Does anyone know the chain of events which occurs when a laptop running X
has its lid opened or closed? I presume it's slightly different to a
desktop machine running a CRT (where there's no concept of a 'lid switch')

Does the hardware tell the kernel ACPI code that a lid event has occurred,
and the ACPI code then notifies X, which then turns DPMS on or off as
appropriate? Or is there some other process which happens?

Is it even a software thing at all, or should the hardware turn off the
display on lid close (enabling it again on open) - with software only
doing display status change as some sort of power management / keyboard
activity?

Notes on specific problem:

I'm trying to chase a bug where my laptop (Dell Inspiron 2200) doesn't
always turn the display (Intel 915GM) off on lid close, or doesn't always
enable it again on lid open (and when it gets into such a state the only
fix [1] is to plug in an external CRT, flip to and and then flip back to
LCD - at which point the LCD comes back to life). So far no amount of
playing around with custom kernel builds has fixed it though, which makes
me worry it's purely a hardware thing...

[1] tell a lie, for changing dpms via xset fixes it too, but it's hard to
open a shell and type in it with a blank display Flipping to console
(via ctrl-alt-Fx) and straight back to X *can* also work - but only when
using the VESA driver for X (it doesn't work if I use the i915 driver)

If the display doesn't shut off on lid close, it *always* then shuts
off when I open the lid again (requiring me to do the CRT trick above).
But once the display is blank, no amount of changing lid state will bring
it back. To me that rules out a simple intermittant lid switch problem.

Kernel (2.4.x and 2.6.x) debug so far has been no help, as logged messages
are the same whether a lid open/shut is successful or not - hence if all
of this *is* the responsibility of software I'm hoping I can add my own
debug to try and trace the problem...

If worst comes to worst, I can 'hotwire' a certain key combination to run
an xset DPMS command (so I can bring the LCD back to life without a CRT),
but that's a little ugly and I'd rather get to the bottom of the actual
problem first :-)

cheers

Jules