Closed Bug 1207048 Opened 9 years ago Closed 9 years ago

[Browser] A video can't be played, even it has been loaded completely.

Categories

(Firefox OS Graveyard :: Gaia::Browser, defect)

ARM
Gonk (Firefox OS)
defect
Not set
normal

Tracking

(b2g-v2.2 unaffected, b2g-master affected)

RESOLVED WORKSFORME
Tracking Status
b2g-v2.2 --- unaffected
b2g-master --- affected

People

(Reporter: yelanying, Unassigned)

Details

(Keywords: regression, smoketest, Whiteboard: [dogfood-blocker])

Attachments

(6 files)

Attached file logcat_2353.txt
[1.Description]:
[Aries KK v2.5][Flame KK v2.5][Browser] Launch browser app, and navigate to "www.youtube.com". Select a video to play, but it can't be played, it has been loaded completely.
Time: 23:53
See attachments: logcat_2353.txt and Aries_KK v2.5.3gp

[2.Testing Steps]: 
1.Launch browser app.
2.Input "www.youtube.com" at search bar.
3.Select a video to play. 

[3.Expected Result]: 
3.The video could be played.

[4.Actual Result]: 
3.The video can't  be played, even it has been loaded completely.

[5.Reproduction build]: 
Flame KK v2.2 (Unaffected)
Build ID               20150921032501
Gaia Revision          95950c9d48ad2fc2da7686c2b133b750a99cd3da
Gaia Date              2015-09-18 09:55:13
Gecko Revision         https://hg.mozilla.org/releases/mozilla-b2g37_v2_2/rev/9644c82a5b88
Gecko Version          37.0
Device Name            flame
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.cltbld.20150921.070345
Firmware Date          Mon Sep 21 07:03:54 EDT 2015
Base Image         v18D v4
Bootloader             L1TC000118D0

Flame KK v2.5 eng(Affected)
Build ID               20150921073455
Gaia Revision          2d370fa35c1a0ee2a637e3772c0843586a5f96c9
Gaia Date              2015-09-21 02:41:31
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/039a8490891595736b16a3ccb17f025f4dcf13eb
Gecko Version          44.0a1
Device Name            flame
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.cltbld.20150921.112037
Firmware Date          Mon Sep 21 11:20:52 EDT 2015
Base Image         v18D v4
Bootloader             L1TC000118D0

Flame KK v2.5  user (Affected)
Build ID               20150921073455
Gaia Revision          2d370fa35c1a0ee2a637e3772c0843586a5f96c9
Gaia Date              2015-09-21 02:41:31
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/039a8490891595736b16a3ccb17f025f4dcf13eb
Gecko Version          44.0a1
Device Name            flame
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.cltbld.20150921.112039
Firmware Date          Mon Sep 21 11:20:55 EDT 2015
Base Image         v18D v4
Bootloader             L1TC000118D0

Aries KK v2.5 eng (Affected)
Build ID               20150921174442
Gaia Revision          29991414eb94b6baa1ec2e63fdb4f6dfae05fb01
Gaia Date              2015-09-21 09:27:10
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/197af2fb7e29ff8e4b3b6ced723b6172e954e17d
Gecko Version          44.0a1
Device Name            aries
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.worker.20150921.170449
Firmware Date          Mon Sep 21 17:04:57 UTC 2015
Bootloader             s1

Aries KK v2.5 user (Affected)
Build ID               20150921174442
Gaia Revision          29991414eb94b6baa1ec2e63fdb4f6dfae05fb01
Gaia Date              2015-09-21 09:27:10
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/197af2fb7e29ff8e4b3b6ced723b6172e954e17d
Gecko Version          44.0a1
Device Name            aries
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.worker.20150921.170449
Firmware Date          Mon Sep 21 17:04:57 UTC 2015
Bootloader             s1


[6.Reproduction Frequency]: 
occasionally Recurrence, 6/10

[7.TCID]: 
6073

[8.Note]:
On Android and PC, the same video can be played successfully according to the STR.
Attached video Aries_KK v2.5.3gp
Keywords: smoketest
QA Contact: jmercado
Whiteboard: [dogfood-blocker]
I've spent about 40 min to investigate this issue. I was able to reproduced that a couple times on Aries 2.5 and Flame 2.5 but repro is still very intermittent. I didn't do anything special just run multiple video clips, launched many different apps and switched between them, closed browser and opened it again etc.  Reboot a phone or restart a browser fix the issue.  

