876772 - [B2G][Inari][Cellular Voice and Data] Cannot connect to cell carrier after flashing to today's v1.1 build

Status: VERIFIED FIXED

(Firefox OS Graveyard :: General, defect, P1)

Description

[Nick Davidson]

Description:
After flashing to today's Inari v1.1 Comm RIL or Mozilla RIL build, device is unable to connect to the cellular carrier. The cell bars just keep flashing as if searching for a network. Device got flashed twice with the same results.

Repro Steps:
1) Updated to Inari Build ID: 20130528070209
2) Try to make a call or send SMS


Actual:
User is unable to make/receive calls and send/receive SMS. The cell bars indicator on a top just keep flashing as if searching for a network.

Expected:
User is able to make/receive calls and send/receive SMS

Environmental Variables:
Occurs on the Inari v1.1 com RIL device.
Build ID: 20130528070209
Kernel Date: Feb 21
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/611d727390a4
Gaia: 5f499ab1e4a274d3d28e20139c9cdc546dc4fc69
Version 18.0

Also occurs on the Inari v1.1 Mozilla RIL device.
 Build: 20130528070209
 Gecko: 611d727390a4
 Gaia: 5f499ab1e4a274d3d28e20139c9cdc546dc4fc69
 mozilla inari 1.1
 Version 18.0

Notes:
Repro frequency: 100%,
Resetting device did not fixed the issue.
Q Analysts Test Team Priority: Priority Levels: Pri 1

Comment 1

[Marcia Knous [:marcia]]

I performed an OTA to:

Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/611d727390a4
Gaia   5f499ab1e4a274d3d28e20139c9cdc546dc4fc69
BuildID 20130528070209
Version 18.0

And on my Inari device, SMS and dialer are working fine. This is using Mozilla RIL.

Comment 2

[Marcia Knous [:marcia]]

One other thing of note: I was using a TMobile card to test.

Comment 3

[Nick Davidson]

I flashed Inari v1.1 mizilla to Build ID: 20130527070208
Kernel Date: Feb 21
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/9cac5a5c4efb
Gaia: 4d10e1297b859cacc174c0a54af61a7678d7c32d

and then OTA to Build ID: 20130528070209
Kernel Date: Feb 21
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/611d727390a4
Gaia: 5f499ab1e4a274d3d28e20139c9cdc546dc4fc69

Experience same issue both on 27 and 28th builds
Same issue with at&t as well as with T-Mobile Sim cards

Comment 4

[Nick Davidson]

Attachment: Log OTA-ing Inari from 27 to 28 build

Comment 5

[Nick Davidson]

Attachment: Screenshot - device is searching for network

Comment 6

[Marcia Knous [:marcia]]

Unfortunately I still see this happening using Mozilla Commercial RIL, Inari: 

Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/4785b1353fd7
Gaia   4de4354e3a99f151a834743c7b03a041ac8db12f
BuildID 20130603070207
Version 18.0

5/24 was the last time this worked properly.

Comment 7

[Allen Maxwell]

This issue is breaking or blocking 9 smoketests. 

Requesting escalation to Pri 1

Comment 8

[Allen Maxwell]

Build reference for comment 7:

Build ID: 20130605070207
Gecko: 09dc1ae3b1b5
Gaia: 92a6e36957145cdb2ac8867e5cdba8ecf12308fc

Comment 9

[Lukas Blakk [:lsblakk] use ?needinfo]

This is blocking QA from daily testing.

Comment 10

[Lukas Blakk [:lsblakk] use ?needinfo]

(In reply to [e-mail] [:lsblakk] from comment #9)
> This is blocking QA from daily testing.

And thus we would consider this a blocker

Comment 11

[David Scravaglieri [:scravag]]

Not reproducible on leo device by leo team.

Comment 12

[John Hu [:johnhu][:johu][:醬糊小弟]]

Bug 880615 seems a duplication of this one.

Comment 13

[Tony Chung [:tchung]]

(In reply to David Scravaglieri [:scravag] from comment #11)
> Not reproducible on leo device by leo team.

This was reported against Inari v1.1, which is one of our backup devices we use to test 1.1 if the Leo's are non functional.

