Compaq nw9440 nvidia QuadroFX 1500M graphics errors

Bug #242382 reported by drink
2
Affects Status Importance Assigned to Milestone
linux-restricted-modules-2.6.24 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

I have long-running problems with my machine. I have mostly worked around them in the past, but now I am hitting a wall. Actually, I am hitting a black screen bug, among many others.

I am using a Compaq nw9440 (P/N EZ901AA#ABA - this is EXTREMELY relevant) and I have been keeping up with BIOS updates from HP, which have not seemed to change my list of problems at all. The only thing that has changed things substantially over time is that moving to hardy-proposed has caused me NEW problems regardless of whether I use nvidia-glx-new or use envy. This system has a QuadroFX 1500M PCIE 16x 256MB.

My longest-running problem is with screen resolutions and EDID. I get the full range of resolutions by default with the 'nv' driver. With the 'nvidia' driver I get only my panel's native resolution, and no others. The skinny from the Xorg output seems to be that the resolutions are rejected because they fall under the maximum pixel clock. I can see from various sources that the max pixel clock is 330, but it is reported under 200 and all modes are normally rejected under the 'nvidia' driver except the native, which again is successfully read from EDID. I have tried to read EDID under DOS but it is a version 1.3 EDID and the get-edid and parse-edid tools are not equipped to deal with it anyway.

My new problems are more serious. The most egregious problem is a black screen problem. It is not fixed by switching to a VC and back to X. The screen just goes black while I'm using it (or not, if I'm not there when it happens) and never comes back. I am using xserver-xgl, but it also happens with Xgl disabled. If I ctrl-alt-backspace then GDM restarts the server as normal and I get a login prompt.

The other serious problem is that the default EDID resolution now results in display trashing. I can run nvidia-settings (with -c :0 when Xgl is running) and manually set the resolution and refresh, which USUALLY solves the problem. Sometimes just repeatedly frying the server and letting it restart helps, but sometimes it just re-trashes the display. It looks like the display is being provided with bad sync; the bits of the display get written all over the screen. I might be able to take a camera picture of it if that would help. It's similar to a semi-static "snow". If I move the mouse around I can kind of see through it sometimes, and if I use nvidia-settings to set the resolution manually, the problem is fixed, usually until the other problem (black screen) crops up at which point I have to fry the server. The server is STILL WORKING in the background - for example, if it happens while I am writing this bug report, I will hit Super-R to open my run dialog, then run gnome-terminal, cat > file, Alt-E, Alt-P, Ctrl-D, Ctrl-D before I Ctrl-Alt-Backspace. I did this recently while authoring a comment on another website and it worked, so obviously the server is still operating in the background, but the screen is no longer being drawn.

I have been specifying modelines gathered under the 'nv' driver; now I am turning off max clock check and useedidresolutions and I get the full set of resolutions, but I get the display trashing. I have dual-link LVDS to my panel, I wonder if the max clock being reported is half of the actual max clock due to the dual-link being inappropriately accounted for? Of course everything works fine on Windows... but you know, I'd have to run windows.

If there is a way to pull a version 1.3 EDID under Windows, I guess I could reinstall from the system restore disks. I have an external disk I can clone my system to. But that doesn't solve the black screen bug and I'd like to see nVidia fix whatever bug is causing them to misdetect my display's capabilities anyway. It's been going on since version 100.something, when I got the machine. But the most serious problems (display trashing, black screen) didn't happen until I updated to hardy-proposed.

Driver: envy 174.14.05
Uname: Linux agamemnon 2.6.24-19-generic #1 SMP Wed Jun 18 14:43:41 UTC 2008 i686 GNU/Linux
Ubuntu: 8.04.1
Video BIOS: 05.71.22.43.a8 (updated with system BIOS image)

Xorg conf attached. Let me know what else I need to post.

Revision history for this message
drink (martin-espinoza) wrote :
Revision history for this message
drink (martin-espinoza) wrote :

Here's an updated xorg.conf that seems to have made the problems crop up less.

Revision history for this message
drink (martin-espinoza) wrote :

Here's the EDID that goes with that last xorg.conf

Revision history for this message
drink (martin-espinoza) wrote :

I am now running Intrepid and I am having only two problems: lockups/hangs, which I don't know if I can blame on the nvidia driver or not; and still need to manually add Modes to my xorg.conf, which I know I Can blame on the nvidia driver (because it works on Windows, whee.)

Bryce Harrington (bryce)
tags: added: hardy
Revision history for this message
bugbot (bugbot) wrote :

This bug report was filed against an old version of Ubuntu.
Can you confirm whether this is still an issue in natty?

If you don't mind, it would be very helpful if you could update the bug
report in launchpad to 'Fix Released' if it is no longer an issue for
you, or if it is still occurring under natty, please tag the bug 'natty'
so it's easier for us to track.

Changed in linux-restricted-modules-2.6.24 (Ubuntu):
status: New → Incomplete
Revision history for this message
drink (martin-espinoza) wrote : Re: [Bug 242382] Re: Compaq nw9440 nvidia QuadroFX 1500M graphics errors

For my part this turned out to be a die bonding error that HP knew was a
problem with my laptop, but it took me literally over a day on the phone to
get someone to admit it and send me another machine... after two
"technician" visits that left my laptop unbootable and not just glitching.
Record as "HP's fault" :)

Revision history for this message
Bryce Harrington (bryce) wrote :

Wow wild. Well glad to hear it's sorted out now! I'll close it out as invalid in this case I think.

Changed in linux-restricted-modules-2.6.24 (Ubuntu):
status: Incomplete → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.