Dear Maintainer,

Today I did a fresh install of debian-8.2.0-amd64 on a Dell Latitude E6520
with an Nvidia NVS-4200M (N12P-NS2) graphics adapter.

After rebooting, the login screen briefly flashes on the screen and then the
 screen goes black.

STFW I see various similar bugs/issues, but not an exact match.

David

Your logs show that the nouveau kernel module loads too late, did you
blacklist it?  This is not going to work, see #700946.

In your setup the situation is especially bad, since the Xorg vesa
driver and the nouveau kernel driver both get loaded and fight over the
card.

Cheers,
       Sven

On 11/05/2015 10:11 AM, Sven Joachim wrote:

Thank you for your reply. :-)

No.

Correct.

So, the Debian project is shipping "stable" product with known Nvidia
DOA bugs OOTB?  That's going to affect a *lot* of computers.

Is there a Debian amd64 version that works OOTB on Nvidia?


David

You need to figure out why it's not loaded earlier by udev then.  This
works for almost everyone else.

It doesn't affect new installs at all, only upgrades from systems that
had non-free packages installed.

Jessie should work on most cards, but some are problematic (especially
old IGPs).

Cheers,
       Sven

No, thanks.


David

Fair enough, I won't bother you again then.  My apologies for
mistakenly thinking you would appreciate help.

Cheers,
       Sven

First, I apologize for my earlier complaint about shipping product with
known bugs.  Companies that refuse to disclose relevant technical
information needed for open-source software development inflict the
problem upon all of us.  And, Dell Latitude E6520 laptop computers like
mine have both Intel HD Graphics and an Nvidia graphics card, which is a
corner case:

https://en.wikipedia.org/wiki/Nvidia_Optimus


I *do* appreciate help.  But, I do not have the knowledge, skills, time,
or interest to "figure out why [the nouveau kernel module] is not loaded
earlier by udev".  If you would like to work together on this, I can run
commands for you, post files, etc., so you can debug it remotely.  Tell
me what you need.


David


p.s. Bumblebee is "a project aiming to support NVIDIA Optimus technology
under Linux":

http://www.bumblebee-project.org/


p.p.s. debian-7.9.0-amd64-xfce-CD1 works OOTB

Apologies accepted.

Thanks, that's important to know.  I'm not experienced with Optimus,
since I don't have any such hardware myself.

My theory was that the nouveau kernel module had been blacklisted, since
udev runs modprobe with the "-b" parameter, while the X server does
not.  But maybe I'm wrong.

Could you please start your system without X? On a standard Jessie
installation with systemd as init and gmd3 as display-manager, you can
add "systemd.unit=multi-user.target" to the kernel command line to
achieve that.  Log in at the console and see if the nouveau kernel
module is loaded ("lsmod | grep ^nouveau").  If not,

1. Run "modprobe -b nouveau" and check again.  If that does not load the
nouveau module,

2. Run "modprobe nouveau" and check again.

If the nouveau kernel module has been loaded in steps 1 or 2, you can
try to run X (start your display manager, or use startx).  Does it work
then?

Cheers,
       Sven

When I power-up, 'Debian GNU/Linux' is highlighted in the GRUB menu.  I
press 'c', get a prompt, and enter:

     grub> systemd.unit=multi-user.target

When I press <Enter>, GRUB gives me another prompt:

     grub>

When press <Enter> again, same as above.

If I let the machine boot by itself and then log in via SSH from another
machine:

     root@i72720qm:~# lsmod | grep ^nouveau
     nouveau              1122419  2


It appears that the nouveau kernel module is already loaded.


Please advise.


David

Sorry for the misunderstanding, this is not the way to append something
to the kernel commandline.  Please press 'e' rather than 'c' in the grub
menu and edit the line starting with "linux".

Instead of editing the grub menu, you can also run
"systemctl set-default multi-user.target" once the machine is up.  To
back out this change, just delete /etc/systemd/system/default.target.

Yes, because the normal boot will start the display manager and trigger
your original problem.

