1226948 - Nightly failed to display webpage after minimizing

Status: RESOLVED DUPLICATE of bug 1234049

(Core :: Graphics, defect)

Description

[2ljpswpih]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:45.0) Gecko/20100101 Firefox/45.0
Build ID: 20151121030232

Steps to reproduce:

Enter some YouTube video. Minimize Nightly.
Go back after some time.


Actual results:

Nightly failed to display webpage. Minimizing or resizing Nightly doesn't help. You can see this here: https://youtu.be/G1A0qfcz9lk
I noticed later that opening new tab solves the problem.
This is very similar to this bug:https://bugzilla.mozilla.org/show_bug.cgi?id=1209734, but it didn't attract serious attention.
My graphics information:
Graphics
Adapter Description	Intel(R) HD Graphics 5500
Adapter Description (GPU #2)	NVIDIA GeForce GTX 850M
Adapter Drivers	igdumdim64 igd10iumd64 igd10iumd64 igd12umd64 igdumdim32 igd10iumd32 igd10iumd32 igd12umd32
Adapter Drivers (GPU #2)	nvd3dumx,nvwgf2umx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um,nvwgf2um
Adapter RAM	Unknown
Adapter RAM (GPU #2)	4095
Asynchronous Pan/Zoom	wheel input enabled
Device ID	0x1616
Device ID (GPU #2)	0x1391
Direct2D Enabled	true
DirectWrite Enabled	true (10.0.10586.0)
Driver Date	8-24-2015
Driver Date (GPU #2)	11-13-2015
Driver Version	10.18.15.4279
Driver Version (GPU #2)	10.18.13.5900
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Subsys ID	229c103c
Subsys ID (GPU #2)	229c103c
Supports Hardware H264 Decoding	Yes
Vendor ID	0x8086
Vendor ID (GPU #2)	0x10de
WebGL Renderer	Google Inc. -- ANGLE (Intel(R) HD Graphics 5500 Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d 1.1
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 1

[2ljpswpih]

I tried also with better graphic card, NVIDIA, but the problem didn't disappear, so it's rather Nightly issue.

Comment 2

[Benoit Girard (:BenWa)]

Thanks for the report. If the problem is not seen using a release build it would be very useful if you could run mozregression:
http://mozilla.github.io/mozregression/

This will help us pin-point which change is responsible so that we can avoid releasing that change.

Comment 3

[2ljpswpih]

I don't understand. 
"If the problem is not seen using a release build" - do you mean stable version of Firefox? Or do I have to create a fresh profile?

Comment 4

[Benoit Girard (:BenWa)]

I mean the stable version of Firefox. If this recently broke we can catch broke it before it makes it to release. mozregression is useful in finding out what that thing is.

Comment 5

[2ljpswpih]

Just to ensure - stable version = Firefox 42 official release?

Comment 6

[2ljpswpih]

Attachment: regression

Ok, I have done that. This tool is very nice. Attachment contains photo of the results.

Comment 7

[u279076]

(In reply to to_du from comment #6)
> Created attachment 8691584 [details]
> regression
> 
> Ok, I have done that. This tool is very nice. Attachment contains photo of
> the results.

Please click the last result, "2015-11-12", and copy&paste the Bisect information to this bug report.

Comment 8

[2ljpswpih]

app_name: firefox
build_date: 2015-11-12
build_file: c:\users\hpp\appdata\local\temp\tmpev6a6z\2015-11-12--mozilla-central--firefox-45.0a1.en-US.win64.zip
build_type: nightly
build_url: https://archive.mozilla.org/pub/firefox/nightly/2015/11/2015-11-12-03-02-38-mozilla-central/firefox-45.0a1.en-US.win64.zip
changeset: 3cc3b1968524248450c465c4ea2ee5596ffa65f2
pushlog_url: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=cc473fe5dc512c450634506f68cbacfb40a06a23&tochange=3cc3b1968524248450c465c4ea2ee5596ffa65f2
repo_name: mozilla-central
repo_url: https://hg.mozilla.org/mozilla-central
task_id: None

Comment 9

[2ljpswpih]

Does somebody apart of me reproduce this?

Comment 10

[u279076]

(In reply to to_du from comment #9)
> Does somebody apart of me reproduce this?

I have not personally seen this issue nor have I heard of reports from other users but that doesn't necessarily mean it's just you.

Benoit, the pushlog above is pretty big but does something jump out at you as a potential cause? If not we may need to get to_du to bisect this further on inbound.

