[Intel-gfx] Tiling problems on i945GM

[Colin Guthrie]

Ping! Anyone here got any thoughts on this or should I go bug the
gnome-shell/mutter people on the assumption that it's compositing related?

Col

'Twas brillig, and Colin Guthrie at 23/08/11 12:56 did gyre and gimble:
> Hi,
> 
> I've been suffering with this problem for a while now, but can't
> specifically remember when it was introduced, but I'm currently using:
> kernel: 3.0.3
> mesa: 7.11
> libdrm: 2.4.26
> x: 1.10.4
> intel: 2.16.0
> 
> 
> The problem presents itself in a couple different ways:
>  1. some tiling issues in some apps (notably firefox when scrolling) [1]
>  2. Not updating the screen generally, e.g. click on an email in
> thunderbird and not have the preview pane update to the new email
> contents (although I know it's just a screen update issue - the email
> *has* been rendered - just not displayed)
> 
> I suspect this problem relates actually more towards some compositing
> problems (perhaps damage related) rather than a driver problem, but as
> I've not heard anyone else complain about compositing issues, I figure
> it should start my investigations here!
> 
> I am currently using gnome-shell (well, mutter I guess) for compositing
> (3.1.5).
> 
> 
> Some other points worth noting:
> 
> 1. I don't seem to get quite the same problem using kwin for compsiting
> under a KDE desktop.
> 2. Under gnome-shell, if I run gnome-terminal, it's quite often
> noticeable (type a command and the previous output is not scrolled up).
> Running knosole instead doesn't seem to have the same problem.
> 3. Connected to the above, the problem seems more apparent in gtk apps -
> can't recall seeing problems under Qt. Libre office is also affected
> tho' - whole buttons disappear a lot in e.g. the ooimpress wizard shown
> on launch. I think Libreoffice is using gtk... don't really know it well
> tho'.
> 4. Whenever any of this happens, triggering a screen refresh "fixes"
> things. e.g. pressing the "win" key in gnome shell will zoom out and
> back in again and the screen contents are fixed. Even taking a
> screenshot fixes things (which makes it a pain to gather evidence!)
> 
> 
> I appreciate this could very easily NOT be a driver issue, so consider
> this my first port of call in the debug trail. The <suck up>very clever
> and intelligent and enlightened</suck up> people on this list are the
> obvious choice to ask for advice :D
> 
> Cheers
> 
> Col
> 
> 
> 
> 
> 
> 1. http://colin.guthr.ie/screen-tiling-issue.jpg
> 
> 
> 
> 
> 


-- 

Colin Guthrie
  http://colin.guthr.ie/