Cheers,
       Sven

Okay:

	linux	/vmlinuz-3.16.0-4-amd64 root=/dev/mapper/sda3_crypt\
  ro  quiet systemd.unit=multi-user.target


Then press Ctrl+x to boot.  I arrive at a text console.  Login as root:

     root@i72720qm:~# lsmod | grep ^nouveau
     nouveau              1122419  1


So, it appears that the nouveau kernel module is loaded.  Logout.  Login
as unprivileged user.  Start X Windows:

     toor@i72720qm:~$ startx


I arrive at the Xfce desktop.


When I log out, I arrive back at the text console where I left off.


So, now I have a work-around (that reminds me of ~10 years ago).


I would still like to get the graphical login manager working.


Please advise.


David

This is nice, but not consistent with the result in your original
report.  Can you please send the complete dmesg output?

Probably something like this would work (see modules-load.d(5)):

# echo nouveau > /etc/modules-load.d/nouveau.conf

This would hopefully load the nouveau kernel module early enough.
However, I still don't understand why this apparently isn't done by
udev.

Cheers,
       Sven

Attached.

That worked -- now it boots to the graphical login manager.


Anything else?


David

I see two notable things here:

- There is no sign of the i915 module being loaded, I guess the on-board
  graphics is disabled in the BIOS/UEFI?

- The boot seems to stall for 27 seconds.  That is probably unrelated to
  this bug, but if you want to find out why, systemd-analyze(1) is your
  friend.

No - I don't think I'll be able to solve the problem, but at least you
have an easy workaround.

Cheers,
       Sven

These are the only things I see in CMOS setup that seem to be related to
graphics:

   Settings ->
     General ->
       System Information ->
         Processor Information ->
           shows nothing about graphics
         Device Information ->
           Video Controller       = NVIDIA N12P-NS2
           Video BIOS Version     = 75.19.6A.01.02
           Video Memory           = 512 MB
           Panel Type             = 15.6" FHD
           Native Resolution      = 1920 by 1080
     Video ->
       Optimus ->
         Enable Optimus           = disabled

Let me try again, using a stopwatch:

     0:00 push power button
     0:07 POST displayed
     0:12 GRUB menu displayed
     0:17 GRUB runs
     0:23 LUKS prompt displayed
     1:00 LUKS passphrase entered
     1:08 login screen displayed
     2:00 login as toor
     2:02 desktop displayed

It looks repeatable:

[    4.966532]  sda: sda1 sda2 sda3
[    4.967006] sd 0:0:0:0: [sda] Attached SCSI disk
[    4.974416] sr0: scsi3-mmc drive: 24x/24x writer dvd-ram cd/rw
xa/form2 cdda tray
[    4.974419] cdrom: Uniform CD-ROM driver Revision: 3.20
[    4.974609] sr 1:0:0:0: Attached scsi CD-ROM sr0
[    4.975611] sd 0:0:0:0: Attached scsi generic sg0 type 0
[    4.975702] sr 1:0:0:0: Attached scsi generic sg1 type 5
[    5.067157] device-mapper: uevent: version 1.0.3
[    5.067287] device-mapper: ioctl: 4.27.0-ioctl (2013-10-30)
initialised: dm-devel@redhat.com
[   32.778218] random: nonblocking pool is initialized
[   34.848447] NET: Registered protocol family 38
[   35.074219] raid6: sse2x1    4711 MB/s
[   35.142138] raid6: sse2x2    7738 MB/s
[   35.210066] raid6: sse2x4   11132 MB/s
[   35.210067] raid6: using algorithm sse2x4 (11132 MB/s)
[   35.210068] raid6: using ssse3x2 recovery algorithm
[   35.210364] xor: automatically using best checksumming function:
[   35.250018]    avx       : 22887.000 MB/sec
[   35.252889] Btrfs loaded


Perhaps it is related to LUKS (?).

You solved the original problem -- the graphical login manager now
works.  Thank you.


David

Here is dmesg output for the above via SSH (so I can't screw it up):

root@i72720qm:~# dmesg
[    0.000000] Initializing cgroup subsys cpuset
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Initializing cgroup subsys cpuacct
[    0.000000] Linux version 3.16.0-4-amd64
(debian-kernel@lists.debian.org) (gcc version 4.8.4 (Debian 4.8.4-1) )
#1 SMP Debian 3.16.7-ckt11-1+deb8u5 (2015-10-09)
[    0.000000] Command line: BOOT_IMAGE=/vmlinuz-3.16.0-4-amd64
root=/dev/mapper/sda3_crypt ro quiet
...
[    4.980369] sd 0:0:0:0: [sda] 117231408 512-byte logical blocks:
(60.0 GB/55.
8 GiB)
[    4.980465] sd 0:0:0:0: [sda] Write Protect is off
[    4.980471] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[    4.980502] sd 0:0:0:0: [sda] Write cache: enabled, read cache:
enabled, does
n't support DPO or FUA
[    4.982171]  sda: sda1 sda2 sda3
[    4.983005] sd 0:0:0:0: [sda] Attached SCSI disk
[    4.990650] sr0: scsi3-mmc drive: 24x/24x writer dvd-ram cd/rw
xa/form2 cdda
tray
[    4.990659] cdrom: Uniform CD-ROM driver Revision: 3.20
[    4.991092] sr 1:0:0:0: Attached scsi CD-ROM sr0
[    4.991593] sd 0:0:0:0: Attached scsi generic sg0 type 0
[    4.991629] sr 1:0:0:0: Attached scsi generic sg1 type 5
[    5.089918] device-mapper: uevent: version 1.0.3
[    5.090119] device-mapper: ioctl: 4.27.0-ioctl (2013-10-30)
initialised: dm-d
evel@redhat.com
[   23.499859] random: nonblocking pool is initialized
[   44.129909] NET: Registered protocol family 38
[   44.360169] raid6: sse2x1    5295 MB/s
[   44.428098] raid6: sse2x2    8011 MB/s
[   44.496017] raid6: sse2x4   10909 MB/s
[   44.496018] raid6: using algorithm sse2x4 (10909 MB/s)
[   44.496019] raid6: using ssse3x2 recovery algorithm
[   44.496292] xor: automatically using best checksumming function:
[   44.535972]    avx       : 22491.000 MB/sec
[   44.538894] Btrfs loaded
...
[   48.495838] e1000e 0000:00:19.0: irq 42 for MSI/MSI-X
[   48.599608] e1000e 0000:00:19.0: irq 42 for MSI/MSI-X
[   48.599726] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[   51.534416] e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow
Control: Rx/Tx
[   51.534470] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[   56.574118] dell_wmi: Received unknown WMI event (0x0)

So, a couple of delays.


David

"In einer kurzen Einführung,

Ich bin ein Rechtsanwalt Meinze Klaus Peter, aus Deutschland, lebe ich
in London, habe ich Ihnen eine E-Mail über meine verstorbene Familie
geschickt, aber ich habe keine Antwort von Ihnen erhalten, der
Verstorbene ist ein Bürger in Ihrem Land mit dem gleichen Nachnamen
bei Ihnen

er ist Goldexporteur hier in London, mein Kunde ist vor ein paar
Jahren gestorben, als seine Familie sein Geschäft verließ und riesige
Geldbeträge von insgesamt 5,7 Millionen Pfund bei derUBS Investment
Bank hier in London deponiert wurden, ich bin der persönliche Anwalt
des verstorbenen Kunden,Ich brauche Ihre volle Kooperation, damit wir
das Geld von der Bank nehmen können, bevor die Regierung endlich doch
greift Ich werde erklären, mehr Details, wenn ich von dir höre


Gott sei mit dir??

Modtog du min sidste e-mail, jeg sendte til dig?

Lasse Boje Nielsen
Politidirektør Enhed For Særlig Kriminalitet