Bugzilla – Bug 35843
Regression: xrandr doesn't offer all resolutions
Last modified: 2011-04-27 01:04:41 UTC
Version "unspecified" because I have no idea what version the intel driver version 2:2.14.902-1+exp1 (Debian package) corresponds to.
Connecting my laptop to my projector, I don't get all resolutions the latter is capable of. This used to work just fine on older versions (checked with 2:2.13.0-5, again Debian package version, and other packages had older versions as well.) Especially I miss the projector's native resolution of 1360x768.
Full details: Please see the corresponding report against the Debian package on http://bugs.debian.org/617419; Laptop is an Acer netbook connected via VGA to a Panasonic PT-AX200E. Also in the Debian report are Xorg.0.log and other data from a working and a non-working configuration.
So it seems that we are not passing on the EDID for the VGA connection. Can you please add drm.debug=0xe to your grub boot parameters and attach the resulting dmesg from boot until X start? And we will take it from there.
Created attachment 45117 [details]
dmesg: drm.debug=0xe until X was started (kdm running)
Created attachment 45118 [details]
Xorg.0.log: drm.debug=0xe until X was started (kdm running)
For both logfiles: nothing attached to the VGA port. If you need it, I can do it with the problematic projector attached. (Not right now, need to be at home for this...)
Funnily enough, I had the laptop attached to a TFT yesterday, where I had the native 1440x900 resolution available. So it seems that the communication with the attached display works in some cases.
(In reply to comment #3)
> For both logfiles: nothing attached to the VGA port. If you need it, I can do
> it with the problematic projector attached. (Not right now, need to be at home
> for this...)
Ah, I wasn't going mad then! Please can you repeat this with the problematic projector. I'm interested in seeing what actually happens when we detect the connection and attempt to probe the EDID.
Created attachment 45302 [details]
* log in (kdm)
* plug in projector
* activate (via krandrtray)
I waited a bit between the steps, I hope it shows in the log timestamps.
Created attachment 45303 [details]
So, it looks like we are missing the projector's EDID.
There are a few patches for a very similar issue on drm-intel-staging [http://git.kernel.org/?p=linux/kernel/git/ickle/drm-intel.git], particularly the GMBUS fixes. Is it possible for you to try a kernel built from -staging?
I'll see what I can do, haven't compiled my own kernel in years. (just from git HEAD?)
I'll just use the config from my distribution kernel, or is there anything special I need to disable/enable?
The commits that I think will help the most have just landed in 2.6.39-rc2, so you might be able to find an experimental kernel with them.
Created attachment 45349 [details]
dmesg with new kernel
Not fixed :-(
This (and the x log) is with drm.debug=0xe in kernel 2.6.39 from drm-intel-staging at commit c97b38bd ("Clear any errors")
At 154.520922, I'm plugging the projector in and at 248.264516 I'm activating it from krandrtray. At around 193.256245, I'm querying it with xrandr (no arguments).
This is still Debian's xserver-xorg-video-intel "2:2.14.902-1+exp1" (Debian version number) -- do I need a newer version of this?
Created attachment 45350 [details]
(In reply to comment #10)
> This is still Debian's xserver-xorg-video-intel "2:2.14.902-1+exp1" (Debian
> version number) -- do I need a newer version of this?
No, the issue is in the kernel. Until i915.ko can see the EDID, we're lost.
Closing as a hw specific issue with no reporter anymore :(