[GM45] [upgrade] [regression] [lucid]: external monitor only usable via DVI, not via DP

[Matthias Andree]

Binary package hint: xserver-xorg-video-intel

I have a Dell Latitude E6400 with docking station on a Dell 2408WFP monitor. This used to be connected via Display Port "DP1" in Ubuntu Karmic and worked to use the monitor's preferred 1920x1200 16:10 mode.

After upgrade to Lucid, the DP1 connection no longer works. The monitor is detected properly, listed in display-settings or xrandr -q, but enabling it via display-settings or xrandr --output DP1 --preferred (or xrandr --output DP1 --auto) does not wake the monitor from energy save, it remains off.

Workaround: connect monitor to DVI port (listed as HDMI2 in xrandr) and configure that.

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: xserver-xorg-video-intel 2:2.9.1-3ubuntu1
ProcVersionSignature: Ubuntu 2.6.32-19.28-generic-pae 2.6.32.10+drm33.1
Uname: Linux 2.6.32-19-generic-pae i686
Architecture: i386
Date: Wed Apr 7 18:54:27 2010
DkmsStatus:

EcryptfsInUse: Yes
MachineType: Dell Inc. Latitude E6400
PccardctlIdent:
 Socket 0:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.32-19-generic-pae root=UUID=1a2166e9-550c-4a0a-8d3d-827d2f623a60 ro quiet nomodeset splash
ProcEnviron:
 LANGUAGE=en_GB:en
 LANG=de_DE.utf8
 SHELL=/bin/bash
SourcePackage: xserver-xorg-video-intel
dmi.bios.date: 12/21/2009
dmi.bios.vendor: Dell Inc.
dmi.bios.version: A20
dmi.board.name: 0W620R
dmi.board.vendor: Dell Inc.
dmi.chassis.type: 8
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.:bvrA20:bd12/21/2009:svnDellInc.:pnLatitudeE6400:pvr:rvnDellInc.:rn0W620R:rvr:cvnDellInc.:ct8:cvr:
dmi.product.name: Latitude E6400
dmi.sys.vendor: Dell Inc.
glxinfo: Error: [Errno 2] No such file or directory
system:
 distro: Ubuntu
 codename: lucid
 architecture: i686
 kernel: 2.6.32-19-generic-pae

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile 4 Series Chipset Memory Controller Hub [8086:2a40] (rev 07)
     Subsystem: Dell Device [1028:0233]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller [8086:2a42] (rev 07)
     Subsystem: Dell Device [1028:0233]

[Matthias Andree (matthias-andree)]

This is the xrandr -q --verbose output when the monitor is attached to the display port rather than the DVI-D (HDMI) output.

[Robert Hooker (sarvatt)]

Does this happen if you don't boot with nomodeset as well? Did you use nomodeset in karmic when it worked?

[Matthias Andree (matthias-andree)]

I did not use nomodeset on Karmic. On Lucid, this GRUB option on the kernel line makes no difference.

[Robert Hooker (sarvatt)]

Can you attach your dmesg, xrandr -q --verbose output and /var/log/Xorg.0.log after booting without nomodeset please?

[Matthias Andree (matthias-andree)]

I've updated all packages with aptitude a couple of minutes ago (~ 15:30 UTC on Wed Apr 14th) and this seems to have fixed the issue (at least if I boot with the laptop docked & opened). Desired files attached.

[Matthias Andree (matthias-andree)]

Further data point, but currently inexplicable, a similar issue haunted Fedora 12 on the same hardware, but not Windows 7, but now has disappeared on either Linux system after rearranging (i. e. replugging the DP cable), but never existed on Windows.

This may mean that signalling between what X.org does and what Windows configures might be subtly different -- so subtle it doesn't break DDC data or WIndows display, but big enough to make a difference to X.org-based display.

[Chris Halse Rogers (raof)]

From your comments it seems that this has fixed itself? I'll mark the bug as closed for the moment, but if it comes back feel free to reopen it.

[Matthias Andree (matthias-andree)]

Am 16.04.2010 07:51, schrieb Chris Halse Rogers:
>>>From your comments it seems that this has fixed itself? I'll mark the
> bug as closed for the moment, but if it comes back feel free to reopen
> it.
>
> ** Changed in: xserver-xorg-video-intel (Ubuntu Lucid)
> Status: Confirmed => Invalid
>

I'm not saying that related bugs are fixed, but at least there isn't a symptom
currently - I know this effectively voids most debugging possibilities. :-(

--
Matthias Andree

[Chris Halse Rogers (raof)]

Yeah. We can't test fixes for a bug that you can't reproduce :).

Again - if you later find you *can* reproduce this, please reopen the bug.

[Matthias Andree (matthias-andree)]

Reopening, problem is back 1 day after official Lucid release. Forcing modprobe option i915 modeset=1, or booting with "nomodeset" make no difference.

Apparently EDID is detected properly and XRandR believes the display to be enabled, yet the monitor remains in power save. Switching from DP1 to DVI1 works, as well as dual-booting into Windows 7.

[Matthias Andree (matthias-andree)]

All the logs in #14 to #17 were obtained with xserver-xorg-video-intel 2:2.9.1-3ubuntu5

[Matthias Andree (matthias-andree)]

OK, back working, but it's unclear why. The dmesg diff is about different timing, clock calibration, device detection order, and hence omitted.

I see the allocated screen in Xorg.0.log is now much smaller, and gdm already enables the cloned display; there's also a related xrandr difference.

I tried booting with i915.modeset=0 which made matters worse, in that the DP outputs were no longer detected at all and prior to mode switching there was garbage, and screen flicker on reboots between linux kernel shutting down and the BIOS reinitializing the gfx chip.

What I also did in between was:

# gconftool-2 --recursive-unset /apps/gnome_settings_daemon/xrandr /apps/gnome_settings_daemon/plugins/xrandr /apps/gdm/simple-greeter/settings-manager-plugins/xrandr

Current state:

gconftool-2 -a /apps/gnome_settings_daemon/xrandr
 show_notification_icon = true
gconftool-2 -a /apps/gnome_settings_daemon/plugins/xrandr
 priority = 1
 active = true
gconftool-2 -a /apps/gdm/simple-greeter/settings-manager-plugins/xrandr
 priority = 1
 active = true

Which parts of the overall system would save state from previous (possibly undocked) boots that might spoil the next startup?

[Matthias Andree (matthias-andree)]

Further observation (because this is so mysterious, but this phenomenon seems plausible):

The splash screen appears on both monitors now; whereas it only appeared on LVDS1 when DP1 wasn't working. Might be related, but it's unclear still.

Might it be that some registers of the gfx chips are not reliably initialized?

Are there any plymouth logs I could look at or enable?

[Chris Halse Rogers (raof)]

The way to get useful information out of the intel kernel driver is to
add “drm.debug=0x04” to the kernel boot commandline, which will cause
the kernel module to print out a bunch of modesetting debugging
information. Since this problem goes back to boot with the plymouth
splash screen, it's clearly a problem in the kernel module's
modesetting.

It's very strange that it appears non-deterministic!

[Matthias Andree (matthias-andree)]

I no longer have access to the hardware in question, so I will not be able to answer further questions.

[Bryce Harrington (bryce)]

Since the hardware is no longer available for doing troubleshooting with, we'll have to close the bug for now. However please feel free to reopen if you or anyone else has the same HW and can reproduce this issue using the latest development version of Ubuntu and is willing to do some troubleshooting with it.

[Rolf Leggewie (r0lf)]

lucid has seen the end of its life and is no longer receiving any updates. Marking the lucid task for this ticket as "Won't Fix".