You are not logged in.
I've just (since yesterday) started seeing a very odd font rendering problem in firefox:
Some things to note:
- this is reproducible with a new profile (moved ~/.mozilla) and no gtkrc (moved ~/.gtkrc-2.0);
- if I do anything to redraw the screen (e.g. scrolling a page down and back up) then the affect areas will change;
- from a quick check, it looks like this isn't a problem in chromium;
- I'm writing this post in a gvim window (as a child of firefox) and just briefly had the same problem here! Standalone gvim just spawned from a terminal seems fine but I guess this suggests a potential gtk issue.
Other info:
- running firefox 11.0-2;
- graphics driver is nouveau-dri 8.0.2-1;
- I'm using gtk-engine-murrine 0.98.1.1-3, but as noted above I was able to reproduce it without using the murrina theme.
Does anyone have any suggestions?
Moderator edit: The included image is too big. Reduced to link.
Last edited by bernarcher (2012-03-27 16:01:12)
Offline
same problem here.
It seems to be a cairo (related) bug, and probably cairo-xcb (from aur) specific.
See: https://bugs.freedesktop.org/show_bug.cgi?id=38904
Offline
After latest cairo-xcb update (1.12) from aur, i have same issues.
// cairo-git - nothing changed
/// 1.10.2-3 and everything seems ok
Last edited by herangr (2012-03-27 18:54:50)
Offline
Thanks for the suggestion folks - I'm also on cairo-xcb 1.12, so it's quite irritating if it's broken given that it's a dependency of awesome.
Has anyone found a decent workaround?
Last edited by pconnell (2012-03-27 13:35:06)
Offline
No help but same issue here...
Had this issue also with older versions of cairo-xcb and the awesome window bar, but far less frequently than with the last update.
Last edited by matse (2012-03-27 14:05:07)
Offline
I've worked around this by moving to cairo-ubuntu-xcb. This does involve replacing a bunch of the font libraries (e.g. libxft) with the Ubuntu patched versions - not quite as nice as infinality IMHO, but at least it works.
Offline
I noticed this issue on one single website and rebuilding cairo-xcb with --enable-xlib-xcb made it disappear. If you can confirm this, I will add this configure option to the AUR package.
Offline
@xduugu: Cannot confirm this, already built cairo-xcb with this option (and also rebuilt firefox afterwards manually) but with no success - firefox and awesome still have these "glitches".
Offline
I also experienced the issue, and not only in firefox. For me the solution was to move from nouveau to the nvidia proprietary drivers (something I wanted to do anyway).
Last edited by silenti620 (2012-03-28 07:24:26)
Offline
@silenti620: Switching to a proprietary driver is never a solution
Nevertheless you pointed me to a new direction and while experimenting with some xorg settings I found out that the (xorg.conf)-option:
Option "EXAPixmaps" "off"
Solves the issue for me (using open source radeon driver).
The manpage says
The default is on with > 32MB VRAM, off with < 32MB.
so I guess it should be good for something and I see it as a (hopefully) temporary workaround.
Greetings
matse
//Edit:
so i opened two bug reports (radeon and cairo, since I don't know whose fault it is):
radeon
cairo
Last edited by matse (2012-03-28 13:01:59)
Offline
@silenti620: Switching to a proprietary driver is never a solution
Nevertheless you pointed me to a new direction and while experimenting with some xorg settings I found out that the (xorg.conf)-option:Option "EXAPixmaps" "off"
I confirm this temporary fix. Working fine here
Offline
I confirm this temporary fix. Working fine here
I guess you are on radeon as well?
So there are several driver-specific bugs that share the same/similar symptoms...
Offline
Hmm yeah forgot to specify my driver. Open-source ATI (xf86-video-ati).
Offline
For me the solution was to move from nouveau to the nvidia proprietary drivers
Well, this is odd...
I am experiencing huge slow-downs in gvim 7.3 with cairo-xcb 1.12.0 and nvidia's proprietary drivers v. 295.33 (@matse: sadly, with fermi cards the nouveau driver isn't much of an option).
Is there a similar fix I could use? Currently I simply downgraded to cairo-xcb 1.10.2, which works just fine.
Offline
huge slow-downs in gvim 7.3 with cairo-xcb 1.12.0 and nvidia's proprietary drivers v. 295.33
Sounds like the old "server-side gradients" patch is still needed.
Get it from the ...debian.tar.gz link at the bottom of the Ubuntu cairo page. In the .tar.gz, it's in debian/patches dir.
Offline
Sounds like the old "server-side gradients" patch is still needed.
Actually, it looks like that's already being applied before building the cairo-xcb package.
Offline
brebs wrote:Sounds like the old "server-side gradients" patch is still needed.
Actually, it looks like that's already being applied before building the cairo-xcb package.
That patch is also included in the main repository's cairo since at least 1.12.0-1. Not sure if it was included before that.
Last edited by Draje (2012-04-02 19:17:09)
Offline
Last firefox (11.0-3) package was compiled without the "--enable-system-cairo" switch and this works fine with no glitches here even with enabled "EXAPixmaps" (radeon driver).
Since firefox was the main program where I noticed those glitches (sometimes, but not very often also in awesome) the problem is now more or less solved for me, how is it with other drivers for you?
