Re: gEDA-user: PCB+GL Testers (please test)

[Peter Clifton <pcjc2@xxxxxxxxx>]

On Wed, 2010-11-10 at 21:34 +0100, Frank Bergmann wrote:

> You find the backtrace at
> 
> http://www.frajasalo.de/frank/projekt/pcb/gdb-pcb.local_customisation_no_pours-backtrace-20101110-1.txt
> 
> I hope it will help, even without the debugging symbols in the system 
> stuff.

To really dig into it, I would need the debugging symbols for the DRI
driver. On Ubuntu/Debian, they are in libgl1-mesa-dri-dbg


>  In my former tests I only benchmark the program, so with deeper 
> looking at it, I found some unexpected behavior:

> On my AMD Mobility Radeon HD3450 (driver: radeonhd) system (debian, 
> stable but old!) and your local_customisation_no_pours branch (without 
> the recent patches) the line ends look wrong
> 
> http://www.frajasalo.de/frank/projekt/pcb/pcb.local_customisation_no_pours-20101110_amd4core.png

The pixel shader which draws the rounded line ends (and vias / holes
etc..) isn't working. Your hardware might not support that GL extension,
and I've been pretty lazy so far about writing code to detect that and
give proper errors.

In the long term, I expect to fall-back to the old rendering in the
"before_pours" branch.

Please send me the output of "glxinfo" from this machine, so I can see
what extensions are supported.

> And on my Intel system I have disturbances comming like rays from the 
> middle of the viewport/drawing area, see:
> 
> http://www.frajasalo.de/frank/projekt/pcb/pcb.local_customisation_no_pours-20101110_intel.ogv

Interesting.. I have no idea what is going on there. It could well be a
driver bug. Please let me know what mesa and kernel version you're
using.

I "THINK" I know what it might be, but I'm not certain. Please let me
know what graphics chipset you have. (Again, glxinfo output would help).
I just had a fix committed upstream to mesa master which gets around a
corruption when uploading data to the card. (An alignment issue).

Let me know what mesa version, and if the bug still applies, I'll see if
I can knock up up a patch. Are you on a Debian / Ubuntu system? (If so,
which version?)

> This effect appears also with before_pours branch - this make me 
> wondering because you are developing on an intel system if I understand 
> right.

Intel GM45 chipset here, with bleeding edge drivers. That is git HEAD
mesa + various patches, 2.6.37 kernel with backported drm-intel-next +
some local fixes ;)

In short.. my kernel / graphics drivers break more often than most
people would put up with, but I will have access to fixes for any known
bugs. (It isn't really practical debugging and submitting fixes without
tying current code first.) It is possible I never saw the issue as I've
been tracking these development drivers for quite a while.

Regarding this bug.. the colour looks like the errors are coming from
one of your polygon layers. Does it go away with certain layers
disabled?

What is the minimum board which exhibits the problem? Does it relate to
certain objects on the board, or is it general?
-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)
Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me)



_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user