I’ll let this one expire
The system in question is currently out of use (and reproducing the bug isn’t straightforward), so I’ll let this one expire.
The system in question is currently out of use (and reproducing the bug isn’t straightforward), so I’ll let this one expire.
FWIW, based on the above comments and my experience I’d lay at least some of the blame on this on nouveau: I have two setups affected by the Plymouth no-show, both with Nvidia graphics:
01:00.0 VGA compatible controller: NVIDIA Corporation G73 [GeForce 7300 GT] (rev a2) (prog-if 00 [VGA controller])
and
00:0d.0 VGA compatible controller: NVIDIA Corporation C61 [GeForce 7025 / nForce 630a] (rev a2) (prog-if 00 [VGA controller])
At the same time all the other systems I have access to that run Precise on non-Nvidia graphics (some half dozen computers, old and new) don’t manifest it, i.e. they display Plymouth’s boot logo screen just fine on every boot without any workarounds. (The Nvidias need either FRAMEBUFFER=y or plymouth:force-drm.)
Also, on the G73, Grub menu shows at a low resolution without GRUB_GFXMODE specified, but on the C61 it’s ”Out of range”. It shows on both if I specify 1280×1024 (the maximum supported by the connected LCDs) as GRUB_GFXMODE. This too is Nvidia-specific, though I suppose the Grub menu is outside nouveau’s jurisdiction.
Most of the problems (”freezing”) seen by the user would probably fall under Bug #994306, which I’ve now witnessed on this system. The one I reported above is different, and apparently much rarer in occurrence.
This happened after about 30 minutes of watching a flash video in fullscreen (using Firefox): the video froze, leaving just the audio playing, and the display no longer responded to user input until I ssh’d in and killed gnome-session for the user (after which LightDM login came up and things worked again).
According to user this freezing has been occurring daily, sometimes multiple times a day, since upgrading to Precise. It never occurred with Lucid.
We’ve yet to try if using 2D desktop or disabling KMS helps.
(Apport wouldn’t send this report without me lying to it that I was referred here by tech support, but I wasn’t.)
Bug #994203 looks similar, but it’s about NV4e whereas this is an NV4c.
Ah, sorry about that, I thought you wanted both updated logs from Lucid *and* tests with newer versions.
I just gave the current 12.04 Beta a spin and am happy to report that the issue seems to have been fixed there: despite my best efforts I couldn’t make it crash with the gallery site. It did remain somewhat sluggish to browse but the Oops never occurred.
I cannot get X working with the latest Mainline build for Lucid (2.6.35-rc1-lucid [1]) nor with linux-image-generic-lts-backport-maverick (which I figured I’d also try, as it’s also a build of 2.6.35). Both end with ”[drm] failed to open device” and ”No devices detected” in the log (I’ll attach one to this comment).
If I’m interpreting Nouveau Wiki’s Troubleshooting document [2] and the logs correctly, KMS is working (fb0 is there), which means there’s ”a version mismatch between the Nouveau DRM and libdrm”. Any help there would (also) be appreciated.
*[1] http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.35-rc1-lucid/
*[2] http://nouveau.freedesktop.org/wiki/TroubleShooting#Xorg_fails_to_start_with_.22.28EE.29_.5Bdrm.5D_failed_to_open_device.22
Timothy, does your harware setup match? You’re using the internal Nvidia graphics and not an add-on card?