1150565 - [App Launching] Launching some apps (Settings, Gallery, Music, etc) will appear blank but the buttons in the app will still respond

Status: VERIFIED FIXED

(Core :: IPC, defect)

Description

[Derek Harris [:DerekH] - Dont need info. Last day 6/5/15]

Attachment: Blank apps Logcat

Description:
Some apps will appear blank when launching them, but all the buttons will still work. This appears on apps such as: Gallery, Video, Music, Settings and sometimes Dialer. This issue occurs after Flashing or OTAing.


Repro Steps:
1) Update a Flame to 20150402063750
2) Launch Settings app

Actual: 
App will appear blank

Expected:
App will open and display all elements

Environmental Variables:
Device: Flame 3.0 (319mb)(Kitkat)(Full Flash)
Build ID: 20150402063750
Gaia: f37be8b44cb7c3a147b9615ab76743b760f08eeb
Gecko: 35046df9df1f
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0

Repro frequency: 70%
See attached: Logcat, Video - https://youtu.be/62gsZigsxD8

Comment 1

[Derek Harris [:DerekH] - Dont need info. Last day 6/5/15]

This issue does NOT occur on Flame 2.2

App will open and display all elements

Environmental Variables:
Device: Flame 2.2 (319mb)(Kitkat)(Full Flash)
Build ID: 20150402002500
Gaia: 1ceca464053dee4a8bf10ea5abeef724d68c2ff2
Gecko: 427b4da96714
Gonk: ebad7da532429a6f5efadc00bf6ad8a41288a429
Version: 37.0 (2.2)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:37.0) Gecko/37.0 Firefox/37.0

Comment 2

[Peter Bylenga [:PBylenga]]

[Blocking Requested - why for this release]:
Functional regression affecting multiple apps, fails smoke tests.

Requesting a window.

Comment 3

[Derek Harris [:DerekH] - Dont need info. Last day 6/5/15]

As an addition to apps not opening correctly, it seems that when trying to take a picture the camera app will close.

Also if this bug is not occuring the the device, it has been seen that restarting the device can sometimes get this bug to start occuring

Comment 4

[No-Jun Park [:njpark]]

Milan, this came up on today's nightly, and it is a ST blocker.  we're working on reg range at the moment, but let me know if this rings a bell to you.  There seem to be some JS warning when this happens:
https://gist.github.com/PeterBylenga/7d18d0f7252a9ccc09ae

Comment 5

[Gregor Wagner [:gwagner]]

Maybe bug 1119878?

Comment 6

[No-Jun Park [:njpark]]

It looks like Bug 1150612 is also related to this bug, although there is no report of SIM failure, (but perhaps that's because some test devices are without SIM)

Comment 7

[Brogan Zumwalt [Inactive]]

Updated STR:

1) Update a Flame to 20150402063750
2) On Homescreen wait ~1 minute for screen to turn off
3) Power on and unlock screen
4) Launch Settings app
5) If issue does not reproduce, kill Settings app from card view then repeat step 4

Central window is below, should post Mozilla-Inbound window shortly.


Central Regression Window:

Last working Central build:
Device: Flame 3.0
Build ID: 20150401030647
Gaia: 4bb3a933bd805e8df1e11827cb247754c3565b0b
Gecko: da2f28836843
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0


First broken Central build:
Device: Flame 3.0
Build ID: 20150401093537
Gaia: 4bb3a933bd805e8df1e11827cb247754c3565b0b
Gecko: e5b72a8edb82
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0


Working Gaia with Broken Gecko issue DOES repro:
Gaia: 4bb3a933bd805e8df1e11827cb247754c3565b0b
Gecko: e5b72a8edb82

Working Gecko with Broken Gaia issue does NOT repro:
Gaia: 4bb3a933bd805e8df1e11827cb247754c3565b0b
Gecko: da2f28836843


Central pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=da2f28836843&tochange=e5b72a8edb82

Comment 8

[Milan Sreckovic [:milan] (needinfo for best results)]

Thanks, and, yes, inbound range would help a lot.  There were some large or high impact changes that went in during this time (bug 847287, bug 1109390, bug 1125848, among others.)

Comment 9

[Brogan Zumwalt [Inactive]]

Mozilla-Inbound Pushlog:

Last working Mozilla-Inbound build:
Device: Flame 3.0
Build ID: 20150401010337
Gaia: 03164bd160809747e6a198e0dba1b7c3ee7789f5
Gecko: 43f8e1806067
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0

First broken Mozilla-Inbound build:
Device:  Flame 3.0
BuildID: 20150401014743
Gaia: 03164bd160809747e6a198e0dba1b7c3ee7789f5
Gecko: aee0f61516c5
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0


Mozilla-Inbound Pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=43f8e1806067&tochange=aee0f61516c5


Issue appears to occur due to changes made in bug 1119878

Comment 10

[KTucker [:KTucker][Inactive 3/4/2016]]

Bob, can you take a look at this please? bug 1119878 seems to be causing a sanity blocker and needs to be backed out.

Comment 11

[Bob Owen (:bobowen)]

As ProcessHandle and ProcessId are the same on posix, this shouldn't have caused a problem for B2G.

However, this is quite a big change and there could be some subtle changes that are causing issues.

Do we have any idea what the errors are here and where things are actually failing.

It's too late for me to look into this further at the moment and I'm not back until Tuesday.
I'm not sure how painful it will be to back this out.
Is there some reason why these tests can't be run as part of normal try and m-i runs?

There were some follow-up changes that landed today, but I don't know if they would sort this problem.

Comment 12

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

I've failed to reproduce this on a debug build of emulator-x86-kk, including with dom.ipc.processPrelaunch.enabled set to true.

Comment 13

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

According to IRC, this is already fixed by bug 1149971.  The patch for bug 1119878 used a constant initialized to the current process ID, which is incorrect when Nuwa is used; the followup patch replaced that by just calling getpid at point of use.  The revision I was testing in comment #12 had that fix, so it wouldn't have had this bug.

Comment 14

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

Is this bug present after the fix for bug 1149971 (gecko git revision 283c8df5507aeec8723518252046eb2583231fd1)?

Comment 15

[Derek Harris [:DerekH] - Dont need info. Last day 6/5/15]

We just ran a smoketest on that mozilla inbound for Flame 3.0 and it doesnt seem to be occuring. I think we should wait until testing next, or tomorrows nightly before closing this as works for me

Environmental Variables:
Device: Flame 3.0 (319mb)(Kitkat)(Full Flash)
Build ID: 20150402103013
Gaia: f37be8b44cb7c3a147b9615ab76743b760f08eeb
Gecko: a16754973862
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (3.0)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0

Comment 16

[Derek Harris [:DerekH] - Dont need info. Last day 6/5/15]

This issue is verified fixed on Flame 3.0

Apps will open and display all elements

Environmental Variables:
Device: Flame Master
Build ID: 20150403010203
Gaia: 7969b367a7da62877c3a24a26d3cb5fda89d766c
Gecko: 70a113676b21
Gonk: b83fc73de7b64594cd74b33e498bf08332b5d87b
Version: 40.0a1 (Master)
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:40.0) Gecko/40.0 Firefox/40.0

Comment 17

[Johan Lorenzo [:jlorenzo]]

Moving the bug to the component where the regression came from.