Device: Aries Master
Build ID: 20150922115027
Gaia: 68361828ae88dffd04b250121b5f2472a63f4bf0
Gecko: a1ccea59e254a88f7bb44b0ad8a58b77b7eca339
Gonk: 2916e2368074b5383c80bf5a0fba3fc83ba310bd
Version: 44.0a1 (Master)
Firmware Version: D5803_23.1.A.1.28_NCB.ftf
User Agent: Mozilla/5.0 (Mobile; rv:44.0) Gecko/44.0 Firefox/44.0
________________________________________________________________

Device: Flame Master
Build ID: 20150922030227
Gaia: 29991414eb94b6baa1ec2e63fdb4f6dfae05fb01
Gecko: 2235e56c94cf61614902fd3a4ac7b837f7154b97
Gonk: c4779d6da0f85894b1f78f0351b43f2949e8decd
Version: 44.0a1 (Master)
Firmware Version: v18D
User Agent: Mozilla/5.0 (Mobile; rv:44.0) Gecko/44.0 Firefox/44.0
Attached file logs_aries.txt
Attached file logs_flame.txt
I am getting this to reproduce 10/10 times on Aries 2.5

The video will not play. It will either be a black screen or endless an endless loading circle.

Device: Aries 2.5
BuildID: 20150922115027
Gaia: 68361828ae88dffd04b250121b5f2472a63f4bf0
Gecko: a1ccea59e254a88f7bb44b0ad8a58b77b7eca339
Gonk: 2916e2368074b5383c80bf5a0fba3fc83ba310bd
Version: 44.0a1 (2.5) 
Firmware Version: D5803_23.1.A.1.28_NCB.ftf
User Agent: Mozilla/5.0 (Mobile; rv:44.0) Gecko/44.0 Firefox/44.0

STR:

Prerequisite: WiFi and data enabled 

1. Restart the device before attempting to reproduce this issue.
2. After the OS loads, open the browser app first and go to www.youtube.com
3. Tap on a video to play it and observe.

Actual result: The video does not start. There will be a black screen or an endless loading circle.

Expected result: The video plays without issue. 

Workaround: After encountering the issue hold down the home button to go into card view and then tap on the browser card to go back into browser. The video will begin to play immediately.
[Blocking Requested - why for this release]:
Major Youtube functionality break.
blocking-b2g: --- → 2.5?
This might be a good opportunity to use the mozregression tool: bug 1205560
Tried mozregression tool and it's not ready yet.  It does lead to the paths of the inbound builds so apparantly the more recent builds do have builds in inbound.  It's just extremely difficult to get to them through manual navigation on taskcluster.
QA Contact: jmercado → pcheng
I'm not sure where exactly it is failing, but turning on some of the video/media related logging may help pinpoint the problem. You can do that by changing the startup script on the device:

adb remount
adb pull /system/bin/b2g.sh .
edit and add the following line near the top of the file: export NSPR_LOG_MODULES=nsMediaElement:6,nsTrackElement:6 nsMediaElementEvents:6,MediaStreamGraph:6,MediaDecoder:6,MediaManager:6,TrackUnionStream:6,OmxDecoder:6
adb push b2g.sh /system/bin
adb shell chmod 755 /system/bin/b2g.sh
Adding qawanted to get the requested logs in comment 9.
Keywords: qawanted
(In reply to Andrew Osmond [:aosmond] from comment #9)
> I'm not sure where exactly it is failing, but turning on some of the
> video/media related logging may help pinpoint the problem. You can do that
> by changing the startup script on the device:
> 
> adb remount
> adb pull /system/bin/b2g.sh .
> edit and add the following line near the top of the file: export
> NSPR_LOG_MODULES=nsMediaElement:6,nsTrackElement:6
> nsMediaElementEvents:6,MediaStreamGraph:6,MediaDecoder:6,MediaManager:6,
> TrackUnionStream:6,OmxDecoder:6
> adb push b2g.sh /system/bin
> adb shell chmod 755 /system/bin/b2g.sh

Oh and reboot the phone afterwards for it to take effect.
Here is the log that was requested.  I had to change the export line slightly to remove the space and put a comma instead before the phone would load though.
Flags: needinfo?(ktucker)
Keywords: qawanted
mozilla-inbound regression window:

