1205971 - [Flame 2.2]Device display black screen when press power key to light screen.

Status: RESOLVED WONTFIX

(Core :: Graphics: Layers, defect)

Description

[xiupinglong]

FFOS 2.2

When press power key to light screen, it will display a black screen other than light immediately.

Comment 1

[Wesly Huang (EPM)]

Hi Xiuping:

I compared 2.2 w/ 2.0 on Flame and my feeling is 2.2 looks slower to display the screen, like attached video (left side=2.2, right side=2.0), is this the same symptom you described here? If not would you mind upload your video for clarity? Thanks!

Hi Jerry:

After discussion w/ Ken he suggested you should be able to assist here, would you help take a look (is it known issue)? Also he pointed out in the test it looks 2.2 "status bar" will be firstly drawn before the rest area is displayed, is it due to the rocketbar (search field) there? Thanks!

Comment 2

[xiupinglong]

Hi Wesly,

Yes, as the comment #1 you said. FFOS2.2 will take more time to display(screen will be black for a while) than FFOS2.0 when you press power key. And the screen will display status bar first.

Comment 3

[Wesly Huang (EPM)]

Video is too big for bugzilla, put elsewhere and link here: https://youtu.be/OhmqOyygHfY

Comment 4

[Jerry Shih[:jerry] (UTC+8) (inactive)]

In v2.2, b2g will discard the layer tree when we turn off the screen[1]. When we turn on the device, b2g should paint all layers again. Thus, the black screen has shown.

[1]
https://hg.mozilla.org/releases/mozilla-b2g37_v2_2/annotate/9644c82a5b88/widget/PuppetWidget.cpp#l183

In master, we discard layer tree only when b2g is under memory-pressure.

Comment 5

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Attachment: Bug 1154231 - Part 1. Reclaim cached resources when memory-pressure occurs. for v2.2 r=mattwoodrow

Comment 6

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Attachment: Bug 1154231 - Part 2. Use LRU to manage remote layers. for v2.2 r=mattwoodrow

Comment 7

[Jerry Shih[:jerry] (UTC+8) (inactive)]

In master(v2.5), b2g discard layer tree only when it is under memory-pressure. And It also preserves at most 10 layer tree at the same time. Please check Bug 1154231.

Hi xiupinglong,
Please try these two patch and test again.

Hi Wesly,
These patch change the layer's memory management policy. We might have risk to use that.

Comment 8

[xiupinglong]

Hi Jerry,

With these two patch, it works for us now. But I just test currently by myself.
Thank you for your strongly support!

Comment 9

[Wesly Huang (EPM)]

Thanks Jerry for the prompt follow up and patch!

Just would you kindly share more information here, about the risk of using these patches (due to the change of layer's memory management policy)? Do you mean more memory consumption, or? 

I think it's needed for partner to well understand the trade-off/risk about adopting the patch. 
Thank you! 

(In reply to Jerry Shih[:jerry] (UTC+8) from comment #7)
> Hi Wesly,
> These patch change the layer's memory management policy. We might have risk
> to use that.

Comment 10

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Since we don't do the aggressively clean up, the memory footprint is higher than previous vision when we launch multiple apps. (But we will delete them with memory-pressure event.)

We have these patch in v2.5.

Comment 11

[Wesly Huang (EPM)]

Thanks Jerry, so the risk is on higher memory consumption when launch multiple Apps as there is more staff cached, but they will still be freed under memory pressure event.

Hi Xiuping, I think this is a trade-off you'll need to be aware of, thanks.

(In reply to Jerry Shih[:jerry] (UTC+8) from comment #10)
> Since we don't do the aggressively clean up, the memory footprint is higher
> than previous vision when we launch multiple apps. (But we will delete them
> with memory-pressure event.)
> 
> We have these patch in v2.5.

Comment 12

[xiupinglong]

Hi Wesly,

Thanks for your help! This is a trade-off and we will use carefully.

Comment 13

[Wesly Huang (EPM)]

Thanks Xiuping. Close this issue as won't fix in Mozilla 2.2 branch, however Jerry has helped make a patch for partner to take into their own codebase locally.

Comment 14

[xiupinglong]

(In reply to Wesly Huang (TAM) from comment #13)
> Thanks Xiuping. Close this issue as won't fix in Mozilla 2.2 branch, however
> Jerry has helped make a patch for partner to take into their own codebase
> locally.

Dear all,
I am very sorry to tell you, with these two patch can solve the screen display black issue,
but they lead to device restart when launch apps.

