I’ve made some progress. By setting up an xorg.conf with a modeline for the external display, I got graphics on the (lightdm) login screen to show up on it. But the login input fields weren’t on this screen (it was just the background), so this was further evidence that there’s monitor misdetection going on (greeter/lightdm happily thinks it’s displaying the input fields on the broken internal LCD).

So I further tweaked xorg.conf to force X to ignore the internal display, and now I’ve got a working login screen back on the external display. I’ll attach my xorg.conf below; it is a satisfactory workaround.

Further details: mouse cursor is briefly visible just prior to the black (login) screen. GDM does show up on the display, but in too big a resolution, so that I only get some of the grey background visible on screen and still have to log in blindly.

It occurred to me that with lightdm, the symptom appears as I would expect if it failed to do monitor detection properly and thinks it’s displaying the login screen on the (broken) internal display only. Well, apart from the funky ”login screen desktop background” phenomenon (which could be a separate issue though).

(Just to be clear: this same combination of broken internal + working external display did display the login screen correctly on the external one in Precise.)

This is a laptop with a broken internal display (”LVDS1” in xrandr output), and therefore connected to a working external LCD (”VGA1”).

After upgrading from Precise to Trusty, boot ends with a black screen (on the external display). The greeter sound effect is played. Typing my password (I’m the sole user) does log me in, and brings up the desktop on the display.

So the login functionality is there, just the visuals are missing. And they do flash by (corrupted) when the desktop is loading; if I set automatic login for myself, I get a desktop with the login screen as the background image (I’ll attach a screenshot).

There are existing, similar reports but all with subtle differences from this (such as Bug #1252748, where the user only gets black screen half the time) or too vague for me to claim to be the same issue with certainty (such as Bug #1242213). A lightdm restart (such as in Bug #969489) doesn’t fix this (i.e. black screen still after lightdm restart).

I’m not sure whether this is a lightdm issue or a unity-greeter one, so my reporting this against the latter is just a guess.

I currently have a 100% reliable recipe for reproducing this in 12.04: trying to restore a saved session of Trusty crashes VirtualBox 4.3.2 (from Oracle), after which Apport’s ”TypeError(Error(’Incorrect padding’,),)” pops up. (The recipe works also after removing crash files from /var/crash.)

Created an attachment (id=259453) [details]

This is in Fedora 19 with current Gthumb 3.2.4. Likewise reproducible in Ubuntu 12.04 with Gthumb 3.1.2 (from a PPA).

Steps to reproduce:
1. Open foo.png
2. Select File > Save As
3. Name the file foo2.png, select Save
4. In the Options dialog, select Cancel
5. Back in the Save As dialog, select Save again

What happens:
The Options dialog now only has Title and the Cancel and Save buttons, no actual options to choose from.

What I expect to happen:
The Options dialog to have options just like at step #4.

More information:
The bug crops up when saving the file as JPEG as well, I just used PNG as an example. It also happens regardless of whether a file is chosen for overwriting or not. After the bug is triggered, subsequent attempts to Cancel and re-enter the Options dialog always bring it up empty, until the Save Dialog is exited (either by Canceling or by going ahead with the Save).

Hi Christopher.

I was unable to test the Saucy images and the latest mainline kernel, as they refuse to boot/install on the Amilo due to lack of PAE support.

jani@kingugidora:~$ LC_ALL=C sudo dpkg –install linux-image-3.11.0-031100rc6-generic_3.11.0-031100rc6.201308181835_i386.deb
(Reading database … 818498 files and directories currently installed.)
Unpacking linux-image-3.11.0-031100rc6-generic (from linux-image-3.11.0-031100rc6-generic_3.11.0-031100rc6.201308181835_i386.deb) …
This kernel does not support a non-PAE CPU.
dpkg: error processing linux-image-3.11.0-031100rc6-generic_3.11.0-031100rc6.201308181835_i386.deb (–install):
subprocess new pre-installation script returned error exit status 1
Examining /etc/kernel/postrm.d .
run-parts: executing /etc/kernel/postrm.d/initramfs-tools 3.11.0-031100rc6-generic /boot/vmlinuz-3.11.0-031100rc6-generic
run-parts: executing /etc/kernel/postrm.d/zz-update-grub 3.11.0-031100rc6-generic /boot/vmlinuz-3.11.0-031100rc6-generic
Errors were encountered while processing:
linux-image-3.11.0-031100rc6-generic_3.11.0-031100rc6.201308181835_i386.deb

I guess the only (unimportant) question remaining is how come I was able to boot -pae kernels on this very same device back in the pre-release times (3.2.0-[<23]) – perhaps those earlier kernels didn’t properly detect the lack of PAE?

The wireless now (with the current 3.2.0-52-generic #78-Ubuntu) seems to somehow work without wistron_btns.

I’m not sure how to deal with the tags & bug status in this case, so I’ll leave it to your discretion. If you have further ideas to test, I’ll be happy to provide, but closing this report (perhaps as ’Invalid’) is also fine by me.

Hi Thomas, sure. Yes, the bug is still reproducible in Precise (gimp 2.6.12-1ubuntu1.2), which is what I’m using on my main desktop. But the good news is I can’t reproduce it in VMs running Raring or Saucy (both of which currently have gimp 2.8.4-1ubuntu1), so it looks like it’s been fixed in the 2.8.4 series at the very latest.

This is basically the same as bug #870297, but for LXDM.

Systems using LXDM have no entries in the utmp record for logins, so tools like ’w’ and ’who’ do not show LXDM sessions.

Testcase:
1. Start Lubuntu system
2. Log into session
3. Open gnome-terminal
4. Run ’who’ command

Expected result:
You should see TWO entries (the first is from LXDM, the second is from gnome-terminal):
tty7 YYYY-MM-DD HH:MM
pts/0 YYYY-MM-DD HH:MM (:0)

Observed result:
You see only one entry:
pts/0 YYYY-MM-DD HH:MM (:0)

Additional info:
Used gnome-terminal just for consistency with the LightDM report; lxterminal also doesn’t seem to update login records so if you use it instead, observed result is actually ”no entries at all” (and issue for another report). Also, bug #1027805 about wtmp records is similarly applicable to LXDM, but that remains to be fixed even in LightDM…

Also reproducible with chromium-browser 28.0.1500.52-0ubuntu1.12.04.2 (in 12.04).

As the original reporter, I’m confirming that 5.20.0-0ubuntu2 from -proposed does contain the fix for this bug, i.e. the issue I reported above is no longer reproducible. I’m only hesitant about tagging this ’verification-done’ because PerformingSRUVerification is pretty adamant about having a solid test case and I’m not sure my sloppy initial report qualifies.

jani@saegusa:~$ LC_ALL=C apt-cache policy unity
unity:
Installed: 5.20.0-0ubuntu2
Candidate: 5.20.0-0ubuntu2
Version table:
*** 5.20.0-0ubuntu2 0
500 http://archive.ubuntu.com/ubuntu/ precise-proposed/main amd64 Packages
100 /var/lib/dpkg/status
5.18.0-0ubuntu2 0
500 http://archive.ubuntu.com/ubuntu/ precise-updates/main amd64 Packages
5.10.0-0ubuntu6 0
500 http://archive.ubuntu.com/ubuntu/ precise/main amd64 Packages