Color transform in pages under Opera (v12.00 alpha build 1174) Linux (Ubuntu 10.04 Xfce)

I have a very strange problem which I wondered if anyone knows about.

I was running Opera v11.52 happily until yesterday; no problems. I then was notified that v11.60 was available. I upgraded (via deb.opera.com). v11.60 was very unstable and crashed consistently within a few seconds of starting; sometimes sooner, sometimes later, but always within say 60 seconds of start. The crash reason was (consistently) an out-of-bounds memory reference.

I wanted to roll back to v11.52, but found that 11.52 was not available anymore on deb.opera.com. Ubuntu's default configuration (apparently) did not leave the 11.52 install package remaining on my machine for rollback purposes.

I saw that v12.00 was available on deb.opera.com, and so I thought to try out 12.00 to see whether the problem still existed there. I installed v12.00 and found that it's quite stable. It doesn't crash (for me, so far, with about 12 hours of use).

However, although v12.00 is usable, I see a very strange problem. In my rendered pages (never mind the Opera interface), it appears that the colors are transformed such that the red gun value is being applied to the blue gun, and vice versa. That is, items that should be solid red are solid blue (e.g. The Economist logo), and items that should be solid blue are solid red, and, e.g., JPEGs with blue sky in them have brownish-yellow sky, and so forth. I don't really know whether the green gun is affected, but I suspect not. It's quite peculiar. (By the way, the Opera interface also is affected this way, but that doesn't really matter to me.)

It occurred to me that this might be a deliberate tweak to make it immediately obvious that an alpha version was being run. I've done stuff like this in the past when working on development software.

Anybody know anything about this, and what I might do to cope with it?

Thanks all.

ruario said: "With regards to the blue you see on startup, it might be related to hardware acceleration, which is available in Opera 12.00 but is in an early state. You could try disabling it and see if it goes away. To do this set opera:config#UserPrefs|EnableHardwareAcceleration to 0, click Save and restarting Opera."

ruario, I did this, but no change resulted. I confirmed that the new setting is zero. I guess hardware acceleration isn't the problem.

I will investigate your other suggestion regarding v11.60 later today. Thanks for your help.

ruario said: "Try running Opera 11.60 with a clean profile with the profile directory switch as follows:
opera -pd /tmp/opera_test_profile &
When run this way does Opera 11.60 still crash on startup?"

OK, I made a fresh install of v11.60. I still have an Opera v11.52/11.60 profile directory in its usual place (in .opera off my home directory), but if I understand you correctly that doesn't matter.

Immediately after installing v11.60 (this time), I opened a shell window and issued "opera -pd /tmp/opera_test_profile &". Within a few seconds, I got the following output on that shell window:

[1] 29930
<my account name>@<my machine name>:~$ opera [crash logging]: CRASH!!
/usr/lib/opera/opera got signal SIGSEGV at address 08962B57

Log was created here:
/var/tmp/crash20111208151900.txt

An opera crash report window also opened, but was visually corrupted and I couldn't manage to fill in its fields with useful information. I did say "Send Report" or whatever the button said to that effect, but I don't know what action might have resulted from that. It is possible that the report might not make it out of my system due to my proxy and firewall situation.

I went to look at /var/tmp/crash20111208151900.txt, but it does not exist. I am even more puzzled than before.

Any ideas?

ruario said: "Can I have the output from the following command:

xdpyinfo


Could you also run the command:

xwininfo


and then click on the Opera window and give me that output as well."

OK, here's the output of xdpyinfo:

name of display: :0.0
version number: 11.0
vendor string: The X.Org Foundation
vendor release number: 10706000
X.Org version: 1.7.6
maximum request size: 16777212 bytes
motion buffer size: 256
bitmap unit, bit order, padding: 32, LSBFirst, 32
image byte order: LSBFirst
number of supported pixmap formats: 7
supported pixmap formats:
depth 1, bits_per_pixel 1, scanline_pad 32
depth 4, bits_per_pixel 8, scanline_pad 32
depth 8, bits_per_pixel 8, scanline_pad 32
depth 15, bits_per_pixel 16, scanline_pad 32
depth 16, bits_per_pixel 16, scanline_pad 32
depth 24, bits_per_pixel 32, scanline_pad 32
depth 32, bits_per_pixel 32, scanline_pad 32
keycode range: minimum 8, maximum 255
focus: window 0x5000004, revert to Parent
number of extensions: 28
BIG-REQUESTS
Composite
DAMAGE
DOUBLE-BUFFER
DPMS
DRI2
GLX
Generic Event Extension
MIT-SCREEN-SAVER
MIT-SHM
RANDR
RECORD
RENDER
SECURITY
SGI-GLX
SHAPE
SYNC
X-Resource
XC-MISC
XFIXES
XFree86-DGA
XFree86-DRI
XFree86-VidModeExtension
XINERAMA
XInputExtension
XKEYBOARD
XTEST
XVideo
default screen number: 0
number of screens: 1

screen #0:
dimensions: 1152x864 pixels (322x241 millimeters)
resolution: 91x91 dots per inch
depths (7): 16, 1, 4, 8, 15, 24, 32
root window id: 0x63
depth of root window: 16 planes
number of colormaps: minimum 1, maximum 1
default colormap: 0x20
default number of colormap cells: 64
preallocated pixels: black 0, white 65535
options: backing-store NO, save-unders NO
largest cursor: 64x64
current input event mask: 0x7a802c
ButtonPressMask ButtonReleaseMask LeaveWindowMask
ExposureMask StructureNotifyMask SubstructureNotifyMask
SubstructureRedirectMask FocusChangeMask PropertyChangeMask
number of visuals: 17
default visual id: 0x21
visual:
visual id: 0x21
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 8 bits
visual:
visual id: 0x22
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 8 bits
visual:
visual id: 0x54
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x55
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x56
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x57
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x58
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x59
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5a
class: TrueColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5b
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5c
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5d
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5e
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x5f
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x60
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x61
class: DirectColor
depth: 16 planes
available colormap entries: 64 per subfield
red, green, blue masks: 0xf800, 0x7e0, 0x1f
significant bits in color specification: 6 bits
visual:
visual id: 0x43
class: TrueColor
depth: 32 planes
available colormap entries: 256 per subfield
red, green, blue masks: 0xff0000, 0xff00, 0xff
significant bits in color specification: 8 bits

