Comment 6 for bug 26194

Hi Daniel,

Although I've uninstalled Ubuntu on my laptop since I filed this bug, I had
installed those DRI snapshots with the fix as per the X.org Bugzilla, and there
were no more segmentation faults when running in 1280x1024. You can mark that
part of this bug resolved ;)

Regarding the "bleeding pixels" (blooming) and complete system lockups, I
identified the issue causing this, which is seperate to the DRI bug. The
Inspiron 8000 has a special Function key (Fn+F7) that virtually resizes the
display to fit to the LCD's native resolution. For example, the 1280x1024
resolution shows a black border when this function is disabled, as the panel's
native resolution is 1600x1200 (or thereabouts), and when enabled, it expands to
fit the entire screen. When I run linux with the function disabled, I experience
no lockups. An interesting behaviour is noted for the r128 driver; when the X
session starts in 1280x1024 (for example), the screen expands as though Fn+F7 is
active, however when I quit X, the screen is no longer expanded. When I use the
vesa driver, the behaviour is different; when X starts, the virtual zoom stays
disabled (as it should be).

The problem, therefore, is that the r128 driver is "banging the hardware",
causing a conflict with the virtual zoom function. If the zoom function is left
disabled system-wide (as can be seen in the bios screens, Ubuntu boot, console,
even in Windows), the screen should always respect the zoom setting as disabled,
but X.org always enables the zoom itself on startup. If the zoom is already
enabled system-wide (which is my preference), random crashes occur when starting
or exiting X.org. I realize this bug shouldn't prevent the segmentation fault
from being marked closed/resolved, so if you want I'll file a new bug with this
seperate issue.

Conn