Re: X11 configuration

[Michael <macallan%netbsd.org@localhost>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Apr 10, 2009, at 8:24 PM, T. M. Pederson wrote:

> Morning,
>
> On Fri, 10 Apr 2009 20:15:26 -0400
> Michael <macallan%netbsd.org@localhost> wrote:
> [...]
> > On Apr 10, 2009, at 6:12 PM, T. M. Pederson wrote:
> >
> > > > if you just start the Xserver itself, then start an xterm via ssh  
> > > > and
> > > > close it again?
> > >
> > > Closing the xterm gets an (almost) immediate lockup. I say  
> > > "almost" as
> > > the pts in which I'd launched xterm had enought time to follow up  
> > > with
> > > the usual prompt before everthing went tango uniform.
> > >
> > > Or, to be explicit:
> > > oasis$ xterm
> > > oasis$ Connection closed by foreign host.
> >
> > Ok, so it's the Xserver exiting.
>
> I take it closing the xterm (I typed exit at the xterm's prompt),
> closes X, too?

Yes, when the last client exits the Xserver will either exit or restart.

> > > > Try this before startx:
> > > > sysctl -w ddb.commandonenter=reboot
> > > > If it's a panic that should reboot the machine and hopefully  
> > > > preserve
> > > > the panic message in the dmesg buffer.
> > >
> > > Oooh. I'd been completely overlooking this (in spite of recent  
> > > threads
> > > elsewhere mentioning it). Thanks!
> > >
> > > Although, sysctl.conf already has ddb.onpanic=0, which I'd expect to
> > > take precedence?
> >
> > I'm not sure, I never used that - please remove it, maybe it has some
> > funny side effects like suppressing the panic message. Maybe the mini
> > gets stuck trying to reboot which wouldn't be too unusual given the
> > circumstances.
>
> That wouldn't surprise me either.
>
> > > Hmm, this gets nothing: The dmesg buffer just has the kernel boot
> > > lines.
> >
> > So it did reboot with this?
>
> Unfortunately, no. It got the same hang, needing a power cycle to  
> clear.

There goes the dmesg buffer.

> > I forgot the syntax to cram more than one command in there - maybe  
> > try
> > 'bt;reboot" hoping that would give us a backtrace?
>
> That looks promising. Although it'll be a bit before I can give that a
> shot.
>
> > Btw. are you using genfb or radeonfb now?
>
> genfb. I'm still using the 5.0_RC3 GENERIC.


Please try radeonfb just to see if it makes any difference.

I have another idea. Please run pcictl pci0 dump -d <whatever device  
number your radeon is on, pcictl pci0 list should tell you> before  
starting X, then start X like before, just with an xterm, run the same  
command and look for differences in the BARs. If the Xserver feels  
like messing with the BARs then whatever console driver we use would  
cause a panic immediately when trying to write video memory. I don't  
really see evidence of this but who knows.

have fun
Michael

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBSd/od8pnzkX8Yg2nAQKuEgf/W3QgrHcB5l8vidvj+NZabeupr7BPd6a8
MO2F+3f6J/Pe5OfT+UgtAB72gPM40KGiYbHjnM59wHHpkDkUxjh1IZbMtiezySOJ
t8PQm7MnqLemEgyKODAjRooVX5UxnatVWd5TCV7L2fVR21mu1eoLG7mAOSupQUUg
sxml0/2HKl5gBTky1bAX0XaoLlM+caOmdcv2EB5hIgn8TPePzfQCmlrSraWfihrF
sBV41RMFJ/oVTE586lnbUh4askQf/GjkgqwQdXRxt3T5LYQ/F1iyN94kOgqjkSzc
D/ix4kl9sAykk3ViXCFmgxYpkg3YiKEhjyhD/fs521VQKNl6a9fnmw==
=Gx/y
-----END PGP SIGNATURE-----