Comment 14

[Ken Chang[:kenkai]]

Edgar, please check if this problem is caused by different ril daemon.

Comment 15

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I think this may be related to bug 880331

Comment 16

[Edgar Chen [:edgar]]

(In reply to Ken Chang from comment #14)
> Edgar, please check if this problem is caused by different ril daemon.

Yes, this is caused by rild daemon can not be launched successfully.
If we replace rild with the version before 5/24, device can connect to network.
Currently have no idea why rild fail to be launched after 5/24 build.

Comment 18

[Vincent Liu[:vliu]]

After the build of 5/24, the rild the inari used is generated from a specific revision in manifest. I think the revision might be wrong. mwu, may I have your comment about the revision we used in Inari? Thanks.

Comment 19

[Edgar Chen [:edgar]]

Thanks :vliu provided some clues.

After did some analyze, the root cause is related to device/inari commit 97858b [1] ,rild and libril.so.
In current configuration, inari uses the rild of source code build + libril.so of android backup. With this setup, rild will get link problem and cannot be executed successfully.

I checked the configurations of other device, they seem use the rild of android backup [2]. Maybe Inari should follow the same setup. I am not sure. Can you help on this, :mwu?

Thanks

[1] https://github.com/mozilla-b2g/device-inari/commit/97858bf095e37b78132f0f98a0e1120e95a88ab1
[2] https://github.com/mozilla-b2g/android-device-otoro/blob/master/extract-files.sh#L217

Comment 20

[Michael Wu [:mwu]]

(In reply to Vincent Liu[:vliu] from comment #18)
> After the build of 5/24, the rild the inari used is generated from a
> specific revision in manifest. I think the revision might be wrong. mwu, may
> I have your comment about the revision we used in Inari? Thanks.

This actually suggests that the binaries we're building against are wrong, rather than the revision. We may need to update the inari backup directory.

Comment 21

[Edgar Chen [:edgar]]

As mention in comment #19, this is caused by rild and libril.so link problem. Transfer it to :vliu. Thanks.

Comment 22

[Vincent Liu[:vliu]]

(In reply to Michael Wu [:mwu] from comment #20)
> (In reply to Vincent Liu[:vliu] from comment #18)
> > After the build of 5/24, the rild the inari used is generated from a
> > specific revision in manifest. I think the revision might be wrong. mwu, may
> > I have your comment about the revision we used in Inari? Thanks.
> 
> This actually suggests that the binaries we're building against are wrong,
> rather than the revision. We may need to update the inari backup directory.

Hi mwu,
 
I pulled a request for the fix. Can you please help me to merge it? Thanks.

https://github.com/mozilla-b2g/device-inari/pull/1

Comment 23

[Michael Wu [:mwu]]

Looks good - thanks for catching this. Please uplift to v1-train.

Comment 24

[Vincent Liu[:vliu]]

(In reply to Michael Wu [:mwu] from comment #23)
> Looks good - thanks for catching this. Please uplift to v1-train.

Hi mwu,

The patch in Comment 22 removing the libril.so. Even so, the |Cellular & Data| in settings app still indicate |Unknown SIM Card State|. It seems that the revision we used in manifest file is still not matched. I think we should pulled libril.so and rild from extract file like the unagi did. How do you think?

Comment 26

[Michael Wu [:mwu]]

Go for it.

Comment 27

[Vincent Liu[:vliu]]

(In reply to Michael Wu [:mwu] from comment #26)
> Go for it.

master:
https://github.com/mozilla-b2g/device-inari/pull/2
v1-train
https://github.com/mozilla-b2g/device-inari/pull/3

Comment 28

[Michael Wu [:mwu]]

Merged.

Comment 31

[Marco Chen [:mchen]]

https://github.com/mozilla-b2g/device-inari/commit/fcad682bd2bd372b7fff5751022dff268ab3f5bd

Comment 32

[Jeni]

Verified fixed on 
Inari ID: 20130621070212
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/a34f6d62cb05
Gaia: f88249cc452c39abd0768ec65729939fb561eca2
Platform Version: 18.0

User is now able to make and receive calls and SMS with Inari device