Last Working
Device: Flame 2.5
BuildID: 20150914020751
Gaia: 4d9b996be4b1935651057d0651461c1a36d98a18
Gecko: c93cd72f8d4057f536fd87cda8088828bd1bd5ef
Version: 43.0a1 (2.5)
Firmware Version: v18Dv4
User Agent: Mozilla/5.0 (Mobile; rv:43.0) Gecko/43.0 Firefox/43.0

First Broken
Device: Flame 2.5
BuildID: 20150914022657
Gaia: 4d9b996be4b1935651057d0651461c1a36d98a18
Gecko: 9a5fd09d1589e8d37c1db651896f5a2e480d9e5d
Version: 43.0a1 (2.5)
Firmware Version: v18Dv4
User Agent: Mozilla/5.0 (Mobile; rv:43.0) Gecko/43.0 Firefox/43.0

Gaia is the same so it's a Gecko issue.

Gecko pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=c93cd72f8d4057f536fd87cda8088828bd1bd5ef&tochange=9a5fd09d1589e8d37c1db651896f5a2e480d9e5d

We have triple checked the window and it is the correct window. However nothing in the pushlog really stands out for us. Changes in bug 236989 looks the most likely but we're unsure.
QA Whiteboard: [QAnalyst-Triage?]
Flags: needinfo?(jmercado)
QA Whiteboard: [QAnalyst-Triage?] → [QAnalyst-Triage+]
Flags: needinfo?(ktucker)
QA Whiteboard: [QAnalyst-Triage+] → [QAnalyst-Triage?]
Andrew can you look at the pushlogs in commment 13 and comment 14 and see if anything stands out to you?
Flags: needinfo?(jmercado) → needinfo?(aosmond)
QA Whiteboard: [QAnalyst-Triage?] → [QAnalyst-Triage+]
Maybe BZ might be able to help?  (based on comment 13)
Flags: needinfo?(bzbarsky)
The window is puzzles me. Hopefully bz has some deeper insights :). The log is even more interesting with that window in context:

09-23 14:58:42.703  1702  1702 I PRLog   : -1225186988[b364a080]: Decoder=b0061e20 ChangeState LOADING => PLAYING
09-23 14:58:42.703  1702  1702 I PRLog   : -1225186988[b364a080]: aeb25000 Ready state changed to HAVE_CURRENT_DATA
09-23 14:58:42.703  1702  1702 I PRLog   : -1225186988[b364a080]: aeb25000 Queuing event mozinterruptbegin
09-23 14:58:42.703  1702  1702 I PRLog   : -1225186988[b364a080]: aeb25000 SuspendOrResumeElement(pause=1, suspendEvents=0) hidden=0

Similar to bug 1202967, the element was paused + a mozinterruptbegin event right after it started playing; that event is triggered when the audio channel is muted. It doesn't look to ever trigger a mozinterruptend event before losing the connection to the source at 09-23 14:59:06.473.
Flags: needinfo?(aosmond)
This bug can't be repro on the latest user/eng build of Flame KK v2.5 and Aires KK v2.5 by the STR in comment 0.

Actual results: The video can be played.
See attachment: Aries_KK_v2.5(2).3gp.
Reproduce rate: 0/10

Device: Aries KK v2.5 user 
Build ID               20150923232654
Gaia Revision          8472f0c736660072799aaae60e4b6001a6aaceb4
Gaia Date              2015-09-23 10:29:02
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/f1dffc8682fbba463cb4bb305f293ddcccbc20b4
Gecko Version          44.0a1
Device Name            aries
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.worker.20150923.224603
Firmware Date          Wed Sep 23 22:46:11 UTC 2015
Bootloader             s1

Device: Aries KK v2.5 eng 
Build ID               20150923233246
Gaia Revision          8472f0c736660072799aaae60e4b6001a6aaceb4
Gaia Date              2015-09-23 10:29:02
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/f1dffc8682fbba463cb4bb305f293ddcccbc20b4
Gecko Version          44.0a1
Device Name            aries
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.worker.20150923.225204
Firmware Date          Wed Sep 23 22:52:12 UTC 2015
Bootloader             s1

