[uae] Re: Pointer trails
- From: "Keith G. Robertson-Turner" <uae-freelists@xxxxxxxxxxxxxxxxxxxxxxx>
- To: uae@xxxxxxxxxxxxx
- Date: Tue, 09 Mar 2004 04:08:17 +0000
On Sat, 2004-03-06 at 19:13, Richard Drummond wrote:
> Hmmm. NewIcons doesn't seem to be a problem here. I had a feeling that the
> problem was somehow related to the way JIT accesses the P96 memory.
Well I think I've just confirmed your theory.
I've spent the last couple of days in testing, and here's my
conclusions:
First a description of the problem, of which there are three related
symptoms:
1) ... The pointer paints a permanent trail on the screen.
2) ... The Workbench backdrop is occasionally not drawn on startup, but
is subsequently "revealed" as objects are moved over it, rubbing through
to the image underneath.
3) ... Windows and Icons are permanently "painted" onto the desktop.
These problems are intermittent both between sessions and during
sessions, and can typically be temporarily "fixed" by opening and
closing a few application windows (client-side). This temporary fix
prevents further "sticky" objects occurring (for a brief period - then
the problem returns), but the screen does not automatically refresh and
remove existing corruption. Manually moving objects over the corrupt
areas "erases" them. Enforcer reveals nothing unexpected on client-side
operation.
The problem manifests on both the SDL and X11 builds, using a wide range
of uae configuration settings, and regardless of Kickstart (kick.rom)
version, Workbench libraries versions, setpatch version or flags, or any
client-side software, with the exception of P96.
At the exact point where the problem occurs, the following output
appears in the uae console:
"NATMEM: Failure to find mapping at <hex number>"
Repeated several times.
Using a non-JIT build of uae, removing P96 from the client, or running
RTG screenmodes less than 16bit, cures the problem.
The problem is 100% reproducible here, but not consistently reproducible
across different Linux environments (from what you and other have said),
so the question is ... what is different about my system? nptl, prelink,
or glibc maybe?
Link to my uaerc, uae-runtime.log, uae-strace.log, text dump of my RPM
database, and screenshots of the issues manifesting on uae:
http://www.genesis-x.nildram.co.uk/filez/uae/
Hope that's useful somehow ;-/
--
Regards,
K.
Other related posts:
