If I restart my device a few times in a row I find that far less than 100% of the time the RIL will fail to connect to the carrier and give a cell connection. Sometimes it just shows up with nothing and "Emergency calls only" and sometimes it comes up with "Scanning" and the antenna signal animated. The worst part is that in either of those states it just never connects regardless of how long you leave it idle. This is new in today's nightly build. STR: 1. Start device 2. Observe RIL state 3. Wait 60 seconds 4. Hold down power, restart device 5. Rinse, repeat We have replicated this on: Unagi 3 UK SIM Unagi O2 UK SIM Unagi Orange Romania SIM Several of our automation devices Build: Gecko http://hg.mozilla.org/mozilla-central/rev/77ed46bf4c1a Gaia ec885264d885d877d68980c0227058231f18ad09 BuildID 20130905040201 Version 26.0a1 Device Unagi

Seems similar to bug 905630.

Can you replicate it as I described Thomas? It has gotten a lot worse in today's build. You don't need to crash the device to replicate it.

PS yes I agree it sounds similar!

(In reply to Zac C (:zac) from comment #2) > Can you replicate it as I described Thomas? No. I tried several times in a row, but everything works as expected. > It has gotten a lot worse in today's build. You don't need to crash the > device to replicate it. I noticed that this sometimes happens after a restart, sometimes after the phone was suspended for a while. And sometimes everything works fine. I've once checked the logcat on my device and it seemed that connecting works at first, but later we get disconnected. I have a T-Mobile SIM, so it's probably unrelated to the network. Maybe it's a driver issue. Did you ever observe this bug on a phone other than the Unagi?

(In reply to Thomas Zimmermann [:tzimmermann] [:tdz] from comment #4) > I noticed that this sometimes happens after a restart Well that's exactly what this bug is describing, so you can replicate it? I haven't tried any other devices, I don't have access to the main dev devices right now.

(In reply to Zac C (:zac) from comment #5) > (In reply to Thomas Zimmermann [:tzimmermann] [:tdz] from comment #4) > > > I noticed that this sometimes happens after a restart > > Well that's exactly what this bug is describing, so you can replicate it? I can replicate it in the sense of 'it happens from time to time', not by following the STR from the description.

That still seems too frequently and it busts a lot of our test automation that uses the RIL to make calls and use cell data. So if you repeat the STR in comment #0 it will fail maybe 3-4 times.

(In reply to Zac C (:zac) from comment #7) > That still seems too frequently and it busts a lot of our test automation > that uses the RIL to make calls and use cell data. > > So if you repeat the STR in comment #0 it will fail maybe 3-4 times. Hi Zac, The log you attached doesn't contain ril debug messages. Would you please attach a log with ril debug messages? Thanks.

In this scenario I hit the "Emergency calls only" state

Here is a log with a different SIM, this time a 3 UK Sim.

We've had this problem in the past but it 'went away' for a while. It has now come back progressively worse in the last fortnight or so. Here is our previous attempt to debug it: https://bugzilla.mozilla.org/show_bug.cgi?id=853489

I cannot reproduce it with gecko m-c 146137:218d4334d29e on unagi. But I saw error messages 'iccInfo is null' in attachment 800642 [details] as well as when zac was showing me the gaia-ui-test failures. I am thinking that might be caused by bug 869778 since it changes the default/initial value of iccInfo, and gaia needs corresponding changes as well. I am going to roll back to the version as comment #0 and apply my gaia patch to see if it helps.

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #13) > I cannot reproduce it with gecko m-c 146137:218d4334d29e on unagi. > But I saw error messages 'iccInfo is null' in attachment 800642 [details] as > well as when zac was showing me the gaia-ui-test failures. I am thinking > that might be caused by bug 869778 since it changes the default/initial > value of iccInfo, and gaia needs corresponding changes as well. > > I am going to roll back to the version as comment #0 and apply my gaia patch > to see if it helps. I've filed bug 914673 to address the error |iccInfo is null|. But since I still cannot reproduce this even I roll back, I am not sure if that gaia patch helps fix this issue.

Not having this problem anymore.