Comment 11

[Benoit Girard (:BenWa)]

Sorry for the delay. Did the tool not offer you to bisect on inbound? This list is too large with several suspect changes in the graphics/layout module that could account for this. It would help greatly if we can bisect down to a single change.

It's possible that the GUI app doesn't support inbound bisection and that you might need to use the command line tool.

Comment 12

[2ljpswpih]

I am able to do it, but nobody ask me to do it ;).

Comment 13

[2ljpswpih]

*asked

Comment 14

[Alice0775 White]

(In reply to to_du from comment #6)
> Created attachment 8691584 [details]
> regression
> 
> Ok, I have done that. This tool is very nice. Attachment contains photo of
> the results.

From attached regression range:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=cc473fe5dc512c450634506f68cbacfb40a06a23&tochange=3cc3b1968524248450c465c4ea2ee5596ffa65f2

The most suspicious:
67da9ced67f1	George Wright — Bug 1098131 - Don't invalidate layers when simply changing SizeMode r=smaug,jimm

Comment 15

[2ljpswpih]

app_name: firefox
build_date: 2015-11-13 05:41:38.849000
build_file: c:\users\hpp\appdata\local\temp\tmpd5cpb9\0c648a1efbe0--mozilla-central--firefox-45.0a1.en-US.win64.zip
build_type: inbound
build_url: https://queue.taskcluster.net/v1/task/joElWb60TayPi1IC-HD_Lw/runs/0/artifacts/public%2Fbuild%2Ffirefox-45.0a1.en-US.win64.zip
changeset: 0c648a1efbe06b5ec866ba058d18256b80808b46
pushlog_url: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=3cc3b1968524248450c465c4ea2ee5596ffa65f2&tochange=0c648a1efbe06b5ec866ba058d18256b80808b46
repo_name: mozilla-central
repo_url: https://hg.mozilla.org/mozilla-central
task_id: joElWb60TayPi1IC-HD_Lw

I hope it will help. I have to test several inbound builds ;).

Comment 16

[2ljpswpih]

Attachment: I don't know if I have pasted rigth informations, because I was testing and from some point every build was bad. It ended with this build checked, but maybe I should paste the last build information.

Comment 17

[2ljpswpih]

I think I have understood it a little. I have noticed that simply this program is still bisecting and can not find another build.
So here are informations from the latest tested build.

app_name: firefox
build_date: 2015-11-13 01:01:10.335000
build_file: c:\users\hpp\appdata\local\temp\tmpd5cpb9\285852de5cd5--mozilla-central--firefox-45.0a1.en-US.win64.zip
build_type: inbound
build_url: https://queue.taskcluster.net/v1/task/GURSZNOgQ7e2X_1wPCeMHg/runs/0/artifacts/public%2Fbuild%2Ffirefox-45.0a1.en-US.win64.zip
changeset: 285852de5cd54891d8f3b1ddbff179e4c9c7ca5f
pushlog_url: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=3cc3b1968524248450c465c4ea2ee5596ffa65f2&tochange=285852de5cd54891d8f3b1ddbff179e4c9c7ca5f
repo_name: mozilla-central
repo_url: https://hg.mozilla.org/mozilla-central
task_id: GURSZNOgQ7e2X_1wPCeMHg


After this build, program can not find earlier build.

Comment 18

[2ljpswpih]

I tried to do it one more time, but the same happened - after build 285852de from 13.11.2015 01:01:10.335000 program keeps searching for next build and can not do it.

Comment 19

[Alice0775 White]

you can download firefox-45.0a1.en-US.win64.zip file inbound build from http://archive.mozilla.org/pub/firefox/tinderbox-builds/mozilla-inbound-win64/

Please try each from 1447277373/ to 1447388376/

Comment 20

[George Wright (:gw280) (needinfo me!)]

This should have been fixed by the fix for bug 1234049. Can you reproduce with the latest nightly? I see from your about:support that the builds are quite old (this fix would have started in the January 18th Nightly)

Comment 21

[2ljpswpih]

Hmm, I have tested another time, this time setting differently days with the similar effect like before, but fortunately, after George's post I tested this in the latest Nightly and it doesn't appear :-D So I think there is no need to continue this bug ;)
This bug was present surely 3 months ago. When Benoit came back to this, I haven't checked whether it still persists ;)

Comment 22

[Benoit Girard (:BenWa)]

Great, sorry for taking so long to get back to this bug but I'm glad it's fixed. Thanks for doing the regression windows that was very useful!