tech-x11 archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

re: Problems after xorg patch to 5.0_STABLE



On Tue, 22 Sep 2009, matthew green wrote:

this is probably the xkbdata -> xkeyboard-config upgrade issue.
we haven't yet implemented a real fix to help this upgrade as it
involves deleting an old subdir before extracting xbase.tgz.  the
simple workaround is to do this:

        # rm -rf /usr/X11R7/lib/X11/xkb/symbols/pc
        # pax -rzpe -f xbase.tgz ./usr/X11R7/lib/X11/xkb/symbols/pc

OK, this fixes the various VT-switching problems.  Except on the i810e
system, where switching to another VT kills the X server.  Killing it
with Ctrl-Alt-Backspace behaves normally and xdm re-launches it.  Haven't
tried Zoom with Ctrl-Alt-KPminus/Ctrl-Alt-KPplus yet.

I was mistaken about the intel video part.  Seems it's an ordinary
i810e (82810), not a 915 as originally stated.

I've collected relevant data on its remaining behavior in:

  http://bobdbob.com/~jdbaker/xorg16errs/i810/

The *.err files show the result of trying to use a configuration file
generated by 'X -configure'.  It's the same for 1.6.3 as for 1.4.2.
The configuration files generated by 'xorgconfig' (in the 1.4.2 system)
erroneously called the driver "i810", but changing it to "intel" made it
work quite well.

The i810e's other quirk is that xdm's login-widget is having trouble
with fonts.  It seems to be rendering the background to black and since
the foreground is also black, the result is black rectangles.  The login
failure message has a red foreground, so that shows up inside a black
rectangle (see the "*.png" files in that directory).

All machines have the same near-stock configuration.  Nothing related
to fonts was changed, but only the i810e machine is having this problem.


The SiS630 system information is in:

  http://bobdbob.com~jdbaker/xorg16errs/sis630/

"Xorg.0.log.old" is the result of using the as-probed "sis" driver.
The Xserver switches video mode and dies.  While one can switch VT's,
the video mode is not restored, so you end up typing blind at a text
console.

Changing the driver to "vesa" gets a working, but glacially-slow graphics
display.  Also, the "Xorg.0.log" file becomes HUGE.

Also includes are the configuration files generated by 'xorgconfig'.
The "sis" driver still kills the server and the "vesa" driver is still
glacially slow.  It was decent with 1.4.2 although it had little sparkly
glitches during disk/network I/O.


The Rage 128 system information is in

  http://bobdbob.com~jdbaker/xorg16errs/r128/

Its only problem at this point is the almost-black-on-black video
when switching away from the VT where the X server is running.


Actually, there's one problem I've seen in common with all of xorg since
the switch was made.  When switching BACK to the X server's VT, the
keystrokes aren't swallowed, but are instead delivered to any application
that happens to have focus when the X display is restored.

For example, with X on VT05 (the default), when I switch away, then
switch back, the XTerm that happened to have focus when the display was
restored beeps and shows a "5~" at the shell prompt.

I never saw this on XF86-4.x.  If there was a setting to tweak it
should have been made more obvious.

For the record this happens on OpenBSD too.


Thanks.

--
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645


Home | Main Index | Thread Index | Old Index