X completely corrupts VesaFB Virtual Terminals' image

.

.

.

Marking as confirmed based on reports at the ubuntuforums, URL: http://www.ubuntuforums.org/showthread.php?t=160006 .

Based on a report at http://www.ubuntuforums.org/showthread.php?t=160006 suggesting that setting vesafb mode to match X resolution would fix this issue, I have tried setting my X resolution
back to 1280x1024 (from 1600x1200) and using a 1280x1024 vesafb mode, to no avail.

As a further note - along with installing the updates, I ran "sudo dpkg-reconfigure xserver-xorg", changing at least the max. resolution from 1280x1024 to 1600x1200. (My current xorg.conf is already attached to this bug report.)

I will attach Xorg.0.log and the output of "sudo hwinfo --framebuffer".

Boot parameter vga=normal used, since I very much prefer to have usable VT's. If a log-file of X while passing the kernel a vesafb boot value would be useful, let me know.

.

Reverting, will let the developers mark this as "confirmed".

Seemingly identical effects present after hibernation at bug 36904. Cause may be different, will not mark either as a duplicate at this time.

There are numerous reports on seemingly identical issues, all related to the framebuffer VTs / consoles not working when X is run. This occurs with varying hardware (and varying kernel versions), so the issue may be wider than previously suspected, and possibly related to something that these occurrences have in common.

However, even as these reports have much in common, I am hesitant to mark any as duplicates, since the hardware varies (e.g. four different graphics cards), and because the cause may or may not be the same.

As I summarised these reports, I find it beneficial to attach the summary to each report. Apologies to recipients of Ubuntu Bugs & Ubuntu X SWAT for having to receive multiple copies of the same.

A brief summary follows:

Four reports, with the common behaviour that the framebuffer consoles stop working _as soon as X starts_:
- Bug 21416 - reported by felix.rommel. Acer Aspire 1350 with a Via KN400 Unichrome graphics card.
- Bug 29407 - unsure if original report is related or caused by a different issue, but all comments are related. Comments by Ignacio Lago Fontán, Compaq Evo N600c, ATI Radeon Mobility M6 LY; also by Thilo Six.
- Bug 40297 - reported by kko. Matrox G450 graphics card.
- Bug 42974 - reported by Richard Laager. HP dv800 with a GeForce Go 7400 graphics card.

A fifth report, with identical behaviour (fb stops working with X running), but only after hibernate + resume):
- Bug 36904 - reported by kamome with "dapper flight5 on a via mII12000".

A further report that may or may not be related, but lacks details (i.e. does the fb not work at all, or only stop working after X has been started? what equipment?):
- Bug 42168 - reported by vertiger.

Is this significant? Both in bug 21416 and bug 40297 the Xorg.0.log shows the following:

(II) Loading sub module "fb"
(II) LoadModule: "fb"
(II) Loading /usr/X11R6/lib/modules/libfb.a
Skipping "/usr/X11R6/lib/modules/libfb.a:fbmmx.o": No symbols found
(II) Module fb: vendor="X.Org Foundation"
 compiled for 6.8.2, module version = 1.0.0
 ABI class: X.Org ANSI C Emulation, version 0.2

On my system, libfb.a has the following details:
 -rw-r--r-- 1 root root 214592 2005-10-10 21:10 libfb.a

Even mc complains of "no symbols" for fbmmx.o if I take a look at libfb.a.

A further comment related to the previous.
File libfb.a belongs to package xserver-xorg-core:
$ dpkg -S libfb.a
xserver-xorg-core: /usr/X11R6/lib/modules/libfb.a

kko: have you tried with later releases? Feisty herd5 was just released, and after that a new mga-driver got in so you could try them.

Thank you for the information.

I must say that I am now comfortably running an install of Kubuntu Dapper (apparently with no special vga options for grub), and so am unlikely to at this point take the trouble of fixing something that isn't broken.

I will report back, however, if I decide to e.g. try the Feisty live-CD.

Bug is not present on current Dapper.

kko: did you mean that this bug isn't present in current Feisty? You already said before that Dapper is fine :)

Umm, no. Haven't tested on Feisty. What my last two comments mean, on Dapper, is this:
- First, with boot parameters amounting to "vga=normal" ("no special vga options", i.e. vesafb not used), this problem isn't actual and can't be tested.
- Second, after I passed "vga=791" to the kernel at boot time, the vesafb worked correctly, showing the expanded console mode (and the splash appeared accordingly smaller), _and_, more importantly, the consoles continued to work even after X had been started.

Thanks for your concern, though.

PS. You may notice I edited the title earlier to remove the reference to Dapper. I hadn't myself met the problem on Dapper, at the time - only reports of it from other people. (I have since stopped referencing other people's experiences in my reports. ;-) So, I can't personally verify that the problem existed on Dapper, but if it did, an update has fixed this.

The activity log doesn't show who added the task for "...-ati" to this bug.

Anyone able to share some light on that, and if the bug is fixed also for ati? (There are probably other reports originally reported against ati, too.)

Since there is no information whatsoever relating to the "...-ati" -task in this bug report, in the interest of bug database maintenance I am going to mark that task as "rejected". As mentioned, there are other reports (also referenced here) on ATI cards, and tracking the possible issue is probably best done where at least some information already exists. (Odds are that also some of those issues may have been fixed, but I can't really tell.) If anyone has cause to continue tracking the ATI issue here, please revert the status and try to provide the relevant information in the report.

Hi, I have an ATI Radeon 7200, using the open source drivers. If I change to virtual console 1 (ctrl + alt +f1) or any of the others, I get very strange things going on. Text in the wrong places, the blinking cursor not in the correct position, strange colors, etc.

All my virtual consoles are corrupted and display very weirdly, and this makes it impossible for me to edit text files like xorg.conf. I'm not sure what info to provide but I can take some pictures and upload them here.

I haven't modified anything but this issue comes up out of the box for me on a fully updated Feisty Fawn. I didn't use Edgy, but I had no issues with this on Dapper Drake (which also had usplash) or previous.

Bug 73826 seems related to this---the open source ati/radeon and fglrx drivers cause unusable virtual terminals in Edgy

I'll see if my advice can be of any use.

From the bug report you referenced it seems that disabling splash from kernel boot options (in grub) has helped the reporters. You could try the same to see if your screen corruption is also caused by the splash, in which case it's probably useful to continue tracking the issue in that report. (Apparently - I don't know - they are still running a _framebuffer_ virtual console at this point.)

If disabling splash doesn't help, you can check if the corruption is instead caused by X by booting to text mode, and verifying that you are still running a framebuffer virtual console (instead of the "normal" or "extended" modes). If the corruption occurs when X is running, but is absent without X, your issue would seem related to this bug.

Hope this helps.

Hi kko, thanks for replying so quickly,

I just installed feisty's new kernel packages (2.6.20-13->2.6.20-14 before reading your response, so I restarted and the problem was gone. Hmm...

I will keep the hints in mind if this happens again and respond back here.

Thanks again