[SOLVED] startx causes black screen freeze after upgrade

johnraff · 2015-10-12 05:41:06

EDIT: in case anyone looks later, the issue was the wheezy-proposed-updates repository I had enabled. The new kernel it brought in broke X. Forcing apt-get to downgrade (and removing that repo) fixed it.
-----------------
This is a testing laptop with two partitions, running Statler and Waldorf. It hadn't been used for a while so still had the #! repos in sources.list of both systems till yesterday.

OK yesterday did the first update/upgrade for some time, forgetting that the #! repos were no longer online and to do the necessary preparations for life on Wheezy (remove tint2, tint2conf, comment out the #! repos). After a big upgrade and a reboot I ended up with:

black screen
no cursor
keyboard has no effect - even the SysRq stuff
only way out is the power button
nothing in ~/.xsession-errors or /var/log/Xorg.0.log

Statler's running fine, as was Waldorf yesterday morning.

I was using autologin + autostartx. After going back to a regular CLI login I can login OK and run commands on the tty with no problem, but if I run startx there's a flash of 5~6 lines (too fast to read) then the black freeze.

Tried:

delete ~/.Xauthority
make ~/.xinitrc (usually don't have one) with 'exec /usr/bin/alternatives/x-session-manager'
run 'startx -- -keeptty' (from here)
run 'X -configure' , copy /root/xorg.conf.new to /etc/X11 and edit out the extra displays

None helped.
I suspect the last one was a distraction - it brought up the error "Number of created screens does not match number of detected devices" and the created xorg.conf.new had three displays listed. I edited them down to one and put the file in /etc/X11 but no good.

I also looked at a lot of the "black screen" threads on #! but nothing clicked.

hwinfo --glxcard showed I have an Intel 945 GM , but since it was working fine yeterday, do I need to look for a driver?

The CLI system is working fine so I'll gladly post the output of any other commands you might suggest. (There's no copy/paste here though.) There's also a working Statler in the next partition.

Any suggestions gratefully received.

Last edited by johnraff (2015-10-13 04:26:27)

tknomanzr · 2015-10-12 06:05:46

 
lspci | grep VGA
lsmod

Check the kernel recognizes the card. Check to make sure intel modules are loaded into the kernel. You need xserver-xorg-video-intel. You can try deleting /etc/X11/xorg.conf and letting it rebuild on next boot.

Head_on_a_Stick · 2015-10-12 06:40:52

You don't need an .xinitrc as long as x-session-manager is set to "openbox-session" and you don't need an xorg configuration file with an Intel card.

Were any critical packages removed by the upgrade?

johnraff · 2015-10-12 09:09:48

Thanks for the feedback!

tknomanzr wrote:

lspci | grep VGA
lsmod

VGA compatible controller: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03)
lsmod is pretty long - what should I grep for? (I'll try googling on that meanwhile)

Check the kernel recognizes the card. Check to make sure intel modules are loaded into the kernel.

OK well i915 is in lsmod. That's the one, right?

You need xserver-xorg-video-intel.

Installed.

You can try deleting /etc/X11/xorg.conf and letting it rebuild on next boot.

There was no such file. I did try rebuilding one as posted above, but when it didn't work I deleted it again. It didn't rebuild automatically. I don't think modern systems have one by default.

johnraff · 2015-10-12 09:25:11

Head_on_a_Stick wrote:

You don't need an .xinitrc as long as x-session-manager is set to "openbox-session"

That's what I thought. When I did try putting one in the only difference it made was that I got a frozen cursor on the black screen.

and you don't need an xorg configuration file with an Intel card.

Ah OK.

Were any critical packages removed by the upgrade?

Looking in /var/log/apt/history.log there's a long list of packages that were upgraded last evening, but nothing removed. (It wasn't a dist-upgrade.) That was still in an X session, based on the update info from the previous day. Anything specific to grep for?

Following that, after the Black Freeze, in an enforced CLI mode, I commented out the #! repos which I'd forgotten about, updated and did another upgrade which triggered a dpkg error because of the tint2 and tint2conf I'd also fogotten to remove first. Removed them, reinstalled tint2 and it went OK.

I did also have the wheezy-proposed-updates enabled (just now commented out). I don't suppose some dodgy package could have got in from there?

johnraff · 2015-10-12 09:41:02

Update:
tried "NoAccel": https://wiki.archlinux.org/index.php/In … tel_driver
NoGood

Will also try "DRI" "False" but not holding my breath.
EDIT: no that didn't work either.

It's probably something simple and obvious...

Last edited by johnraff (2015-10-12 09:58:48)

Head_on_a_Stick · 2015-10-12 09:53:50

johnraff wrote:

I did also have the wheezy-proposed-updates enabled (just now commented out). I don't suppose some dodgy package could have got in from there?

Yes, that is a possibility.

That repository has a bad reputation over at forums.debian.net (never used it myself).

The kernel was updated from there recently:
http://ftp.debian.org/debian/dists/whee … 64.changes

What is the output of:

apt-cache policy linux-image-$(uname -r|sed 's,[^-]*-[^-]*-,,')

tknomanzr · 2015-10-12 12:44:46

Sorry, I wasn't sure what to grep for with lsmod. Video would have done the trick. Are you perchance using a display manager?

johnraff · 2015-10-13 02:47:17

@HoaS
linux-image-$(uname -r|sed 's,[^-]*-[^-]*-,,')
is
linux-image-686-pae
and apt-cache policy reports only one version 3.2+46 from wheezy/main

BUT
linux-image-3.2.0-4-686-pae
reports the installed version as 3.2.71-2 in /var/lib/dpkg/status only.
I've just commented out wheezy-proposed-updates so I'm assuming this version came from there.

The version in wheezy/updates/main is 3.2.68-1+deb7u4 so it looks as if going back to that kernel is definitely something to try.

In the past the grub menu used to keep old kernels, but I have only the one choice for this partition. Is it still available here do you think, or will I have to force a downgrade?

@tknomanzr no display manager on this system. Graphical (auto)login, which was working fine till the day before yesterday.

tknomanzr · 2015-10-13 03:27:52

Ok. I vaguely remember having to do some stuff with slim when I went this route oh so many months ago. I'd have to go back and check the walkthrough I posted, though.

johnraff · 2015-10-13 03:53:13

^The only change to this system was the apt-get update and upgrade I did two days ago. Up till then everything was fine, so it's looking as if it might have been that kernel upgrade from wheezy-proposed. slim was uninstalled long ago.

johnraff · 2015-10-13 04:21:53

That was it.
Forced the kernel (and linux-libc-dev) back to 3.2.68-1+deb7u4 and everything works again. smile

Thanks to both of you.

whalefish · 2015-12-12 00:06:54

could you post some instructions of how the downgrading was achieved from the black screen. I don't quite have the knowledge to implement these changes and am having difficulty researching how to do so.

johnraff · 2015-12-12 00:26:07

Hi whalefish - are you using CrunchBang Waldorf/Debian Wheezy with the wheezy-proposed-updates repository enabled? That's what was causing my issue.

If you're using BunsenLabs or Debian Jessie, could you start a new topic in "Basic Help and Support" with your problem?

whalefish · 2015-12-12 00:38:51

Yes i am using crunchbang Waldorf with the exact same issues after upgrading via repos after an install.

johnraff · 2015-12-12 00:48:38

You are using the "wheezy-proposed-updates" repositories?

What is your current kernel version number?
If you can get a tty and login (without X) then please post the output of

uname -v

whalefish · 2015-12-12 00:58:24

How can i find out if i am using the wheezy proposed repositories? I am not very experienced with using crunchbang so i am unsure.

Kernel is 3.2.0-4-686-pae

User@crunchbang:~$ uname -v
#1 smp debian 3.2.68-1+deb7u6

Last edited by whalefish (2015-12-12 00:59:50)

johnraff · 2015-12-12 02:11:47

OK then you're using the standard kernel, the same as on this machine I'm using right now with no problems, so I think your issue is different from what I had.
If you run

cat /etc/apt/sources.list{,.d/*}

in a terminal you will see what repositories you are using.

Anyway, I think you should start a new topic for your problem - you'll get better help that way.

#1 2015-10-12 05:41:06

[SOLVED] startx causes black screen freeze after upgrade

#2 2015-10-12 06:05:46

Re: [SOLVED] startx causes black screen freeze after upgrade

#3 2015-10-12 06:40:52

Re: [SOLVED] startx causes black screen freeze after upgrade

#4 2015-10-12 09:09:48

Re: [SOLVED] startx causes black screen freeze after upgrade

#5 2015-10-12 09:25:11

Re: [SOLVED] startx causes black screen freeze after upgrade

#6 2015-10-12 09:41:02

Re: [SOLVED] startx causes black screen freeze after upgrade

#7 2015-10-12 09:53:50

Re: [SOLVED] startx causes black screen freeze after upgrade

#8 2015-10-12 12:44:46

Re: [SOLVED] startx causes black screen freeze after upgrade

#9 2015-10-13 02:47:17

Re: [SOLVED] startx causes black screen freeze after upgrade

#10 2015-10-13 03:27:52

Re: [SOLVED] startx causes black screen freeze after upgrade

#11 2015-10-13 03:53:13

Re: [SOLVED] startx causes black screen freeze after upgrade

#12 2015-10-13 04:21:53

Re: [SOLVED] startx causes black screen freeze after upgrade

#13 2015-12-12 00:06:54

Re: [SOLVED] startx causes black screen freeze after upgrade

#14 2015-12-12 00:26:07

Re: [SOLVED] startx causes black screen freeze after upgrade

#15 2015-12-12 00:38:51

Re: [SOLVED] startx causes black screen freeze after upgrade

#16 2015-12-12 00:48:38

Re: [SOLVED] startx causes black screen freeze after upgrade

#17 2015-12-12 00:58:24

Re: [SOLVED] startx causes black screen freeze after upgrade

#18 2015-12-12 02:11:47

Re: [SOLVED] startx causes black screen freeze after upgrade

Board footer