FC5 workaround for i810 video

[ikluft]

My FC5 install on my laptop ran into this i810 freeze problem which has already been observed in freedesktop.org's Bugzilla site. https://bugs.freedesktop.org/show_bug.cgi?id=5443

This uses a workaround from that bugzilla discussion which is not considered the final fix for the problem. But it makes the difference between "doesn't work at all" and "seems to work fine so far." I made a patched version of the xorg-x11-server-Xorg rpm at http://sw.sbay.org/fc5/ - the RPMs at this URL will go away when the long-term fix is distributed. The current patch is based on the xorg-x11-server-Xorg-1.0.1-9 RPM security update from today. Source RPM included.

[jim]

have you submitted this to redhats bugzilla as well ?

[ikluft]

I haven't submitted it to RedHat. This was the first public post of my patched RPMs' URL after telling some others at sbay.org about it.

I can't take credit for any of the discussion on the freedesktop.org bugzilla discussion - I simply found it when looking for the fix for my laptop. I suppose I could post it to RedHat with this caveat. The problem is that so far the fix is just a workaround. On the freedesktop.org bugzilla, there is a valid comment that this is not the right long-term fix, commenting out a call to ValidatePCI() in the core server to fix an i810 problem. But it's what people have done to get this working for now. So that's why I'm also calling this just a workaround.

[ikluft]

OK, it's opened as RedHat's Bugzilla #186016 at https://bugzilla.redhat.com/bugzilla....cgi?id=186016

[knocker]

can you tell me if this bug is also the reason that i810 don't work with intel 945gm? i only get a black screen with gdm and cannot start a x-server, but the starting-screen is visible during boot.


would really like to know


knocker

[ikluft]

I don't have a 945gm so I can't really confirm for you. But your symptoms sound exactly the same as everyone is reporting with the i810 driver for Xorg 7.0. So that's probably the cause for you too. If you're on Fedora (which is likely considering this is FedoraForum) then you can use my workaround RPMs listed above. In looking around I found this was also identified as a critical bug by Debian and Ubuntu - they have their own implementations of the workaround.

The reason this change is not being called a solution is because it simply comments out a call to the ValidatePCI function where the crash occurs for machines using the i810 driver. But that function is actually part of the core Xorg server. So this change is probably not beneficial to other hardware's users. So we need to work with Xorg and RedHat on finding a long-term solution that makes the Xorg server work for all the different types of hardware.

[knocker]

Ok,

i tried it with your packages right now.

For 945GM this is no solution, it still don't work with i810 and switches back to console.

[ikluft]

Well, you may have had more than one problem. If after running X you see the console at all again without rebooting, then that's not the same problem.

The symptom of this problem is that the i810 driver locks up the X display with a black screen or some garbage, and won't release it even with CTRL-ALT-BACKSPACE, kill -1, kill -15 or even kill -9. Though the machine is still running and can be accessed normally via the network.

If you didn't see the console before and now you do, then it's possible this solved one problem and now you have another.