Comment 15

[Guoqiang.CHEN]

Dears,

When restart it occur, we can find some mozalloc error in "01-01 00:49:09.639"

01-01 00:49:09.559 I/Messages( 4682):     at startListener (app://sms_stc.gaiamobile.org/shared_stc/js/softkey_panel.js:62:4)
01-01 00:49:09.559 I/GeckoDump( 3650): [system] [HierarchyManager][155.000] system-resize
01-01 00:49:09.559 I/GeckoDump( 3650): [system] [HierarchyManager][155.001] handover system-resize to AppWindowManager
01-01 00:49:09.569 I/GeckoDump( 3650): [system] [AppWindow][Messages][AppWindow_14][155.004] request RESIZE...active? ,true
01-01 00:49:09.569 I/GeckoDump( 3650): [system] [AppWindow][Messages][AppWindow_14][155.005]  will resize... 
01-01 00:49:09.569 I/GeckoDump( 3650): [system] [AppWindow][Messages][AppWindow_14][155.005] force RESIZE...
01-01 00:49:09.569 I/GeckoDump( 3650): [system] [AppWindow][Messages][AppWindow_14][155.005] publishing internal event: withoutkeyboard
01-01 00:49:09.579 I/Messages( 4682): Content JS LOG: Softkeys engine: close group menu 
01-01 00:49:09.579 I/Messages( 4682):     at closeMenu (app://sms_stc.gaiamobile.org/shared_stc/js/softkey_panel.js:333:4)
01-01 00:49:09.589 I/GeckoDump( 3650): [system] [HierarchyManager][155.023] windowclosed
01-01 00:49:09.589 I/GeckoDump( 3650): [system] [InputWindow][Built-in Keyboard][InputWindow_3][155.024] publishing internal event: close
01-01 00:49:09.589 I/GeckoDump( 3650): [system] [InputWindow][Built-in Keyboard][InputWindow_3][155.024] publishing external event: close
01-01 00:49:09.639 I/Gecko   ( 3650): [Parent 3650] ###!!! ABORT: corrupted actor state: file /local/gc910/out/target/product/go_flip/obj/objdir-gecko/ipc/ipdl/PCompositor.cpp, line 36
01-01 00:49:09.639 E/Gecko   ( 3650): mozalloc_abort: [Parent 3650] ###!!! ABORT: corrupted actor state: file /local/gc910/out/target/product/go_flip/obj/objdir-gecko/ipc/ipdl/PCompositor.cpp, line 36
01-01 00:49:09.969 I/GeckoFocus( 4682): <<SetFocus begin>>
01-01 00:49:09.969 I/GeckoFocus( 4682): Shift Focus: li
01-01 00:49:09.969 I/GeckoFocus( 4682):  Flags: 0 Current Window: 0xb2bac3a0 New Window: 0xb2bac3a0 Current Element: 0x0
01-01 00:49:09.969 I/GeckoFocus( 4682):  In Active Window: 1 In Focused Window: 1 SendFocus: 1
01-01 00:49:09.969 I/GeckoFocus( 4682): <<Blur begin>>
01-01 00:49:09.969 I/GeckoFocus( 4682): Element (none) has been blurred
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0004 code=0x0004 value=0x00000020 when=578257493000
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0001 code=0x009e value=0x00000001 when=578257493000
01-01 00:49:10.179 I/Gonk    ( 3650): inputReader scanCode: 158
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0000 code=0x0000 value=0x00000000 when=578257493000
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0004 code=0x0004 value=0x00000020 when=578376132000
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0001 code=0x009e value=0x00000000 when=578376132000
01-01 00:49:10.179 I/Gonk    ( 3650): inputReader scanCode: 158
01-01 00:49:10.179 I/Gonk    ( 3650): Input event: device=0 type=0x0000 code=0x0000 value=0x00000000 when=578376132000
01-01 00:49:10.179 I/Gonk    ( 3650): audio_hw_primary get the key, scanCode: 158, isKeyDown: 1
01-01 00:49:10.179 I/Gonk    ( 3650): audio_hw_primary get the key, scanCode: 158, isKeyDown: 0
01-01 00:49:10.179 E/libsuspend( 3650): Error reading from /sys/power/wakeup_count: Interrupted system call
01-01 00:49:10.179 I/Gecko   ( 3650): fileReadLE: /sys/go_flip_key/lib_value
01-01 00:49:10.179 I/Gecko   ( 3650): <TZM_LOG> -*- JrdExtension -*-: _isCurrentCommand: keys = FlipState
01-01 00:49:10.189 I/GeckoDump( 3650): lxp:: permission manager return because it is hide
01-01 00:49:10.189 I/GeckoDump( 3650): [system] [AppWindow][Messages][AppWindow_14][155.630] handling keydown
01-01 00:49:10.189 I/GeckoDump( 3650): lxp:: ValueSelector -> keydown BrowserBack
01-01 00:49:10.199 I/GeckoDump( 3650): cgq BrowserBack
01-01 00:49:10.199 I/GeckoDump( 3650): lxp:: handleKeydown BrowserBack
01-01 00:49:10.199 I/GeckoFocus( 4682): Update Caret: 0 1
01-01 00:49:10.199 I/GeckoFocus( 4682): <<Focus begin>>
01-01 00:49:10.209 I/Gecko   ( 3650): fileReadLE: /sys/go_flip_key/lib_value
01-01 00:49:10.209 I/Gecko   ( 3650): <TZM_LOG> -*- JrdExtension -*-: _isCurrentCommand: keys = FlipState
01-01 00:49:10.249 I/ServiceManager(  235): service 'media.resource_manager' died
01-01 00:49:10.249 I/ServiceManager(  235): service 'SurfaceFlinger' died
01-01 00:49:10.249 I/ServiceManager(  235): service 'permission' died
01-01 00:49:10.249 I/ServiceManager(  235): service 'display.qservice' died
01-01 00:49:10.249 I/ServiceManager(  235): service 'android.security.keystore' died
01-01 00:49:10.249 E/PerMgrSrv(  244): GPS registered for modem has died
01-01 00:49:10.249 D/PerMgrSrv(  244): modem num voters is 3
01-01 00:49:10.249 D/PerMgrSrv(  244): modem state: is on-line, remove client GPS
01-01 00:49:10.249 E/QC-QMI  (  357): qmuxd: RX on fd=21 returned error=0 errno[2:No such file or directory]
01-01 00:49:10.249 E/QC-QMI  (  357): qmi_ctl_rx_msg.c Can't find txn info for txn_id = 102
01-01 00:49:10.859 D/VM_BMS  (  359): wake_lock acquired
01-01 00:49:10.859 D/VM_BMS  (  359): New BMS data - Size=56
01-01 00:49:10.859 D/VM_BMS  (  359): Start process_bms_data
01-01 00:49:10.859 D/VM_BMS  (  359): Battery charge_cycle=0 fcc_adjust=0 rbatt_adjust=0 cycle_changed=false
01-01 00:49:10.869 D/VM_BMS  (  359): For batt_temp=297 fcc_mah=1764
01-01 00:49:10.869 D/VM_BMS  (  359): Lookup fcc_mah=1764.000000  adjusted fcc_mah=1764.000000
01-01 00:49:10.869 D/VM_BMS  (  359): For ocv=4309698 batt_temp=297 soc=99
01-01 00:49:10.869 D/VM_BMS  (  359): fifo_count=5 acc_count=0
01-01 00:49:10.869 D/VM_BMS  (  359): INIT: ocv=4309698 last_ocv=4309698 batt_temp=297 soc=99
01-01 00:49:10.869 D/VM_BMS  (  359): For soc=99 batt_temp=297 slope=13000

