Hi Felix,
it seems like that this autoconnect feature is causing some regressions and a
headaches to few people :-) I'm talking about this commit in particular:
Author: Felix Fietkau <n...@nbd.name>
Date: Thu Sep 22 20:07:45 2016 +0200
uqmi: re-enable autoconnect which was dropped without explanation
Fixes a regression in commit 8f24ee638275:
"uqmi: Add proper IPv6 support"
I'm using Quectel EC20 miniPCIe modem mainly in Slovakia and Czech Republic.
Till now, without autoconnect enabled it was working fine in both networks, but
now after this change has been pulled into my device I'm no more able to
initiate the connection in Slovakia:
netifd: wan (1681): Waiting for network registration
netifd: wan (1681): Starting network internet
netifd: wan (1681): Unable to connect IPv4
netifd: wan (1681): Unable to connect
To get it working again, I've to just remove autoconnect from start-network
command:
--- a/package/network/utils/uqmi/files/lib/netifd/proto/qmi.sh
+++ b/package/network/utils/uqmi/files/lib/netifd/proto/qmi.sh
@@ -112,8 +112,7 @@ proto_qmi_setup() {
${auth:+--auth-type $auth} \
${username:+--username $username} \
${password:+--password $password} \
- --ip-family ipv4 \
- --autoconnect`
+ --ip-family ipv4`
To me it seems like this autoconnect option is being propagated throught the
network and the network operator could refuse connection with this option. I
don't have any other logical explanation.
How to fix it for both use cases, with and without autoconnect feature?
Introduce qmi_autoconnect config option?
Someone likely affected also[1].
Thanks.
1. https://bugs.lede-project.org/index.php?do=details&task_id=148
-- ynezz
_______________________________________________
Lede-dev mailing list
Lede-dev@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/lede-dev
