937773 - Call screen displayed when no calls ongoing

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Vendcom, defect)

Description

[Michael Schwartz [:m4]]

Steps:
# Establish a multiparty call to 3 numbers
# Press the End key to send all of these at once
Expected: all calls are ended and the call screen exits
Observed: all calls are ended but the call screen remains

Comment 1

[Joe Cheng [:jcheng] (please needinfo)]

triage: koi+ for CS blocker and also seem like a regression. etienne, can you take a look? Thanks

Comment 2

[Etienne Segonzac (:etienne)]

I wasn't able to reproduce the bug.

- made an outgoing call
- made a second outgoing call
- merged the 2 calls
- got an incoming cal
- merged the new incoming call with the 2 others
- pressed "end"

The call screen closed and all the calls were ended.

Comment 3

[Jason Smith [:jsmith]]

QA Wanted to see if we can reproduce this.

Comment 4

[Sarah Parsons]

I was able to reproduce this issue on the Buri 1.2 Build ID: 20131114004004

Gaia   3e67ddf7d2680d1ef01441641f4ca2e0e4006887
SourceStamp 948e30162eac
BuildID 20131114004004
Version 26.0

Comment 5

[bhavana bajaj [:bajaj]]

:etienne, not sure if you had tried the same device/ build version as QA and hope the recent QA information in comment#4 helps here. 

Passing this onto you as assignee, please feel free to clear if you are not the right person to look at this one. Thanks !

Comment 6

[Etienne Segonzac (:etienne)]

I was able to get a buri in the same state and partially reproduce the bug.
It looks like a vendor RIL issue.

Here's my STR:
- call my voicemail
- get an incoming call
- accept the extra call
- merge it in a conference call
- tap the "add call" button
- make another call
- merge it in the conference call
- tap the end button (1)
- tap the end button (2)
- tap the end button (3)

I have to press the end button 3 times, each time only one of the calls would end.
*Note that once all the calls ended the call screen closed as expected* which differs from the Description.

I logged all the telephony.calls each time I got an oncallschanged, and all the telephony.conferenceGroup.calls on conferenceGroup.oncallschanged here's what I got:

=== First call
+++ call 0 - 888 - dialing
+++ call 0 - 888 - dialing
+++ call 0 - 888 - connected

=== Second call
+++ call 1 - +33184XXXXXX - incoming
+++ conf call 0 - 888 - connected       ??? why do we get a conferenceGroup.oncallschanged here?
+++ call 0 - +33184XXXXXX - connected
+++ call 0 - +33184XXXXXX - connected

=== Merging the first 2 calls in a conference call
+++ conf call 0 - 888 - connected
+++ conf call 1 - +33184XXXXXX - connected

=== Third call
+++ call 0 - 0674XXXXXX - dialing
+++ call 0 - 0674XXXXXX - connected

=== Merging the all 3 calls in the conference
+++ conf call 0 - 888 - connected
+++ conf call 1 - +33184XXXXXX - connected
+++ conf call 2 - 0674XXXXXX - connected

=== Ending all 3 calls, only 1 get's remove from the group, no callschanged :/
+++ conf call 0 - +33184XXXXXX - connected
+++ conf call 1 - 0674XXXXXX - connected

=== Ending all 2 calls, only 1 get's remove from the group, no callschanged :/
+++ conf call 0 - +33184XXXXXX - disconnecting

=== Ending the last call
+++ call 0 - 0674XXXXXX - connected
+++ call 0 - 0674XXXXXX - connected

Looks like we don't get all the oncallschanged, and that the vendor RIL doesn't like us calling hangUp() on all the calls at the same time.
The interesting part is that only the the last call to hangUp() in the tick got executed (we start at the end on the array [1]).

Sorry I wasn't able to get RIL debugs but I can only safely flash gaia on top of those teleweb builds.

[1] https://github.com/mozilla-b2g/gaia/blob/master/apps/communications/dialer/js/calls_handler.js#L631-637

Comment 7

[Jason Smith [:jsmith]]

mvines - Sounds like this is a bug on your end. Can someone on your team look into this?