Device: Flame KK v2.5  user 
Build ID               20150923150203
Gaia Revision          8472f0c736660072799aaae60e4b6001a6aaceb4
Gaia Date              2015-09-23 10:29:02
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/4a46de29431baa621d98d8f1168e43297ce1a916
Gecko Version          44.0a1
Device Name            flame
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.cltbld.20150923.184517
Firmware Date          Wed Sep 23 18:45:38 EDT 2015
Base Image             v18D v4
Bootloader             L1TC000118D0

Device: Flame KK v2.5 eng 
Build ID               20150923150203
Gaia Revision          8472f0c736660072799aaae60e4b6001a6aaceb4
Gaia Date              2015-09-23 10:29:02
Gecko Revision         https://hg.mozilla.org/mozilla-central/rev/4a46de29431baa621d98d8f1168e43297ce1a916
Gecko Version          44.0a1
Device Name            flame
Firmware(Release)      4.4.2
Firmware(Incremental)  eng.cltbld.20150923.184655
Firmware Date          Wed Sep 23 18:47:17 EDT 2015
Base Image             v18D v4
Bootloader             L1TC000118D0
I can confirm that this bug is not occurring on today's Aries build. However I noticed that there is one bug in comment 13 pushlog that was backed out yesterday, bug 1152046, which could have been the cause of this temporary no reproduction.

NI the author of 1152046 to take a look and see if 1152046 could have caused this bug.

Bug not occurring on:
Device: Aries 2.5
BuildID: 20150924111215
Gaia: 4bb17b24620818cbda0ba0c0d69e0ce3f914e1b7
Gecko: 001942e4617b2324bfa6cdfb1155581cbc3f0cc4
Gonk: 2916e2368074b5383c80bf5a0fba3fc83ba310bd
Version: 44.0a1 (2.5) 
Firmware Version: D5803_23.1.A.1.28_NCB.ftf
User Agent: Mozilla/5.0 (Mobile; rv:44.0) Gecko/44.0 Firefox/44.0
Flags: needinfo?(dd.mozilla)
STR of this bug (this STR was used to find the regression window):

1) Freshly flash the device
2) On the FTU, connect to wifi, and finish FTU without changing anything
3) Open browser, and access www.youtube.com
4) Tap on any video on the youtube page.

Do NOT search anything on youtube as it will make the bug go away. Do NOT utilize the task manager (long pressing Home button) because it will make the bug go away as well.

When the bug occurs, it is as if the video is stuck on the first frame. Or sometimes, the loading icon (NOT the animated spinning icon, if you see the spinning icon you're having internet connection issue) will appear indefinitely. Within the same session all videos have this issue. But once user uses task manager or searches on youtube, the bug will disappear and I can't figure out a way to make it happen again unless device is reflashed.
This bug should not cause such a error. on 15. September a patch that introduced a new thread for closing sockets was pushed. So all PR_Close calls are dispatched on a separate thread. Yesterday I pushed a patch that was disabling this for firefox os. So it should behave as before 15.September.

But on the 15. September another patch was pushed that my patch some how hides its failure (bug 1127270), probably because it changes timings. See bug 1207979 and 1208019. Can something like this cause this problem?

So bug 1152046 should not cause this. 
But we have here a bit strange mix of bugs here.
Flags: needinfo?(dd.mozilla)
We are no longer reproducing this issue on the latest Aries 2.5

Youtube videos are playing without issue. 

Device: Aries 2.5
Build ID: 20150924111215
Gaia: 4bb17b24620818cbda0ba0c0d69e0ce3f914e1b7
Gecko: 001942e4617b2324bfa6cdfb1155581cbc3f0cc4
Gonk: 2916e2368074b5383c80bf5a0fba3fc83ba310bd
Version: 44.0a1 (2.5)
Firmware Version: D5803_23.1.A.1.28_NCB.ftf
User Agent: Mozilla/5.0 (Mobile; rv:44.0) Gecko/44.0 Firefox/44.0
> Changes in bug 236989 looks the most likely but we're unsure.

Those changes affect the "save web page, complete" functionality in Gecko.  That's not involved in any way here, so I rather doubt those changes are relevant.
Flags: needinfo?(bzbarsky)
QA Whiteboard: [QAnalyst-Triage+] → [QAnalyst-Triage+][MGSEI-Triage+]
Closing, as its no longer reproducible
Status: NEW → RESOLVED
blocking-b2g: 2.5? → ---
Closed: 9 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: