Closed Bug 915811 Opened 11 years ago Closed 11 years ago

Defect - touch scrolling results in mostly black page or black rects at view bounds

Categories

(Firefox for Metro Graveyard :: Pan and Zoom, defect)

Product:
Firefox for Metro Graveyard
Component:
Pan and Zoom
Version:
27 Branch
Platform:
x86_64
Windows 8.1
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

(Not tracked)

Status:
RESOLVED WORKSFORME

People

(Reporter: moz-bugzilla, Unassigned)

References

Details

(Whiteboard: [block28])

Attachments

(5 files)

screenshot of 2013-09-07
164.15 KB, image/png
Details
graphics section of about:support on sony vaio that does NOT reproduce the problem
104.40 KB, image/png
Details
about:support on VAIO Duo 11 where the bug DOES show
2.84 KB, text/plain
Details
half of the screen completely white
325.21 KB, image/jpeg
Details
home page black bug
38.11 KB, image/png
Details 
User Agent: Mozilla/5.0 (Windows NT 6.2; Win64; x64; rv:26.0) Gecko/20100101 Firefox/26.0 (Beta/Release)
Build ID: 20130906030202

Steps to reproduce:

On any Nightly after 2013-09-07 (or so, running -06 here), whenever I scroll down on a page, the content will correctly shift up, but the content that is supposed to be scrolled into view is not there, instead, it's all black. Scrolling with the cursor keys works fine.

Also, when I "touch scroll", it only seems as if it was scrolling, but when I then pressed cursor down, it starts scrolling from the top of the page.
Severity: normal → major
Can you post a good example page?

(This is probably a dupe of bug 907100.)
Blocks: 915723
Really, any page which is longer than the screen is high will do. I can try to get a sreenshot. I did look at #907100 but it seemed slightly different to me, because it sounded like the view will recover after panning. I'm not even able to pan down anywhere without getting the black bars.
Latest nightly. I was just trying this on www.macrumors.com and I can see top and bottom sections of the page in black, depending on the scrolling direction. In my case, after a few scrolls up and down the page went blank and I had to reload.
Here's a scrceenshot. You can see that only the initial view of the page has been rendered correctly, the rest isn't shown at all.

This is on a Full HD 11" display (Sony VAIO Duo 11).
Oh and forgot to add: works fine in 2013-09-06, broken in 2013-09-07 and all newer builds.
It also happens whenever you want to scroll (left, up, right or down) as long as the scrolling is possible.
For example if you are in the top left corner of the page, black bands appear when you scroll right or down.
There is no issue with a mouse or when the scroll bars are used, though black areas can be seen but they disappear instantly when the page stops scrolling.
I have tried with the latest Nightly and the problem is still very much present. I can also confirm comment 6. The problem does NOT occur when Nightly is started from the desktop with -metrodesktop. I'll try to search the commits between 09-06 and 09-07 to see if I can find anything, but I don't really know what to look for.

It is also the case that touches in the black area are not recognized, so it seems that really only the rendered part of the website is moved away but the notification to the renderer (?) that the newly visible areas should be rendered do not come through.
Severity: major → critical
OS: All → Windows 8 Metro
Hardware: All → x86_64
Version: 26 Branch → 27 Branch
When I set layers.async-pan-zoom.enabled to false, things work again, although it feels a bit slow. Can I suspect that there's an incompatibility with the (Intel) graphics drivers which causes this?
(In reply to Stefan Seidel from comment #8)
> When I set layers.async-pan-zoom.enabled to false, things work again,
> although it feels a bit slow. Can I suspect that there's an incompatibility
> with the (Intel) graphics drivers which causes this?

I think this is just a bug in our new apz code we need to track down.
This is specific to black area / redraw issues with touch input. Other redraw issues are being dealt with in other bugs.
Status: UNCONFIRMED → NEW
Ever confirmed: true
The bug (or rather the effects of it) made it into the news. Since Mozilla decided to release a public Preview of Firefox for Metro, heise online, one of the biggest news site in Germany, tested it:
http://www.heise.de/newsticker/meldung/Firefox-Preview-fuer-Windows-8-Tablets-1964872.html
They found this bug on a Win8 Desktop PC with touch and on a Samsung tablet, stating that only on the Surface Pro touch scrolling worked, but was sluggish.

So I don't know who's responsible for release/announcements - but that wasn't good timing.
Still have this bug in the latest version of Nightly on my ASUS VivoTab Smart ME400C.
I can't reproduce this on my Surface Pro and botond can't repro on his touchscreen ThinkPad Carbox C1. Is there anybody who can reproduce this who would be able to build a local build of Firefox to test some things? I can guide you through the diagnosis of the problem.
(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #13)
> I can't reproduce this on my Surface Pro and botond can't repro on his
> touchscreen ThinkPad Carbox C1. Is there anybody who can reproduce this who
> would be able to build a local build of Firefox to test some things? I can
> guide you through the diagnosis of the problem.

Also, if there's a device in the Toronto office that can repro this, I'm happy to try debugging this on it myself.
I can reproduce this problem with the Lenovo X1 Carbon, when scrolling through long websites, I see flickers of black and sometimes the majority of the screen quickly goes black. I have the source and VS2012 installed on this laptop.

Maybe we can setup a remote debugging session? Someone could login into this machine and take a look at it.
Hi Botond, I can bring a device for you tomorrow.
(In reply to Kamil Jozwiak [:kjozwiak] from comment #15)
> I can reproduce this problem with the Lenovo X1 Carbon, when scrolling
> through long websites, I see flickers of black and sometimes the majority of
> the screen quickly goes black. I have the source and VS2012 installed on
> this laptop.

Do you only see the black strip for a short time (like a fraction of a second)? IIUC, that's a different (known) issue - the issue in this bug in that the black strip stays there.

I have an X1 Carbon too, by the way, and I only see the black strips for a fraction of a second.
(In reply to Botond Ballo [:botond] from comment #17)
> (In reply to Kamil Jozwiak [:kjozwiak] from comment #15)
> > I can reproduce this problem with the Lenovo X1 Carbon, when scrolling
> > through long websites, I see flickers of black and sometimes the majority of
> > the screen quickly goes black. I have the source and VS2012 installed on
> > this laptop.
> 
> Do you only see the black strip for a short time (like a fraction of a
> second)? IIUC, that's a different (known) issue - the issue in this bug in
> that the black strip stays there.
> 
> I have an X1 Carbon too, by the way, and I only see the black strips for a
> fraction of a second.

Yup, only see it a fraction of a second when scrolling through long websites. Assumed that it was the same issue, my apologies! If I see anything like this I will add more information into the ticket.
(In reply to Botond Ballo [:botond] from comment #17)

> Do you only see the black strip for a short time (like a fraction of a
> second)? IIUC, that's a different (known) issue - the issue in this bug in
> that the black strip stays there.

In my case (ASUS VivoTab Smart ME400C) these black parts doesn't disappear. So I can't use Firefox on my tablet at all.
I can test at least daily. I'm a software developer myself so don't be afraid to ask me some technical details. I can't compile stuff though on that machine, also I probably wouldn't have time to do so.
I tried reproducing this with a Samsung ATIV Smart PC device that Marco gave me, but wasn't able to.
(In reply to Max Tr from comment #19)
> In my case (ASUS VivoTab Smart ME400C) these black parts doesn't disappear.
> So I can't use Firefox on my tablet at all.

Does it work when you go to about:config and apply the workaround from my comment #8 ?
Sorry, didn't see that post. 

Yes, it helped. And yes, it causes the reduction of speed on my tablet too.
(In reply to Botond Ballo [:botond] from comment #21)
> I tried reproducing this with a Samsung ATIV Smart PC device that Marco gave
> me, but wasn't able to.

Nor with an Acer Iconia W5.
(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #13)
> botond can't repro on his touchscreen ThinkPad Carbox C1.

That was a typo on IRC, by the way. It's Carbon X1 :)
(In reply to Stefan Seidel from comment #20)
> I can test at least daily. I'm a software developer myself so don't be
> afraid to ask me some technical details. I can't compile stuff though on
> that machine, also I probably wouldn't have time to do so.

I got a hold of a Sony VAIO Duo 11. Updated to latest nightly build (2013-10-15), went to macrumors.com and scrolled down at various speeds, but never saw anything more than a fraction-of-a-second black flicker. Tried with a few other websites as well, same thing.

Stefan, could you confirm that the same page on the same device with the same build produces this problem for you?
Flags: needinfo?(moz-bugzilla)
If you can get about:support to load properly, please post the graphics section too. Alternatively you can post the graphics section from desktop, although that's not as helpful.
I can get about:support to load properly. Attached is a screenshot of the graphics section.
Here's my about:support. Just tested with 2013-10-10
Flags: needinfo?(moz-bugzilla)
I have the black bands on any page, even in about:support and even if I change the layers.async-pan.zoom on about:config to false. I have a Microsoft Surface Pro 128Mo.
(In reply to kpoirot from comment #30)
> I have the black bands on any page, even in about:support and even if I
> change the layers.async-pan.zoom on about:config to false. I have a
> Microsoft Surface Pro 128Mo.

brief black bands while scrolling or issues where the entire window doesn't paint during a scroll?
(In reply to Stefan Seidel from comment #29)
> Created attachment 818051 [details]
> about:support on VAIO Duo 11 where the bug DOES show
> 
> Here's my about:support. Just tested with 2013-10-10

From both of the about:supports, looks like you are running a newer driver. Botond, maybe you could try updating to see if you can reproduce with the same driver Stefan has?

works:
7-20-2012
9.17.10.2817

doesn't:
12-12-2012
9.17.10.2932
(In reply to Jim Mathies [:jimm] from comment #32)
> Botond, maybe you could try updating to see if you can reproduce with the
> same driver Stefan has?
> 
> works:
> 7-20-2012
> 9.17.10.2817
> 
> doesn't:
> 12-12-2012
> 9.17.10.2932

Device Manager tells me that I already have the newest driver. Should I try to install the 9.17.10.2932 version anyways? Is there an official place to download it from?
(In reply to Stefan Seidel from comment #34)
> I got mine from here :
> http://www.sony.co.uk/support/en/product/SVD1121C5EB/updates

I tried with the new driver - no difference.
(In reply to Jim Mathies [:jimm] from comment #31)
> (In reply to kpoirot from comment #30)
> > I have the black bands on any page, even in about:support and even if I
> > change the layers.async-pan.zoom on about:config to false. I have a
> > Microsoft Surface Pro 128Mo.
> 
> brief black bands while scrolling or issues where the entire window doesn't
> paint during a scroll?

I have permanent black bands. Only the first screen which appears when I load the page is painted. If I use touch, I have only permanent black bands. And when I use the scroll bar, I have also black bands but these one disappear after few seconds.
could this be related? I am getting this issue here and there where half of the screen just turns white making it unusable (usually have to restart the browser. Always occurs vertically while I am scrolling using touch.
(In reply to Kamil Jozwiak [:kjozwiak] from comment #37)
> Created attachment 818160 [details]
> half of the screen completely white
> 
> could this be related? I am getting this issue here and there where half of
> the screen just turns white making it unusable (usually have to restart the
> browser. Always occurs vertically while I am scrolling using touch.

Doesn't sound like it at all unless it occurs when you scroll using touch screen.
(In reply to Botond Ballo [:botond] from comment #35)
> (In reply to Stefan Seidel from comment #34)
> > I got mine from here :
> > http://www.sony.co.uk/support/en/product/SVD1121C5EB/updates
> 
> I tried with the new driver - no difference.

Just to be 100% sure - you're starting it from the metro UI, not from the desktop (using -metrodesktop)? Because with the latter, the bug is not present on my system, but when I run FF as intented (makeing it the default browser, fully filled tile), the bug does occur.

Somehow yesterday it also said "Nightly is up to date" even though the version was only 2013-10-10. I'm pretty sure I'm using the version I downloaded from the firefox-central directory. Also, IIRC it's the x64 version, not i686. Is that the same you're using or should I be using another version?
(In reply to Stefan Seidel from comment #39)
> (In reply to Botond Ballo [:botond] from comment #35)
> > (In reply to Stefan Seidel from comment #34)
> > > I got mine from here :
> > > http://www.sony.co.uk/support/en/product/SVD1121C5EB/updates
> > 
> > I tried with the new driver - no difference.
> 
> Just to be 100% sure - you're starting it from the metro UI, not from the
> desktop (using -metrodesktop)? Because with the latter, the bug is not
> present on my system, but when I run FF as intented (makeing it the default
> browser, fully filled tile), the bug does occur.
> 
> Somehow yesterday it also said "Nightly is up to date" even though the
> version was only 2013-10-10. I'm pretty sure I'm using the version I
> downloaded from the firefox-central directory. Also, IIRC it's the x64
> version, not i686. Is that the same you're using or should I be using
> another version?

Can you install the 32-bit build and test to see if you still experience the problem? We currently don't officially support 64-bit builds, although we're working on changing that.
(In reply to Stefan Seidel from comment #39)
> (In reply to Botond Ballo [:botond] from comment #35)
> > (In reply to Stefan Seidel from comment #34)
> > > I got mine from here :
> > > http://www.sony.co.uk/support/en/product/SVD1121C5EB/updates
> > 
> > I tried with the new driver - no difference.
> 
> Just to be 100% sure - you're starting it from the metro UI, not from the
> desktop (using -metrodesktop)?

Yup.

> Somehow yesterday it also said "Nightly is up to date" even though the
> version was only 2013-10-10. I'm pretty sure I'm using the version I
> downloaded from the firefox-central directory. Also, IIRC it's the x64
> version, not i686. Is that the same you're using or should I be using
> another version?

I followed these instructions [1] to install Nightly, or if it was already present, I updated it in desktop mode and then ran it again from the Metro UI.
 
[1] https://support.mozilla.org/en-US/kb/how-do-i-install-windows-8-metro-style-firefox
(In reply to Kamil Jozwiak [:kjozwiak] from comment #37)
> Created attachment 818160 [details]
> half of the screen completely white
> 
> could this be related? I am getting this issue here and there where half of
> the screen just turns white making it unusable (usually have to restart the
> browser. Always occurs vertically while I am scrolling using touch.

I saw this once on the Sony VAIO device, after changing the screen's orientation and changing it back, but I couldn't reproduce it again. Probably a different problem.
Ok I installed 2013-10-18, x86 aka 32 bit and - no difference. I'll wait for Win 8.1 to install and test again. After that, somebody will probably need to tell me whether I can compile a local build for debugging using only free software (e.g. Visual Studio Express), then I could do some testing. I wish there was a way to reproduce this on some hardware you can get your hands on.

The only thing which might be different on my VAIO than on the one you had is that I installed my personal copy of Windows 8 Pro from scratch, wiping the previous Windows 8 (non-Pro) install and commanding the Windows Setup to ignore the device built-in license key.
I'm using an Dell XPS 12 (Haswell-Version) with Windows 8.1 and the newest graphics driver from Intel (10.18.10.3316 from Windows Update) and still see this problem. Scrolling results in big black areas and the zoom makes everything blurry (see Bug 902505) making the metro version unusable. So I don't think that Windows 8.1 will make a difference.
I have exactly the same issue here on Lenovo ThinkPad EDGE E325 with latest video driver (AMD, integrated into E450 CPU) from Microsoft. Windows 8.1 on amd64.
(In reply to Max Tr from comment #19)
> (In reply to Botond Ballo [:botond] from comment #17)
> 
> > Do you only see the black strip for a short time (like a fraction of a
> > second)? IIUC, that's a different (known) issue - the issue in this bug in
> > that the black strip stays there.
> 
> In my case (ASUS VivoTab Smart ME400C) these black parts doesn't disappear.
> So I can't use Firefox on my tablet at all.

I have the same problem on my vivotab smart. I have updated intel graphic driver without result
Attached image home page black bug — 

    
    

    
  
(In reply to Dmitrij D. Czarkoff from comment #45)
> I have exactly the same issue here on Lenovo ThinkPad EDGE E325 with latest
> video driver (AMD, integrated into E450 CPU) from Microsoft. Windows 8.1 on
> amd64.

Forgot to mention - I don't have touch screen (classic laptop), but bug still happens with any way of scrolling here.

    
  
Summary: Defect - touch scrolling results in mostly black page → Defect - touch scrolling results in mostly black page or black rects at view bounds

    
  
Whiteboard: [blocker]
Whiteboard: [blocker] → [block28]

    
    

    
  
I update nightly yesterday 10pm and bug has disappear

    
    

    
  
Well, disappeared is a strong word. It has kinda morphed into a version like bug 907100 - the browser with APZ is definitely usable now, but black bars appear for sub-second periods when touch scrolling (and occasionally when scrolling very fast with the keyboard/scrollwheel, too), but then the black bars are getting replaced by the content.

It is nicely visible during the "momentum" phase of the touch "fling", it seems like the rendered is just always a little behind of the actual view.

    
    

    
  
(In reply to Stefan Seidel from comment #50)
> Well, disappeared is a strong word. It has kinda morphed into a version like
> bug 907100 - the browser with APZ is definitely usable now, but black bars
> appear for sub-second periods when touch scrolling (and occasionally when
> scrolling very fast with the keyboard/scrollwheel, too), but then the black
> bars are getting replaced by the content.
> 
> It is nicely visible during the "momentum" phase of the touch "fling", it
> seems like the rendered is just always a little behind of the actual view.

Same for me.

    
    

    
  
Ok, so it sounds like the "permanent" black bars are gone, so I'm going to close this bug as WFM. The "temporary" black sections are checkerboarding (i.e. we're not painting fast enough to keep up with scrolling). Bug 907179 covers this case.
Status: NEW → RESOLVED
Closed: 11 years ago
Resolution: --- → WORKSFORME

    
    

    
  
My laptop is not touch enabled, but I randomly get black bars (particularly frequently on this bugzilla). It only happens in new tabs, which render fine but get covered with black bars from top or bottom of display depending on scrolling direction. The contents of the clean portion of the screen (if any remains) doesn't get scrolled, but gets redrawn on scrolling. Dragging the "window" with a mouse (the gesture used for snapping and closing apps, I don't know how it's called) fixes the issue.

    
    

    
  
Disregard my previous comment, please, - I was on Aurora, and Nightly as of now seems OK.

    
  
No longer blocks: 915723

    
  
Blocks: metro-apzc
OS: Windows 8 Metro → Windows 8.1

          You need to log in
          before you can comment on or make changes to this bug.
        






  

    
  







  

    

      
Attachment

      

      
      
      
      
    

    

      

        

          

            
General

            

              Creator: 



            

            
Created: 

            
Updated: 

            
Size: