Closed Bug 901336 Opened 11 years ago Closed 8 years ago

Rounded button using border-radius rendered incorrectly

Categories

(Core :: Graphics, defect)

Product:
Core
Component:
Graphics
Version:
22 Branch
Platform:
x86_64
Windows 7
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: john.foreign, Unassigned)

References

(Blocks 1 open bug)

Details

Attachments

(2 files)

testcase
321 bytes, text/html
Details
Outout from about:support on buggy system
3.59 KB, text/plain
Details 
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:22.0) Gecko/20100101 Firefox/22.0 (Beta/Release)
Build ID: 20130618035212

Steps to reproduce:

On Windows 7 (64-bit) the follwing shows the problem:
<html>
<head>
  <title>Firefox border-radius bug demo</title>
  <style type="text/css">
    BUTTON {
      border-radius: 2em 2em 2em 2em;
      font-size: 110%;
      padding: 0.5em 1em;
      min-width: 7em;
      padding: 1px 1em;
    }
  </style>
</head>
<body>
  <button type="submit">Submit Exercise</button>
</body>
</html>


Actual results:

The result is shown here: https://dl.dropboxusercontent.com/u/18254707/button_firefox.jpg
The top right & bottom left borders appear to be rotated or inverted.
This was reported originally in https://bugzilla.mozilla.org/show_bug.cgi?id=431176.
Maybe related to https://bugzilla.mozilla.org/show_bug.cgi?id=613266 (?)



Expected results:

The same thing with Chrome renders correctly: https://dl.dropboxusercontent.com/u/18254707/button_chrome.jpg
Attached file testcase — 
As said in Bug 431176 comment 65, WFM in Fx 23-25 on Mac OS X 10.8.4. So this may be a platform-specific issue and/or already fixed. 

Please download Firefox beta (https://mozilla.org/firefox/beta/) and check the attached testcase again.
Attachment #785506 - Attachment mime type: text/plain → text/html
The testcase renders correctly both in Firefox 22 (same build as the reporter's one) and in latest Firefox 23 Beta (20130729175331) under Windows 7 Home Premium x64.

Probably the issue is caused by some extension, or the provided testcase is different from a real web page discovering the issue.
(In reply to Marat Tanalin | tanalin.com from comment #2)
> The testcase renders correctly both in Firefox 22 (same build as the
> reporter's one) and in latest Firefox 23 Beta (20130729175331) under Windows
> 7 Home Premium x64.
> 
> Probably the issue is caused by some extension, or the provided testcase is
> different from a real web page discovering the issue.

On my system the testcase renders incorrectly, exactly as shown in the image. So it's no different behaviourally from the real pages where I first noticed it.

What else can I tell you that might help you to reproduce it on YOUR system? The only extension I have installed is FireBug 1.11.4; apart from that I have a handful of add-ins (Acrobat 9.5.1.283, Google Earth 7.1.1.1888, Java 7u25, Shockwave Flash 11.7.700.224 and Shockwave for Director 11.5.7.609).
(In reply to comment #3)
I would try to:

1. disable ALL extensions and plugins (to guarantee that they do not affect pages in any way);

2. update videocard's drivers to latest version (drivers installed currently may be outdated/buggy);

3. disable antivirus and any other third-party software that may intercept HTTP traffic and introduce nonstandard elements to web pages (e.g. Skype).

By the way, did you try the testcase copy available at this bugzilla ticket itself (this would prevent some sort of cache that may affect your original page rendering)? Or at least did you try to reload your buggy page/testcase completely by pressing Ctrl+F5 in the browser?
(In reply to Marat Tanalin | tanalin.com from comment #4)
> 1. disable ALL extensions and plugins (to guarantee that they do not affect
> pages in any way);
> 2. update videocard's drivers to latest version (drivers installed currently
> may be outdated/buggy);
> 3. disable antivirus and any other third-party software that may intercept
> HTTP traffic and introduce nonstandard elements to web pages (e.g. Skype).