Comment 8

[Anshul]

(In reply to Etienne Segonzac (:etienne) from comment #6)
> I was able to get a buri in the same state and partially reproduce the bug.
> It looks like a vendor RIL issue.
> 
> Here's my STR:
> - call my voicemail
> - get an incoming call
> - accept the extra call
> - merge it in a conference call
> - tap the "add call" button
> - make another call
> - merge it in the conference call
> - tap the end button (1)
> - tap the end button (2)
> - tap the end button (3)
> 
> I have to press the end button 3 times, each time only one of the calls
> would end.
> *Note that once all the calls ended the call screen closed as expected*
> which differs from the Description.
> 
> I logged all the telephony.calls each time I got an oncallschanged, and all
> the telephony.conferenceGroup.calls on conferenceGroup.oncallschanged here's
> what I got:
> 
> === First call
> +++ call 0 - 888 - dialing
> +++ call 0 - 888 - dialing
> +++ call 0 - 888 - connected
> 
> === Second call
> +++ call 1 - +33184XXXXXX - incoming
> +++ conf call 0 - 888 - connected       ??? why do we get a
> conferenceGroup.oncallschanged here?
> +++ call 0 - +33184XXXXXX - connected
> +++ call 0 - +33184XXXXXX - connected
> 
> === Merging the first 2 calls in a conference call
> +++ conf call 0 - 888 - connected
> +++ conf call 1 - +33184XXXXXX - connected
> 
> === Third call
> +++ call 0 - 0674XXXXXX - dialing
> +++ call 0 - 0674XXXXXX - connected
> 
> === Merging the all 3 calls in the conference
> +++ conf call 0 - 888 - connected
> +++ conf call 1 - +33184XXXXXX - connected
> +++ conf call 2 - 0674XXXXXX - connected
> 
> === Ending all 3 calls, only 1 get's remove from the group, no callschanged
> :/
> +++ conf call 0 - +33184XXXXXX - connected
> +++ conf call 1 - 0674XXXXXX - connected
> 
> === Ending all 2 calls, only 1 get's remove from the group, no callschanged
> :/
> +++ conf call 0 - +33184XXXXXX - disconnecting
> 
> === Ending the last call
> +++ call 0 - 0674XXXXXX - connected
> +++ call 0 - 0674XXXXXX - connected
> 
> Looks like we don't get all the oncallschanged, and that the vendor RIL
> doesn't like us calling hangUp() on all the calls at the same time.
> The interesting part is that only the the last call to hangUp() in the tick
> got executed (we start at the end on the array [1]).
> 
> Sorry I wasn't able to get RIL debugs but I can only safely flash gaia on
> top of those teleweb builds.
> 
> [1]
> https://github.com/mozilla-b2g/gaia/blob/master/apps/communications/dialer/
> js/calls_handler.js#L631-637
Etienne, the issue you mentioned is already resolved in the recent AUs. The build you are using is probably old. Could you please share a patch with debug messages so we can run the test and share the logs with you?

Comment 9

[Etienne Segonzac (:etienne)]

(In reply to Anshul from comment #8)
> Etienne, the issue you mentioned is already resolved in the recent AUs. The
> build you are using is probably old. Could you please share a patch with
> debug messages so we can run the test and share the logs with you?

Don't have it handy, but can you reproduce the bug in question here on the new build?

Comment 10

[Preeti Raghunath(:Preeti)]

Moving to koi? till reproducibility and regression are confirmed.

Comment 11

[Michael Schwartz [:m4]]

I'm still seeing the issue on koi.  Here are my sha1's:

Gaia: a6484b1e6fc07cf6bd8d6fcf9aeebb14b7e8869d
Gecko: 0112f6b636c5042baa3fd43b9eaf3bc049d18481

Comment 12

[Jason Smith [:jsmith]]

Moving back to dialer based on the above discussion implying this is a Gaia bug now.

Comment 13

[Etienne Segonzac (:etienne)]

(In reply to Jason Smith [:jsmith] from comment #12)
> Moving back to dialer based on the above discussion implying this is a Gaia
> bug now.

I don't see the part implying that it's gaia... just that it is reproducible.

Comment 14

[Michael Schwartz [:m4]]

(In reply to Etienne Segonzac (:etienne) from comment #13)
> (In reply to Jason Smith [:jsmith] from comment #12)
> > Moving back to dialer based on the above discussion implying this is a Gaia
> > bug now.
> 
> I don't see the part implying that it's gaia... just that it is reproducible.

I checked that commercial ril is calling the TelephonyProvider callbacks correctly.  I can triage this further but I thought the Moz folks might be more effective at that.

Comment 15

[Joe Cheng [:jcheng] (please needinfo)]

Hsinyi, is it something that on the Moz RIL that you can look into? thanks

Comment 16

[Hsin-Yi Tsai (she/her) [:hsinyi]]

(In reply to Joe Cheng [:jcheng] from comment #15)
> Hsinyi, is it something that on the Moz RIL that you can look into? thanks

Sure, I'll take a look at it to see if anything I could help!

Comment 17

[bhavana bajaj [:bajaj]]

(In reply to Hsin-Yi Tsai  [:hsinyi] from comment #16)
> (In reply to Joe Cheng [:jcheng] from comment #15)
> > Hsinyi, is it something that on the Moz RIL that you can look into? thanks
> 
> Sure, I'll take a look at it to see if anything I could help!

passing this to in that case, thank s!

Comment 18

[Szu-Yu Chen [:aknow] (inactive)]

I cannot reproduce it on unagi device with moz ril.

My steps
- make an outgoing call
- make an incoming call, accept
- merge them into a conference call
- tap the "add call" button
- make another outgoing call
- merge it into the conference call
- tap the end button
- all calls are disconnected and call screen closed

Comment 19

[Michael Vines [:m1] [:evilmachines]]

Can you please share a radio and main log, and confirm the Gecko/Gaia SHA1s are from v1.2.  Thanks

Comment 20

[Szu-Yu Chen [:aknow] (inactive)]

Gaia
commit 6e66cb790d5b931d7cfb3256b8cb90a4471c1a3c                     
Merge: 5a0802f 978b57e                                              
Author: Timothy Guan-tin Chien <[e-mail]>                 
Date:   Sun Nov 24 17:57:34 2013 -0800                              
                                                                    
    Merge pull request #13968 from timdream/keyboard-pref-off2-v1.2 

Gecko
commit 5fec3090bdb3c2a4e85de90d9b1c1628f3d1333f              
Author: Gaia Pushbot <[e-mail]>          
Date:   Sun Nov 24 18:10:55 2013 -0800                       
                                                             
    Bumping gaia.json for 3 gaia-1_2 revision(s) a=gaia-bump

Comment 21

[Szu-Yu Chen [:aknow] (inactive)]

Attachment: main.log

Comment 22

[Szu-Yu Chen [:aknow] (inactive)]

Attachment: radio.log

Comment 23

[Preeti Raghunath(:Preeti)]

QA,

Can we get a regression window here?

Comment 24

[Jason Smith [:jsmith]]

(In reply to Preeti Raghunath(:Preeti) from comment #23)
> QA,
> 
> Can we get a regression window here?

I think you are meaning to ask if this reproduces on 1.1 or not. We can work on a window after that.

Comment 25

[Sarah Parsons]

This issue does not repro on the Leo 1.1 Build ID: 20131202041200

Gaia   19c9ff3a46a4389e40253c97b359763243af4531
SourceStamp 617eb9d9bcc2
BuildID 20131202041200
Version 18.0

-1.2 Regression Window- 

This issue started to occur on the Buri 1.2 Build ID: 20130905185329

Gaia   b77dc7d399c14ac2bafe5c89566b292bc6772d32
SourceStamp df8f342e9a6b
BuildID 20130905185329
Version 26.0a1

Last working Buri 1.2 Build ID: 20130904040205

Gaia   b6c5bf1d24230bfed4a8f680e625fa2175001f82
SourceStamp 7ff96bd19c1c
BuildID 20130904040205
Version 26.0a1

Comment 26

[Preeti Raghunath(:Preeti)]

Sarah,

Please provide info if this test was done on com RIl or moz RIL?

Comment 27

[Sarah Parsons]

This issue was on both Com and Moz RIL.

Comment 28

[Preeti Raghunath(:Preeti)]

Ken

Please review.

Comment 29

[Jason Smith [:jsmith]]

Moving back to koi+ per confirmation that this is a regression reproducing on COM & MOZ RIL.

Comment 30

[Ken Chang[:kenkai]]

Aknow, can you please use the build id provided in comment 25 and Buri to reproduce this bug?

Comment 31

[Szu-Yu Chen [:aknow] (inactive)]

OK. I will try on buri.

Comment 32

[Szu-Yu Chen [:aknow] (inactive)]

Sarah,

I could not reproduce the issue with buri v.12

gaia  b77dc7d399c14ac2bafe5c89566b292bc6772d32  (as you mentioned, issue started to occur from this commit)
gecko 492afacb36aef5515d1eb3d2e3c4fc3c298560f5

steps
1. make outgoing call a
2. make outgoing call b
3. merge a and b
4. make outgoing call c
5. merge c into the group
6. press end key

All calls are disconnected. Call screen is disappeared. So I see the dialling keypad now.
Could you give me the details about the issue, including your reproduce steps and the difference between the good one and the bad one. There might be something I missed.

Comment 33

[Sarah Parsons]

I was able to repro this issue on the Buri 1.2 Build ID: 20131205004003

Gaia   0659f16b9790b1cf9eba4d80743fcc774d2ffe3a
SourceStamp af2c7ebb5967
BuildID 20131205004003
Version 26.0

STR:

1. From phone A, Call phone B.
2. When call is connected, add call phone C. 
3. When phone A is connected to both  B and C, try to disconnect call.

Acutal: Phone A hangs on callscreen, pressing disconnect does nothing. The other phones hang up, but the buri will show the call screen as still connected.

Comment 35

[Kai-Chih Hu [:khu]]

Aknow, are you the best one to own this bug?

Comment 36

[Kai-Chih Hu [:khu]]

Aknow, if you're not the best one, please talk with Ken and see who can help. Thanks.

Comment 37

[Szu-Yu Chen [:aknow] (inactive)]

Sarah,

I really can't reproduce it here. Could you provide the log of the issue with ril debugging info. Please follow the instruction here to turn ril debugging on.
https://wiki.mozilla.org/B2G/QA/Tips_And_Tricks#Changing_preferences

Then record the main log and radio log at the same time and attach the results here.
1. adb logcat -v time 
2. adb logcat -b radio -v time

Kevin,
OK. Currently, we don't know where the problem is, but I could be the one to analyze the problem first.

Comment 38

[Sarah Parsons]

Attachment: bug937773_logcat

Comment 39

[Sarah Parsons]

Attachment: bug937773_radiolog

Comment 40

[Sarah Parsons]

I can still repro on today's buri 1.2 Build ID: 20131205004003

Gaia   0659f16b9790b1cf9eba4d80743fcc774d2ffe3a
SourceStamp af2c7ebb5967
BuildID 20131205004003
Version 26.0

Please see attached main logcat and radio log.

Comment 41

[Sarah Parsons]

This issue seems to only occur on the Moz RIL on 1.2. I tried to repro on Com, but couldnt get the issue to occur.

Comment 42

[Jason Smith [:jsmith]]

This sounds like the same problem seen in https://bugzilla.mozilla.org/show_bug.cgi?id=943117.

Comment 43

[Szu-Yu Chen [:aknow] (inactive)]

Sarah,
Sorry, I can't find the related log in the attach file. Thus, I have sent you a email to confirm some testing details.

Comment 44

[Sarah Parsons]

Hi Aknow,

Thank you for your helpful email. I went ahead and replied. I will also attach the correct log here. 

Regards,

Sarah

(In reply to Szu-Yu Chen [:aknow] from comment #43)
> Sarah,
> Sorry, I can't find the related log in the attach file. Thus, I have sent
> you a email to confirm some testing details.

Comment 45

[Sarah Parsons]

Attachment: 937773_12_09_log

Comment 46

[Jason Smith [:jsmith]]

QA wanted - can this be retested with the latest buri base image with gaia/gecko on top?

Comment 47

[Szu-Yu Chen [:aknow] (inactive)]

Attachment: main.log #2

Log provided by Sarah.

Comment 48

[Szu-Yu Chen [:aknow] (inactive)]

Hi Sarah,

Please update your buri with new base image.
See https://bugzilla.mozilla.org/show_bug.cgi?id=943117#c20

Then help retest the issue to check whether it still has the problem.
Thank you.

Comment 49

[Sarah Parsons]

I can no longer repro this issue on Buri 1.2 Build ID: 20131212004004

BuildID: 20131212004004
Gaia: 6d02039072a2ae5cf9225a6f4c78ed49decfab5c
Gecko: 8bae10bb0aed
Version: 26.0
Base: V1.2_20131209

Also, I can no longer repro this issue on Buri 1.3 Build ID: 20131212004003

BuildID: 20131212004003
Gaia: 588a3e02c4ace3b3341ba1f6bb7274120b53b2b3
Gecko: 031270be3702
Version: 28.0a2
Base: V1.2_20131209

There is however a separate issue I ran into on 1.3 when adding another call using the + button, the options to use the Dialer pad or Call Log disappear, only the address book pops up.

Comment 50

[Jason Smith [:jsmith]]

Can you file a separate bug for the issue you saw in comment 49?

Closing as fixed as the new base image resolved this problem.

Comment 51

[Sarah Parsons]

See bug 949201


(In reply to Jason Smith [:jsmith] from comment #50)
> Can you file a separate bug for the issue you saw in comment 49?
> 
> Closing as fixed as the new base image resolved this problem.