Bug#464015: xserver-xorg-video-mga: doesn't resume properly after s2ram
On Mon, Feb 04, 2008 at 08:40:16PM +0100, Brice Goglin wrote:
> Nikolaus Schulz wrote:
> > testing the mga driver from experimental I find that indeed DVI now
> > works with my G550 card (cooool, finally!!),
>
> Good to know.
>
> > but unfortunately the
> > screen doesn't resume into a sane state after suspending to ram (s2ram).
> > This is not the case with the driver in unstable (plus the proprietary
> > HAL module).
After looking into this more again, I have to correct the last
statement: the drivers in unstable do not load the HAL module, and
neither the driver in Etch. -- I've been running the unofficial driver
from Alexander Griesser <tuxx@tuxx-home.at> for years now; I'm sure
people involved with the mga driver know Alexander's driver.
I've made a Debian package of it long ago and forgot some details about
it, sorry. It is only this driver that loads the proprietary HAL lib,
works with DVI, *and* did successfully suspend.
So it's not a regression in the free driver.
> > Suspend to disk works like a charm.
> >
> > After s2ram the machine seems to be entirely alive, but the screen is
> > all white. I can access the console, and at least once even my xterm
> > was still working, but I'm blind.
> >
>
> Did you ever get s2ram to work on this box ? (when not using DVI?)
See above. I did another heap of tests, and can now say that without
Matrox HAL (and possibly a patched X.org driver that actually loads it),
the drivers in Etch and unstable do not successfully suspend to RAM with
both DVI and analogue connectors.
> What happens with driver 1.4.8 from unstable?
Blank screen with DVI. Works with analogue connector, but doesn't
suspend to ram (didn't test s2disk).
> > Restarting X with ctrl-alt-backspace doesn't help: the screen flashes a
> > couple of times (trying to initialize?) then enters text mode, still
> > flashing and unusable. I am attaching two X logfiles, one captured
> > right after resuming, and another one after restarting X (as described).
> > Obviously X sees only 2MB video RAM when restarted, and finally
> > crashes...
> >
>
> Ok there's a bug in the way the driver retrieves the VideoRAM after suspend.
>
> The crash comes from the pixman lib and is similar to upstream bug
> #12398 which is supposed to be fixed.
Yes, apparently another bug in the pixman library...
> What is your server doing after
> startup? Loading a real window manager with shiny effects? Or just
> loading a simple xdm?
Erm, can this have any effect here? I'm mostly running Windowmaker and
Gnome. One stable system, one unstable.
> > I've tried to get the driver from upstream's git repo working but
> > failed, I ended up with a blank screen. Please say if you think I
> > should still report this upstream.
> >
>
> Let's see what Tilman as to say first :)
>
> Brice
Okay! Hey Tilman, it would be really *awesome* if we could finally get
rid of that damn binary blob from Matrox. :-) It has been a PITA for
years now.
Nikolaus
Reply to: