Closed Bug 952916 Opened 11 years ago Closed 10 years ago

[Inari][System] Blank white screen after opening any application

Categories

(Firefox OS Graveyard :: Vendcom, defect)

Product:
Firefox OS Graveyard
Component:
Vendcom
Platform:
ARM
Gonk (Firefox OS)
Type:
defect
Priority:
Not set
Severity:
blocker

Tracking

(Not tracked)

Status:
RESOLVED DUPLICATE of bug 947504

People

(Reporter: viorela, Unassigned)

References

Details

(Keywords: regression)

Attachments

(2 files)

VIDEO0025.3gp
8.85 MB, video/3gpp
Details
logcat.txt
27.69 KB, text/plain
Details
Attached video VIDEO0025.3gp — 
After flashing the device with the latest master, I get a blank white screen when opening an app.

Gecko  http://hg.mozilla.org/mozilla-central/rev/530209bff5e8
Gaia  9a8757985e663978f9ea5c85e47ff6ccef096ec9
BuildID 20131222040201
Version 29.0a1

You can see the issue in the video I attached, where I opened several apps randomly.
Attached file logcat.txt — 
The blank white screen also appears after flashing the device in the FTU app, and manually we can't do anything.
Qa Wanted - Can this be reproduced on a buri device?
Keywords: qawanted
I know Askeing has a Buri device
Flags: needinfo?(fyen)
I can not reproduce it on buri with pvt m-c 20131223040210 and 20131222040201 builds.
Flags: needinfo?(fyen)
Keywords: qawanted
I can replicate this on my Inari but not on my Buri.

It makes the Inari next to unusable.
I can replicate this on Inari:

Gaia      711d8bf4efde6b1aaaf1f61bd422312c17435810
Gecko     http://hg.mozilla.org/mozilla-central/rev/cf80c0d4f46e
BuildID   20140105040201
Version   29.0a1
ro.build.version.incremental=eng.cltbld.20140105.073001
ro.build.date=Sun Jan  5 07:53:36 EST 2014
I'm making this a blocker as it makes Inari devices unusable on the latest builds
Severity: normal → blocker
Keywords: regression
I suspect this is just HWComposer issue. As long as you disable HWComposer it should be back.
But how do we disable it?
My ZTE Open reproduces this bug with the Inari build and v1.0.0B02 blobs, but works fine with v1.1.0B04 blobs.
I think this bug is caused by the GPU driver blobs and already fixed on recent version.
$ git grep 'composer' build/
build/config/common-settings.json:   "layers.composer2d.enabled": true,
(In reply to Yasuhiro Fujii from comment #10)
> My ZTE Open reproduces this bug with the Inari build and v1.0.0B02 blobs,
> but works fine with v1.1.0B04 blobs.
> I think this bug is caused by the GPU driver blobs and already fixed on
> recent version.

That seems to be reasonnable. My Inari that exposes the issue has a quite old set of blobs.
(In reply to Hubert Figuiere [:hub] from comment #9)
> But how do we disable it?

You may disable it inside the settings app -> Device Information -> More information -> Developer -> Enable hardware composer

Or use

adb shell mv /system/lib/hw/hwcomposer.msm7627a.so /system/lib/hw/hwcomposer.msm7627a.s_
Component: Gaia::System → Vendcom
(In reply to John Hu [:johnhu] from comment #13)
> (In reply to Hubert Figuiere [:hub] from comment #9)
> > But how do we disable it?
> 
> You may disable it inside the settings app -> Device Information -> More
> information -> Developer -> Enable hardware composer
> 

It is funny how I get told to do something the UI to display the UI. It's like a chicken and egg problem, but not funny.


> Or use
> 
> adb shell mv /system/lib/hw/hwcomposer.msm7627a.so
> /system/lib/hw/hwcomposer.msm7627a.s_

At least there is an alternative to try.

Nothing to do in the build instead? Something more permanent?
(In reply to Alexandre LISSY :gerard-majax from comment #11)
> $ git grep 'composer' build/
> build/config/common-settings.json:   "layers.composer2d.enabled": true,

I thought I tried this yesterday, and it didn't work.

Changing it to false will do the trick. I had to "reset-gaia" to make it work. Thanks,
With bug 950800 the toggling of HWC was transferred from a gecko pref to a gaia setting. Perhaps Inari builds were overriding the pref?
(In reply to Hubert Figuiere [:hub] from comment #15)
> (In reply to Alexandre LISSY :gerard-majax from comment #11)
> > $ git grep 'composer' build/
> > build/config/common-settings.json:   "layers.composer2d.enabled": true,
> 
> I thought I tried this yesterday, and it didn't work.
> 
> Changing it to false will do the trick. I had to "reset-gaia" to make it
> work. Thanks,

Is it working now? I too have the same problem...
> Is it working now? I too have the same problem...

The workaround is working.
I think this is going to be fixed by what's being worked on in bug 947504, as that fix is focusing on only turning the hw composer on for supported devices only.
Depends on: 947504
inari-wise, talking about the v1.1.0B04 binary blobs...  Are those something that need to be flashed into a raw partition (versus just the adrendo .so libs?)  I have a pre-ZTE-open device (ikura?) that I tried to make happy by making its backup-inari the contents of the v1.1.0B04 open image.  That did not work.  I tried to flash the device to be more ZTE-openy, but the device seems to simply lack the fastboot partitions expected on a ZTE open device.  Is there a way to get my ikura device low-level formatted/etc. or will it never support hw composer properly?
(In reply to John Hu [:johnhu] from comment #13)
> (In reply to Hubert Figuiere [:hub] from comment #9)
> > But how do we disable it?
> 
> You may disable it inside the settings app -> Device Information -> More
> information -> Developer -> Enable hardware composer


Today's built is also having the same problem but somehow able to manage to unchecked the option and device is working now.
This is also affecting eideticker, which is currently using inari devices. The workaround in comment 13 works though.
(In reply to Dave Hunt (:davehunt) from comment #22)
> This is also affecting eideticker, which is currently using inari devices.
> The workaround in comment 13 works though.

I can't seem to move the .so file as it complains that the /system is read-only?
nevermind, I forgot that I rooted the device. I had to remount 'system' and everything works now.
The gecko/gaia patches in bug 947504 just landed on m-c. With them HWC should be disabled by default on devices that don't support it. Please check if it solves this bug.
Seems the patch not been build in PVT yet...

To temporally disable it in local dev environment, 
you can add a file `custom_pref.js` in `gaia/build/` folder with following content

    // Disable HWComposer
    user_pref("layers.composer2d.enabled", false);
(In reply to Diego Wilson [:diego] from comment #25)
> The gecko/gaia patches in bug 947504 just landed on m-c. With them HWC
> should be disabled by default on devices that don't support it. Please check
> if it solves this bug.

Still the issue persists...
(In reply to Fred Lin [:gasolin] from comment #26)
> Seems the patch not been build in PVT yet...
> 
> To temporally disable it in local dev environment, 
> you can add a file `custom_pref.js` in `gaia/build/` folder with following
> content
> 
>     // Disable HWComposer
>     user_pref("layers.composer2d.enabled", false);

Will give it a try today...
(In reply to Abinash Bishoyi from comment #27)
> (In reply to Diego Wilson [:diego] from comment #25)
> > The gecko/gaia patches in bug 947504 just landed on m-c. With them HWC
> > should be disabled by default on devices that don't support it. Please check
> > if it solves this bug.
> 
> Still the issue persists...

Do you have the patches for both gecko and gaia? What do you get when you run |adb shell getprop ro.display.colorfill| ?
> Do you have the patches for both gecko and gaia? What do you get when you
> run |adb shell getprop ro.display.colorfill| ?

$ adb devices && adb shell getprop ro.display.colorfill
List of devices attached 
full_inari	device


$

Today's build(20140118065730) is working fine, May be my last build didn't have the patch.
(In reply to Abinash Bishoyi from comment #30)
> > Do you have the patches for both gecko and gaia? What do you get when you
> > run |adb shell getprop ro.display.colorfill| ?
> 
> $ adb devices && adb shell getprop ro.display.colorfill
> List of devices attached 
> full_inari	device
> 
> 
> $
> 
> Today's build(20140118065730) is working fine, May be my last build didn't
> have the patch.

Cool. Duping into bug 94750.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: