied, thanks.
Regards,
-Denis
1073
days inactive
1083
days old
2 comments
3 participants
participants (3)
-
Denis Kenzior -
Richard Röjfors -
richard.rojfors＠gmail.com