Re: Starting X on NetBSD 9.0

[Tom <tgrom.automail%nuegia.net@localhost>]

On Fri, 13 Mar 2020 17:40:47 -0400
Greg Troxel <gdt%lexort.com@localhost> wrote:

> Clay Daniels <clays.shell%SDF.ORG@localhost> writes:
> 
> > I like the new NetBSD 9.0. It works great at the command line level,
> > and this is where I am writing this email using alpine. But I have
> > yet to get the X server to work. I've been trying to use the twm
> > window manager which I installed with pkgsrc, and have made
> > multiple attempts at creating an .xinitrc file that works. So I
> > could use some help.  
> 
> It is highly likely that your problem is not about twm.
> 
> > The odd thing to me is that when I type startx, it seems to try to
> > open xorg. I get appropiate color fuzz on the screen, and a display
> > with three faint and barely readable error messages, and oddest of
> > all, there is a partial copy of my MSI bios splash screen right in
> > the mddle of the page, and then it hangs and freezes up. I have to
> > shutdown with the power button as the ctl/F* does not switch
> > consoles. Anyway, I would appreciate any suggestions.  
> 
> That sounds like the X server is troubled.
> 
> I would advise that you set up networking and use some other computer
> to get a remote shell before you start X11, and use that to
> shutdown/reboot.   Or you can type "shutdown -r +10" (I think; read
> the man page) before you start.
> 
> THe next advice is to read the /var/log/X* logs.  Others will probably
> give more specific advice, but you may be able to disable some aspects
> of your graphics card or reduce acceleration.   It would help to post
> the dmesg lines about the card or from the log.
> 
> For instance, in dmesg I have
> 
>   radeon0 at pci5 dev 0 function 0: vendor 1002 product 68b8 (rev.
> 0x00)
> 
> (which is from 2014).

The msi bios screen post-boot comes from a security vulnerability
around VRAM. It's not as protected as the system memory and you can
pull images even across reboots by displaying uninitialized vram. Some
have even been able to pull images of the host machine from inside a
virtual machine when acceleration is enabled. This is part of the
reason the video devices in devfs require you to be part of a special
group. I have experienced similar things in the past with an Xorg
screenshot tool and using two monitors of differing resolution. The
areas that there was no monitor for was filled with corrupted images
and textures from various 2d and 3d accelerated software I was using
hours earlier.

-- 
 _____________________________ 
< When all else fails, EAT!!! >
 ----------------------------- 
\
 \
   /\   /\   
  //\\_//\\     ____
  \_     _/    /   /
   / * * \    /^^^]
   \_\O/_/    [   ]
    /   \_    [   /
    \     \_  /  /
     [ [ /  \/ _/
    _[ [ \  /_/