Greetings
matse
Offline
Uli Schlachter has a patch here that (hopefully) re-enables system cairo build: https://bugzilla.mozilla.org/show_bug.cgi?id=722975 . I wanted to test it and make an AUR package, it seems that it compiles but I have not enough RAM to the linker...
Just use abs to download Arch's official Firefox 12.0 build files, modify mozconfig (uncomment --enable-system-cairo) and PKGBUILD (add patch -Np1 -i "$srcdir/system_cairo.patch" line where system_cairo.patch is Uli's attachment etc.), and it should work.
I would appreciate if someone could build an i686 package and put it to an unofficial repo. :-)
Offline
ngaba: I tried compiling firefox 12.0 as you suggested (on an x86_64 machine), but it segfaults upon starting.
Here is a backtrace; sorry for the missing symbols from libxul.so, I compiled firefox as described on the wiki, but it didn't help with that and I don't know what to do about it.
#0 0x00007ffff5da99ab in ?? () from /usr/lib/firefox/libxul.so
#1 0x00007ffff5da99f8 in ?? () from /usr/lib/firefox/libxul.so
#2 0x00007ffff5dbadc5 in NS_InvokeByIndex_P () from /usr/lib/firefox/libxul.so
#3 0x00007ffff529b9c9 in ?? () from /usr/lib/firefox/libxul.so
#4 0x00007ffff52a1319 in ?? () from /usr/lib/firefox/libxul.so
#5 0x00007ffff5343afa in ?? () from /usr/lib/firefox/libxul.so
#6 0x00007ffff533537c in ?? () from /usr/lib/firefox/libxul.so
#7 0x00007ffff5343bf5 in ?? () from /usr/lib/firefox/libxul.so
#8 0x00007ffff5344192 in ?? () from /usr/lib/firefox/libxul.so
#9 0x00007ffff5ee803c in js::ProxyHandler::call(JSContext*, JSObject*, unsigned int, JS::Value*) () from /usr/lib/firefox/libxul.so
#10 0x00007ffff5efc4e1 in js::Wrapper::call(JSContext*, JSObject*, unsigned int, JS::Value*) () from /usr/lib/firefox/libxul.so
#11 0x00007ffff5efcbdb in js::CrossCompartmentWrapper::call(JSContext*, JSObject*, unsigned int, JS::Value*) () from /usr/lib/firefox/libxul.so
#12 0x00007ffff5ee8733 in ?? () from /usr/lib/firefox/libxul.so
#13 0x00007ffff5ee879a in ?? () from /usr/lib/firefox/libxul.so
#14 0x00007ffff5343d22 in ?? () from /usr/lib/firefox/libxul.so
#15 0x00007ffff533537c in ?? () from /usr/lib/firefox/libxul.so
#16 0x00007ffff5343bf5 in ?? () from /usr/lib/firefox/libxul.so
#17 0x00007ffff53190ff in ?? () from /usr/lib/firefox/libxul.so
#18 0x00007ffff5343afa in ?? () from /usr/lib/firefox/libxul.so
#19 0x00007ffff533537c in ?? () from /usr/lib/firefox/libxul.so
#20 0x00007ffff5343bf5 in ?? () from /usr/lib/firefox/libxul.so
#21 0x00007ffff53039e4 in ?? () from /usr/lib/firefox/libxul.so
#22 0x00007ffff5343afa in ?? () from /usr/lib/firefox/libxul.so
#23 0x00007ffff533537c in ?? () from /usr/lib/firefox/libxul.so
#24 0x00007ffff5343bf5 in ?? () from /usr/lib/firefox/libxul.so
#25 0x00007ffff5344192 in ?? () from /usr/lib/firefox/libxul.so
#26 0x00007ffff53007ad in JS_CallFunctionValue () from /usr/lib/firefox/libxul.so
#27 0x00007ffff5295479 in ?? () from /usr/lib/firefox/libxul.so
#28 0x00007ffff529208d in ?? () from /usr/lib/firefox/libxul.so
#29 0x00007ffff5dbb9cf in ?? () from /usr/lib/firefox/libxul.so
#30 0x00007ffff5dbaee3 in ?? () from /usr/lib/firefox/libxul.so
#31 0x00007fffe618de20 in ?? ()
#32 0x0000000000000000 in ?? ()
Offline
donbex: Patched version works fine here with radeon drivers (x86_64), no idea about your missing symbols, do you use radeon drivers from official repos? Tried to remove your local firefox profile?
Offline
No, I use the nvidia closed source driver (I have a Fermi adapter). I also tried starting firefox from a fresh profile, although it seemed that firefox tried loading pentadactyl anyway (could be because I installed it system-wide with the AUR package).
Offline
oh damn... thought i'm in the radeon thread here ok no idea then, as i said works fine here with all plugins i use (pentadactyl is also one of them), you set the -g, -O1 and !strip flag in your PKGBUILD or globally in makepkg.conf?
Offline
[...] you set the -g, -O1 and !strip flag in your PKGBUILD or globally in makepkg.conf?
I set those flags in the PKGBUILD.
Offline
donbex: So would you pastebin it anywhere?
Offline