Closed Bug 693056 Opened 14 years ago Closed 14 years ago

Firefox window title, border and controls corrupted

Categories

(Core :: Graphics, defect)

Product:
Core
Component:
Graphics
Platform:
x86_64
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: kdevel, Unassigned)

Details

(Whiteboard: [CLOSEME 2011-11-15])

Attachments

(4 files)

Screenshot 1 program title bar has two titles
11.00 KB, image/jpeg
Details
Screenshot 2 Program window border with short horizontal stripes
33.64 KB, image/jpeg
Details
Screenshot 3 horizontal pattern one pixel high
20.97 KB, image/jpeg
Details
Screenshot 4 corrupted title bar, fresh profile
25.46 KB, image/jpeg
Details
User Agent: Steps to reproduce: 1. Open three tabs with many and large graphics one of which shall be http://www.theatlantic.com/infocus/2011/10/occupy-wall-street-spreads-beyond-nyc/100165/ 2. Play with tabs, scroll, resize window, switch from full-screen to normal view. Click at 1280px option etc. Actual results: Some program border elements become corrupted. Expected results: No corruption.
OS: Other → Linux
Hardware: Other → x86_64
Please attach a screenshot.
Screenshots 1 & 2 were produced with Fx built from revision b4da2d439cbc. Screenshot 3 was procduced with Fx built from current revision 6d5fd5a30c71.
Is there a race condition when switching between two tabs (with the mouse) regarding the update of the program title in the title bar? I can reproduce the situation of Screenshot 1 with nightlies from 2011-01 and with 3.6.23.
What is the fish head doing there?:) I do not see it. What desktop environment are you using (KDE, Gnome, ... ) what is the window manager? Are you using hardware acceleration? Paste Graphics section of about:support .
Summary: program window's border and controls corrupted → Firefox window title, border and controls corrupted
(In reply to aceman from comment #8) > What is the fish head doing there?:) I do not see it. Pardon? Anyway. The cause seems to be the graphics driver. So this bug can be closed.
In all the screenshots, I see head of a fish in the window title bar after the window title text. The fish does not show on my system when I visit that page. If you think it is the driver, please explain, we'd like to hear the solution.
(In reply to aceman from comment #10) > The fish does not show on my system when I visit that page. It's not a fish, it's a gecko: http://www.google.de/search?q=suse+gecko http://de.opensuse.org/skins/opensuse/images/common/geeko.jpg > If you think it is the driver, please explain, we'd like to hear the > solution. The problem shows up with xf86-video-ati 6.13.0 and 6.13.2 while 6.12.99 seems to be OK.
OK, so it is the head of a gecko:) Is it supposed to be there in the title? Is that part of the bug or not? That is interesting. I also use these ATI drivers and I do not see the bug. I am on version 6.14.1. Can you try that?
(In reply to aceman from comment #12) > OK, so it is the head of a gecko:) Is it supposed to be there in the title? > Is that part of the bug or not? There is usually one of these heads to the right of the program title: http://alternativenayk.files.wordpress.com/2008/07/googledocsproblem.png > That is interesting. I also use these ATI drivers and I do not see the bug. > I am on version 6.14.1. Can you try that? Unfortunately not (yet). There has been an ABI change after 6.13.2 which prevents my Xorg from starting with 6.14.2. I'll stay with 6.12.192. The error (not the ABI one but the one of this bug) is in That's the changeset which proce http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=53ac06161eb2b8cffb1b88e24a9a21cfd12e5883
What is your Xorg version?
X.Org X Server 1.7.4 Release Date: 2010-01-08 X Protocol Version 11, Revision 0
Ah OK, I am on 1.9. And what is your card model?
It's an RV730. Now I have recompiled the whole xorg sources and the error is gone. BTW: WebGL is still not running natively on the ati card. My system reports the same Fx error and glxinfo as in Bug 673699. Is yours working?
I have RV710. I will check if WebGL is currently enabled (it gets better with each FF version). I can post my glxinfo too in 10 hours. Meanwhile, what do you get in the Graphics section of about:support ? See here for what you need: https://wiki.mozilla.org/Blocklisting/Blocked_Graphics_Drivers .
I get this: Graphics Adapter Description X.Org -- Gallium 0.4 on AMD RV710 Driver Version 2.1 Mesa 7.11-rc1 WebGL Renderer X.Org -- Gallium 0.4 on AMD RV710 -- 2.1 Mesa 7.11-rc1 GPU Accelerated Windows 0/1 So it seems WebGL is enabled.
Here WebGL is enabled when I start the kernel with KMS. Does this http://learningwebgl.com/lessons/lesson03/index.html test work? I have stripe and other patterns with the current Mesa (git master).
Yes, I use KMS permanently. It is better than UMS for me. The demo seems to work fine. Only the bottom line of the black "canvas" sometimes flashes with black/white stripes (1-2px high, 20px long). Otherwise the image seems correct. There is a blue rectangle and rainbow colored triangle, both rotating on black canvas.
Here with KMS the demo runs with Mesa 7.10.3 but it crashes Fx in demo 16 and probably in all of these samples: http://code.google.com/p/webglsamples/. With KMS I cannot have two X servers running on one card.
If 2 X servers on the same card means "startx -- :1" or "Start->Switch user->Start new session" in KDE 3, then that has always worked for me, also with KMS. Demo (Lesson 16) doesn't crash for me, the laptop is spinning there. From the demos on the linked page, I tried Blobs and Aquarium. Both seem to work. I get 10fps on Aquarium with 50 fish. I tested this on FF 10a1. Mesa 7.11. ATI driver 6.14.99 (git master from about 2 weeks ago). Jump from Mesa 7.10 to 7.11 was very good for me, the R600g (gallium) driver jumped from unusable (was using R600c because of that) to very good and enabled by default. I can only recommend it.
OK, one program that regressed in 7.11 was Google earth 6. That just crashes on startup. If you don't need that, then really try the new Mesa. I am on 32bit system, so I don't know if it all will work for you the same, but otherwise the system components and cards are similar.
(In reply to aceman from comment #23) > If 2 X servers on the same card means "startx -- :1" or "Start->Switch > user->Start new session" in KDE 3, then that has always worked for me, also > with KMS. In kdmrc I have enabled two "static servers". The second one does not (always) start, I have to restart kdm manually. It also happend that :1 was on vt9 instead of vt8. Interestingly there are two different durations when switching between the virtual terminals with ctrl-alt+n, n in {1, 7, 8}. (In reply to aceman from comment #24) > OK, one program that regressed in 7.11 was Google earth 6. That just crashes > on startup. If you don't need that, then really try the new Mesa. Mesa 7.11 and the git master both don't crash here (haven't tried Google earth yet) but show these stripes. I'll try to bisect the code ...
Shall I close this bug with WORKSFORME?
Oh, yes, if the drivers/X.org fully fixed it.
(In reply to aceman from comment #27) > Oh, yes, if the drivers/X.org fully fixed it. This is not yet the case. https://bugs.freedesktop.org/show_bug.cgi?id=42175
But comment 5 there (yours?) seems to say it is working now. Still, if the drivers cause this, mozilla can't do much about that. And also, this bug is about window title/border/controls, not about WebGL.
Component: General → Graphics
Product: Firefox → Core
QA Contact: general → thebes
Whiteboard: [CLOSEME 2011-11-15]
(In reply to aceman from comment #29) > But comment 5 there (yours?) seems to say it is working now. Sorry, forgot to close this here.
Status: UNCONFIRMED → RESOLVED
Closed: 14 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: