Bug 5584

Summary: Vertical lines in text (OOo, Wine) - RenderAccel on Radeon IGP345M?
Product: xorg Reporter: James Le Cuirot <chewi>
Component: Driver/RadeonAssignee: Xorg Project Team <xorg-team>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: high CC: ghepeu, michael.gaber
Version: 7.0.0   
Hardware: x86 (IA32)   
OS: Linux (All)   
URL: http://www.openoffice.org/issues/show_bug.cgi?id=48494
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Screenshot of vertical lines none

Description James Le Cuirot 2006-01-13 04:04:18 UTC
Hi. I'm getting strange vertical lines appearing in text. See the attached
screenshot. I've only seen it in OpenOffice (for a long time now) and the
CrossOver Office version of Wine (just once but I don't use it much). It tends
to affect certain letters like 'r' but you can also see it affecting some other
letters in the screenshot. The problem goes away when I set RenderAccel to false
so that must be where the problem lies. It's probably also worth mentioning that
X seems that little bit faster when RenderAccel is set to false. Is that because
of this bug? And does the fact that the Radeon driver cannot do accelerated
subpixel rendering have anything to do with it? I have subpixel rendering
enabled in my fontconfig.

I'm using Xorg 7.0.0 with 20051223 CVS snapshots of Mesa and the DRM modules. I
have a Radeon IGP345M.
Comment 1 James Le Cuirot 2006-01-13 04:05:11 UTC
Created attachment 4333 [details]
Screenshot of vertical lines
Comment 2 Francesco Biscani 2006-02-21 00:54:18 UTC
Yes! Me too! 
 
Actually I get other many random corruptions with my IGP340M (HP notebook).   
What I've discovered so far is that there seems to be a major b0rkage of the  
RenderAccel code, at least on this board. Configurations I've tested: 
 
- EXA+RenderAccel: visualization is fine in 24bit mode, but there are major  
performance drops. Even without shadows and transparent windows. Scrolling  
large webpages in konqueror & firefox hogs the system down, and after a while  
overall performance starts to suck (especially when switching workspaces). In  
16 bit mode it is a tiny bit faster, but I get the same corruptions as you.  
Anti-aliased text does not make any appreciable difference. 
 
- EXA+NoRenderAccel: it is WAY faster! Scrolling pages in web-browsers is  
on-par with XAA (i.e. blazingly fast). Of course shadows and especially  
transparencies are slower due to lack of acceleration, but I can live with  
that. At least xcompmgr -a is now usable ;) After a while performance drop a  
bit, but nothing compared to the previous case (I understand this may be and  
issue with EXA memory management Eric Anholt is tackling). 
 
- XAA+RenderAccel: speed is ok, but I get strange corruptions near  
anti-aliased text. For example with some letters there is a "smear" effect,  
which overlaps with nearby characters. And in konsole with white background  
sometimes pale grey lines appears over and below text. If text is not  
anti-aliased there are no corruptions. 
 
- XAA+NoRanderAccel: visually everything is ok. Most usable configuration as  
of now. 
 
So to summarize the issue seems strictly connected to RenderAccel, regardless 
of the acceleration method used. Can anyone look into this please? 
Comment 3 Francesco Biscani 2006-02-21 00:57:51 UTC
Oh, some missing info. 
 
lspci: 
 
01:05.0 VGA compatible controller: ATI Technologies Inc Radeon IGP 
330M/340M/350M 
 
Xorg is modular 7.0, but the problem is there since a long time (at least 
6.8). 
Comment 4 Giacomo Perale 2006-02-21 02:09:45 UTC
I had a similar problem, see https://bugs.freedesktop.org/show_bug.cgi?id=1424

I'm still appling the latest patch made by Kevin E. Martin here, on gentoo xorg
7.0 (xf86-video-ati 6.5.7.3) but I'm using only exa 
Comment 5 Michel Dänzer 2006-07-02 13:30:33 UTC
(In reply to comment #4)
> I had a similar problem, see https://bugs.freedesktop.org/show_bug.cgi?id=1424

The screenshot certainly looks like this. James, if you agree, please resolve
this as a duplicate of bug 1424.

> I'm still appling the latest patch made by Kevin E. Martin here, on gentoo xorg
> 7.0 (xf86-video-ati 6.5.7.3) but I'm using only exa 

The patches from that bug don't have any effect with EXA though.
Comment 6 James Le Cuirot 2006-07-04 15:01:18 UTC
I couldn't say for sure. I haven't seen anything like the first screenshot and 
I don't think I've seen it affect 'i' the way it has in the third screenshot. 
The 'R' in the third screenshot, however, is exactly what I get.
Comment 7 Michel Dänzer 2006-07-17 05:40:02 UTC
Yeah, the/one trouble with that bug is that it refers to at least two separate
issues... what a mess.
Comment 8 Timo Jyrinki 2007-02-22 14:28:11 UTC
Marking broken (status null/blank) bugs in xorg with no activity in a long time as fixed. Please reopen if you think it's necessary, but first do a search if a similar bug report is already filed and in a NEW/ASSIGNED state. These bugs do not currently show in most search results as they do not have any status.

Sorry for this janitorial spam, you know where to send hate mails to when your inbox gets full of bugs you're subscribed to.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.