Comment 16

[Wesly Huang (EPM)]

Hi Jerry:

Would you kindly help check and advise? Thanks!

Comment 17

[Jerry Shih[:jerry] (UTC+8) (inactive)]

I don't see the mozalloc_abort related message in adb log.

This is my test env:
device: flame-kk
gecko: git commit id: 4f370e7cf8e7ab2879944d29fc8315ce371f4cbb with these two patches.

I launch sms, contacts, music, browser and setting apps, but I still can't see that message.

What's your STR?

Comment 18

[xiupinglong]

(In reply to Jerry Shih[:jerry] (UTC+8) from comment #17)
> I don't see the mozalloc_abort related message in adb log.
> 
> This is my test env:
> device: flame-kk
> gecko: git commit id: 4f370e7cf8e7ab2879944d29fc8315ce371f4cbb with these
> two patches.
> 
> I launch sms, contacts, music, browser and setting apps, but I still can't
> see that message.
> 
> What's your STR?

We open an app and kill it, repeat these steps about more than 10 times, the restart issue occurs.

Comment 19

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Attachment: Bug 1154231 - Part 1. Use LRU to manage remote layers. for v2.2 r=mattwoodrow

Comment 20

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Attachment: Bug 1154231 - Part 2. Reclaim cached resources when memory-pressure occurs. for v2.2 r=mattwoodrow

Comment 21

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Please try these 2 patches again.

Comment 22

[xiupinglong]

Hi Jerry,

Thank you very much!
I have tested with these new two patch, it can work normal now.
I test about half an hour and the device works well.

Comment 23

[Jerry Shih[:jerry] (UTC+8) (inactive)]

Please reopen this bug if it still has problem.