And here's the output of xwininfo, clicking on the Opera v12.00 window:

xwininfo: Please select the window about which you
would like information by clicking the
mouse in that window.

xwininfo: Window id: 0x4600005 "Color transform in pages under Opera (v12.00 alpha build 1174) Linux (Ubuntu 10.04 Xfce) - Opera for *nix - Linux/FreeBSD - ... - Opera Next"

Absolute upper-left X: 0
Absolute upper-left Y: 42
Relative upper-left X: 0
Relative upper-left Y: 22
Width: 1152
Height: 796
Depth: 16
Visual: 0x21
Visual Class: TrueColor
Border width: 0
Class: InputOutput
Colormap: 0x20 (installed)
Bit Gravity State: NorthWestGravity
Window Gravity State: NorthWestGravity
Backing Store State: NotUseful
Save Under State: no
Map State: IsViewable
Override Redirect State: no
Corners: +0+42 -0+42 -0-26 +0-26
-geometry 1152x796+0+20

Now, given all this information, I hope it confirms what you surmised and wrote about in your next post after the one I responded to above. I see "Depth: 16", which I presume must be the "running X in 16-bit colour mode" you were talking about.

I would like to try your suggested workaround of "switch to 24-bit", but I am somewhat unfamiliar with X configuration. The research I've done online is somewhat contradictory and confusing, and doesn't match well to what I actually see in my machine. Perhaps I am several versions of X beyond what some of the authors were writing about. Are you familiar enough with X configuration to provide specific guidance? I do apologize and I do realize that it is not your job to educate random users about the internals of their systems. Never the less, I do hope you can point me in the right direction, and I thank you for whatever effort you choose to provide.

I, markblattner, said: "I would like to try your suggested workaround of "switch to 24-bit", but ...".

I have, after some experimentation and definite error after trial, inserted a "DefaultDepth 24" line into the "Section "Screen"" section of my /etc/X11/xorg.conf file. It does work, though I see some rather worrying video effects on my monitor. I am willing to risk my (cheapo) monitor; I have another identical unit standing by if needed.

Anyway, ruario, if this is what you would have suggested, then I suppose I am successfully working around the color problem.

If you are still willing to pursue the v11.60 crash problem, I would like to pursue it too, and will cooperate however you direct.

ruario said: "could you expand on that?" referring to my statement that "I see some rather worrying video effects on my monitor" after I "inserted a "DefaultDepth 24" line into the "Section "Screen"" section of my /etc/X11/xorg.conf file".

I see, systemwide, not confined to Opera, sparkling pixels arranged in vertical lines the entire height of the display. The effect is most prominently seen on my desktop, which has a predominantly dark blue background consisting of a .png image that has been color- brightness- and saturation-adjusted ("on the fly"). When I am in an application window such as Opera, the effect is mostly not noticeable, perhaps due to white or light-colored backgrounds being prevalent. It is possible that the effect is particular to image (e.g. PNG, JPEG, etc.) rendering, as I have noticed it occurring inside news photographs but not generally elsewhere. I never see it in title bars, icons, or similar graphic elements.

My screen is about 325 mm wide, and the sparkly lines appear at quite regularly spaced intervals of about 6 mm. The lines vary in terms of how many pixels vertically appear to sparkle; some lines have almost all pixels sparkling, while others have only one pixel sparkling every 4.5 mm vertically. The effect, in total, is quite geometric and not random-looking at all. Toward the left and right edges of the screen there are many places that one would expect to see a sparkly line, where no such line is evident (that is, no sparkly pixels).

I was worried about possible hardware damage to my monitor because X documentation specifically warns about this possibility, if one tells X that a monitor can handle frequencies and such that it cannot in fact handle. My fear has subsided a bit, though, given that I have now subjected my monitor to this signal for about 16 hours and no apparent damage has occurred.

By the way, now that the color situation is temporarily resolved, I went back and turned on the "Enable Hardware Acceleration" in opera:config. It caused the title bar of the opera window to become garbled and unreadable, but the rest of the opera window was not affected. I didn't see any great improvement in rendering speed, so I went back and turned off the "Enable Hardware Acceleration". At this point I do seem to have a workable system, and I can live with the sparkly lines given that it gets me proper colors in my browser.

I notice this morning that deb.opera.com reports "Changes for the versions: 12.00.1174 12.00.1191" (new versions are available). How will I know when a version appears that would run correctly with 16-bit color? Is there some sort of development log or change log or something that I could check every day or so, watching for such a change? I automatically monitor many web pages every day, and I could easily add one or several at opera.com if that would be the way to become aware of the fix I need. If you intend to post to this forum thread when this specific fixed version is available, of course that's good too; I'm already monitoring this...

Thanks again, ruario, for the quick help you've given me.