1083454 - Aborting marionette tests is causing bootup failure

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Runtime, defect)

Description

[Anshul]

[Blocking Requested - why for this release]: This is causing a lot of grief as the only way to resolve the boot up issue is to reflash the build again.

STR.
1. Run a bath of marionette scripts one after the other
2. While the test is running, press ctrl + c to abort the test
3. Reboot the device

Observed: The phone get stuck on the firefox logo and I see the following error in the logs

0-15 13:41:45.906 I/Gecko   (  215): *************************
10-15 13:41:45.906 I/Gecko   (  215): A coding exception was thrown and uncaught in a Task.
10-15 13:41:45.906 I/Gecko   (  215): 
10-15 13:41:45.906 I/Gecko   (  215): Full message: TypeError: invalid 'in' operand aManifest
10-15 13:41:45.906 I/Gecko   (  215): Full stack: this.DOMApplicationRegistry.updateDataStore@resource://gre/modules/Webapps.jsm:680:1
10-15 13:41:45.906 I/Gecko   (  215): this.DOMApplicationRegistry.updateDataStoreForApp<@resource://gre/modules/Webapps.jsm:380:1
10-15 13:41:45.906 I/Gecko   (  215): TaskImpl_run@resource://gre/modules/Task.jsm:314:40
10-15 13:41:45.906 I/Gecko   (  215): Handler.prototype.process@resource://gre/modules/Promise.jsm -> resource://gre/modules/Promise-backend.js:865:23
10-15 13:41:45.906 I/Gecko   (  215): this.PromiseWalker.walkerLoop@resource://gre/modules/Promise.jsm -> resource://gre/modules/Promise-backend.js:744:7
10-15 13:41:45.906 I/Gecko   (  215): 
10-15 13:41:45.906 I/Gecko   (  215): *************************
10-15 13:41:45.906 I/Gecko   (  215): *************************
10-15 13:41:45.906 I/Gecko   (  215): A coding exception was thrown and uncaught in a Task.
10-15 13:41:45.906 I/Gecko   (  215): 
10-15 13:41:45.906 I/Gecko   (  215): Full message: TypeError: invalid 'in' operand aManifest
10-15 13:41:45.906 I/Gecko   (  215): Full stack: this.DOMApplicationRegistry.updateDataStore@resource://gre/modules/Webapps.jsm:680:1
10-15 13:41:45.906 I/Gecko   (  215): this.DOMApplicationRegistry.updateDataStoreForApp<@resource://gre/modules/Webapps.jsm:380:1
10-15 13:41:45.906 I/Gecko   (  215): TaskImpl_run@resource://gre/modules/Task.jsm:314:40
10-15 13:41:45.906 I/Gecko   (  215): Handler.prototype.process@resource://gre/modules/Promise.jsm -> resource://gre/modules/Promise-backend.js:865:23
10-15 13:41:45.906 I/Gecko   (  215): this.PromiseWalker.walkerLoop@resource://gre/modules/Promise.jsm -> resource://gre/modules/Promise-backend.js:744:7
10-15 13:41:45.906 I/Gecko   (  215): 
10-15 13:41:45.906 I/Gecko   (  215): *************************
10-15 13:41:45.916 E/GeckoConsole(  215): [JavaScript Error: "invalid 'in' operand aManifest" {file: "resource://gre/modules/Webapps.jsm" line: 680}]

Comment 1

[Anshul]

This issue is reproducible both on 2.1 and Moz central.

Comment 2

[Wesley Huang [:wesley_huang] (EPM) (NI me)]

qawanted to check 2.0 branch and confirm if we can reproduce on Flame.

Comment 3

[howie [:howie]]

Hi Tim, I don't see this is settings issue. Can you confirm and change to the right component? Thanks.

Comment 4

[Andrew Overholt [:overholt]]

2.1 CC blocker = 2.1+

Comment 5

[bhavana bajaj [:bajaj]]

(In reply to Anshul from comment #1)
> This issue is reproducible both on 2.1 and Moz central.

Does 2.0 run as expected?

Comment 6

[Anshul]

Bhavana, yes it does.

Comment 7

[Tim Guan-tin Chien [:timdream] (please needinfo)]

This is either an FxOS: Runtime or DOM: Apps bug, judging by quick reading the log posted.

Comment 8

[Anshul]

This is much more easily reproducible on moz central than it is on 2.1.

Comment 9

[Joshua Mitchell (Inactive)]

clearing QA-Wanted tag (which was addressed in comment 6) and setting appropriate tracking-flags

Comment 10

[Candice Serran (:cserran)]

Hey James...should ownership of this bug fall on your team? If so, can we get assignee please :)...thanks!

Comment 11

[James Lal [:lightsofapollo]]

Maybe but this is really vague right now... The logs point to Webapps.jsm (which has nothing to do with marionette) and "marionette scripts" does not mean anything in particular. 


Can we get an up to date STR with the actual test framework name and some details as to what is actually being done.

Comment 12

[Anshul]

I can also reproduce the issue if I run |flash gaia| command while the phone hasn't finished booting up. The issue is not 100% reproducible and takes a few attempt.

1. Run adb reboot
2. Before the homescreen shows up run |flash gaia|
3. Run adb reboot

Comment 13

[Wesley Huang [:wesley_huang] (EPM) (NI me)]

Hi Fabrice,
Not sure who can look into this. 
Could you forward to needinfo out?

Comment 14

[[:fabrice] Fabrice Desré]

Attachment: restart-block.patch

I was only able to reproduce once. This patch will prevent the error from happening, but I'm not sure this will lead to a sane boot anyway because this condition should not happen.

Anshulj, can you test with this patch?

Comment 15

[Anshul]

Comment on attachment 8510570 [details] [diff] [review]
restart-block.patch

This patch does fix the issue.

Comment 16

[[:fabrice] Fabrice Desré]

https://hg.mozilla.org/integration/b2g-inbound/rev/4bfa433c0437

Comment 17

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/4bfa433c0437

Comment 18

[Ryan VanderMeulen [:RyanVM]]

Please request b2g34 approval on this when you get a chance.

Comment 19

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-b2g34_v2_1/rev/8b1c06f68711

Comment 20

[Yeojin Chung [:YeojinC]]

Unable to verify at this location as this issue involves an automated test.