Re: Problems after xorg patch to 5.0_STABLE

[Jukka Salmi <j+nbsd%2009.salmi.ch@localhost>]

Hello,

not having followed development, pull-ups etc. for the last two months,
I built a i386 netbsd-5 release today and installed it on my ThinkPad
X40 to see whether the problem described below still exists.  It does,
except that now I see a black 'X' mouse cursor with a white outline on a
black screen after running startx(1).  While the mouse seems to be
working (I can move the cursor), my ~/.xinitrc seems not to be executed,
the screen stays dark, I can't kill the X server using
Ctrl+Alt+Backspace (yes, I've set the "DontZap" option to "false"),
switching to a VT (Ctrl+Alt+Fn) doesn't work, and sending a SIGKILL to
the Xserver leaves the display garbled.  So, more or less the same
problems as before...

When using the vesa(4) driver instead of intel(4), everything works
fine (and slowly...).

Any hints about how to proceed?  Try the latest intel driver (seems to
be 2.9.1 ATM)?  Try with the intel driver used before the Xorg pull-ups
(2.4.0)?  Something else?

Help is appreciated.

TIA, Jukka

Jukka Salmi --> port-i386 (2009-09-21 19:36:50 +0200):
> John D. Baker --> port-i386 (2009-09-20 19:57:30 -0500):
> > I updated my sources this weekend to pick up the xorg patch in 5.0_STABLE
> > and a number of things don't work right.  I've so-far installed on three
> > different i386 platforms w/different video devices.
> > 
> > Generic/OEM ATI rage128-32MB (r128 or ati driver)
> > Intel i915 w/8MB? (intel driver)
> > SiS 630AGP/shared-memory (16MB) (vesa driver)
> > 
> >   Can't switch vt's when X server is running (all systems)
> > 
> >   SiS/vesa system is glacially slow to paint the screen (you can watch
> >   it draw).  It used to be quick/normal.  The SiS driver caused X to die
> >   and leave the console unusable prior to the patch, so that's why the
> >   vesa driver is being used.  SiS driver worked fine in XF86-4.x.
> > 
> >   i915 system shows black squares for login/password prompt (and username)
> >   and status/error message appears in black box on xdm login widget.
> > 
> >   r128 system leaves text console in nearly-black-on-black when X server
> >   exits (or back when vt's could still be switched).  It was doing this
> >   before the patch as well, but I was hoping this would have been fixed.
> > 
> > All systems are still using the "/etc/X11/xorg.conf" file that was tailored
> > for them before the patch.
> > 
> > I tried generating one with "X -configure" on the i915 system, but it
> > failed.  I'll make note of the error when I get access to the machine
> > again--I ran out of time this weekend.
> > 
> > 'xorconfig' is missing after the patch.  Is this deliberate?  It was
> > the only way to get a working configuration file on these machines.
> > 
> > Sorry for not including data.  I won't have access to those machines
> > again until next weekend.
> > 
> > Has anyone else had problems like these?  I didn't see any reports after
> > the request for tests (in which I couldn't participate--having no
> > experimental machines at present).
> 
> I just updated a netbsd-5 i386 system (ThinkPad X40) and I'm seeing the
> problems you describe for the intel driver: black screen as soon as I
> run startx(1), can't switch to another VT afterwards.  Logging in via
> ssh and sending X a SIGKILL just leaves the display garbled.
> 
> And yes, xorgconfig indeed seems not to be included in the sets;
> pcitweak, scanpci, in[blw], out[blw] and ioport are also missing.
> 
> I've made [1]old (before the recent Xorg related pullups) and [2]new
> (todays netbsd-5 sources) Xorg.0.log files from the system in question
> available, in case anybody wants to have a look at this problem.
> 
> BTW, `X -configure' succeeds here, but starting X with the resulting
> config file shows the problem described above...
> 
> 
> Regards, Jukka
> 
> [1] http://salmi.ch/~jukka/nbsd/netbsd-5-Xorg/Xorg.0.log.old
> [2] http://salmi.ch/~jukka/nbsd/netbsd-5-Xorg/Xorg.0.log.new
> 
> -- 
> This email fills a much-needed gap in the archives.

-- 
This email fills a much-needed gap in the archives.