I disabled all the add-ons and plugins, closed and restarted FF, and it made no difference. The video driver is the latest, and I'm not running Skype etc.  Locally I'm just opening an HTML file with FF, and not accessing it on a server. In any case remember that the same page renders correctly on the same system in both Chrome and IE, so it's unlikely to be an external process interfering with HTTP traffic. The site I noticed it on uses HTTPS, btw.

> By the way, did you try the testcase copy available at this bugzilla ticket
> itself (this would prevent some sort of cache that may affect your original
> page rendering)? Or at least did you try to reload your buggy page/testcase
> completely by pressing Ctrl+F5 in the browser?

Both, and it makes no difference either way.
(In reply to John from comment #5)
> Both, and it makes no difference either way.

I just tried it out on the other two systems I have avaiable at the moment: FF 22 running on WinXP, and FF 15.0.1 running on an old Ubuntu. Both of these render the page correctly. It seems to be something very system-specific for some reason...
Try to create a new Firefox profile:
https://support.mozilla.org/kb/Managing-profiles
A low-level graphics issue seems unlikely to be related to extensions and/or profiles.

Could you go to the URL about:support , use the "copy text to clipboard" button, and paste the results into this bug (or into an attachment on this bug).  If you're not comfortable pasting the whole thing, then at least the "Application Basics" and "Graphics" sections?
Blocks: border-radius
Status: UNCONFIRMED → NEW
Component: Untriaged → Graphics
Ever confirmed: true
Product: Firefox → Core
Also, do you know if this worked in previous Firefox versions on the problematic system?  If it did, can you figure out when it broke?
Flags: needinfo?(john.foreign)
(In reply to Marat Tanalin | tanalin.com from comment #7)
> Try to create a new Firefox profile:
> https://support.mozilla.org/kb/Managing-profiles

Tried that. No change.
Flags: needinfo?(john.foreign)

    
    

    
  
(In reply to David Baron [:dbaron] (don't cc:, use needinfo? instead) from comment #9)
> Also, do you know if this worked in previous Firefox versions on the
> problematic system?  If it did, can you figure out when it broke?

Yes it did. I added the rounded button style to the CSS (using this machine for the development) about 8 months ago (21 Dec 2012 according to the subversion log), so I would have noticed it at the time while testing. So sometime this year, basically. I download and install updates when advised by the update manager.

    
    

    
  
And both of these may in turn be related to bug 851764 and bug 849517.  All are reports of problems with rounded borders on Intel graphics cards with device ID 0x24a2.

    
    

    
  
(In reply to David Baron [:dbaron] (don't cc:, use needinfo? instead) from comment #12)
> This seems likely to be the same as bug 894542.

Sounds like it. However I can't seem to do anything about hardware acceleration; the Change Settings button under Control Panel -> Display -> Adjust settings -> Advanced is greyed out... I hate Windows 7... :-(

I wonder how Chrome and IE do the same rendering?

    
    

    
  
John, did you try disabling hardware acceleration in Firefox itself? (in Options > Advanced > General)

    
    

    
  
(In reply to mjh563 from comment #16)
> John, did you try disabling hardware acceleration in Firefox itself? (in
> Options > Advanced > General)

Excellent, that fixes it. So that is the problem after all.

    
    

    
  
I can't replicate it (anymore!?). Is this still a problem for you John?
Flags: needinfo?(john.foreign)

    
    

    
  
I followed mjh563's advice and disabled hardware acceleration, and then the buttons displayed properly, but that was over 2 years ago. Since 2013 I have changed computer and OS a couple of times. On my current machine, hardware acceleration is enabled and it displays correctly, but I can't tell you any more than that -- sorry!

    
    

    
  
Oops, forget to tick the "clear needinfo" box.
Flags: needinfo?(john.foreign)

    
    

    
  
Thanks John!

David, I have quickly tested it on my Win7 VM with FF 47 and had no issues. Does that count as fixed?
Flags: needinfo?(dbaron)

    
    

    
  
As nobody is able to reproduce it (probably due to bugfixes about HWA), let's close it.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → WORKSFORME

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






  

    
  







  

    

      
Attachment

      

      
      
      
      
    

    

      

        

          

            
General

            

              Creator: 



            

            
Created: 

            
Updated: 

            
Size: