May 05, 2013 - 07:57pm
I have just acquired the newly Samsung NP900X3E-A02US which includes HD4000. See lspci below. The situation is really not great regarding the graphics. The screen is garbaged by horizontal black lines that follow movement of windows. Some horizontal grey lines also appear over time and they are even sometimes persistent across reboot! See this picture. With Windows 8, there is no problem at all.
I have installed Ubuntu 12.04.2:
user@laptop:/tmp$ uname -a
Is it a known bug? How could I fix this problem?
Many thanks in advance for any advice,
Appreciate the information, gregoiregentil. Investigation is underway on our end. It would be helpful if you could also send the output of the following commands:
Also, if you could pass along the contents of these files:
If you are not comfortable posting some (or all) of that information here on the forum, feel free to send it to me in a private message. These commands are strictly queries-- they do not modify your system or its configuration.
Just to give more visual data point, when I start LibreOffice, the screen becomes completely garbaged as each line separating the rows leaves a horizontal white line. Even after a reboot and going to the BIOS settings (no OS started), some lines persist! There is no way to remove the battery of this computer, so I can not really shut completely off the system so that all internal capacitors get emptied. After a while, the white lines start to disappear.
dkms status reports nothing.
Hi - You mention that dkms status did not report anything - looking at the logs, I see you have
kernel 3.5.x installed, so I would expect that (the 12.04 installer has a backport from 3.4.x to the
3.2.x kernel that 12.04 originally shipped with - your kernel is already more up to date than that).
So if the problem lies in the i915 kernel module, it's a problem with the vanilla ubuntu 3.5.x kernel
and not the backport from the Intel graphics stack (for you anyway).
You _might_ be able to work around this by getting the i915-3.6-3.5 package from the Ubuntu 12.10
Intel graphics repository here:
If you want to try this, do the following:
• Download the .dsc and .tar.gz from there.
• dpkg-source -x *.dsc
• in the unpacked directory, dpkg-buildpackage -rfakeroot -b
• install the resulting -dkms package.
If it builds, you will have the backported kernel 3.6.x i915 driver.
However, since I see in the other posts that people are seeing this bug in 3.8.x kernels,
I wouldn't hold out too much hope for the 3.6.x driver fixing things.
It could be a regression in the i915 driver in the kernel, or it could be a quirk in this particular
Also, glxgears works very well. Even with the window maximized in 1080p, I get 60fps...
Like, I'm not dreaming/inventing an issue!!! http://bit.ly/16QrFs2
the same issue is happening to me with that same laptop, under Debian unstable, experimental kernel 3.8.3. I posted about it on the debian forums (but the spam filter doesn't want to let me post the link here)
Not that starting with noacpi or nomodeset makes it work, it seems to use the vesa fbdev driver for X instead of the intel one.
Any suggestion on what to try to get this to work is welcome
Here is the address of my post on the debian forum that shows the same issue:
Thanks nabajour and gregoiregentil for the information. I appreciate everyone's responsiveness. As analysis progresses, we may ask this thread for additional information in order to arrive at a solution.
nabajour and gregoiregentil: Thanks again for all the information you've provided. We (vivijim and I) are in the awkward position of not owning that particular laptop, so we cannot reproduce the issue or test any fixes on it ourselves. Should you choose to-- and I sincerely hope you do-- you can help us fix this quickly by escalating the issue.
If either of you are game, file a quick a bug with my colleagues who actively develop drivers for Linux. You can provide a link back to this thread, since there is a lot of good information here. If you're comfortable building kernels from source and applying patches, all the better-- you can help verify our work and expedite the process. Have a look at these guidelines-- bear in mind this issue is likely categorized as a "DRM kernel bug":
I entered a bug report in the bugzilla following the guidelines you showed:
I hope it contains all the needed info, I can give a go at compiling a kernel, I have done that a long time ago, haven't tried patching, though...
But hey, if that's what's needed to get my laptop working, I'll learn :-D
Thanks a lot, thac0.