On dim, 2009-03-01 at 07:39 +0000, Sheridan Hutchinson wrote: > Package: pm-utils > Version: 1.2.4-2 > Severity: important > > Hello, > > I'm a longtime pm-utils user on my Thinkpad X40. This laptop reached S3 and > S4 > nirvana a couple of years ago with the minimum of fuss. > > I'm using squeeze and a nice shiny new pm-utils package turned up this week. > I > installed it and didn't think much of it. A consequence of this new version > is > however that my backlight no longer comes back on when I resume from S3 > (haven't > tested S4 yet). It was working reliably prior to this. > > I do not have uswsusp installed (as I've never needed it) and have been > contently using > the in-kernel suspend mechanism. > > The only thing this laptop previously needed to work was to pass > acpi_sleep=s3_bios,s3_mode > hints to the kernel in the /boot/grub/menu.lst file. They are still in there. > > What should I do to get my backlight working again? This is a production > machine > that is very compromised by this bug. If I hold the screen under intense > light > I can make out images on the screen and it is perfectly possible to kill X > with an CTRL ALT BACKSPACE however this doesn't bring the backlight back >on.
I have the same kind of problems. Well, in fact, my backlight is not
correctly switched off when suspending using pm-suspend on my shiny
“new” X31. I tried to debug the /usr/lib/pm-utils/sleep files, and think
I may have found the problem.
In /usr/lib/pm-utils/sleep.d/98smart-kernel-video there's a check to
detect if the system uses KMS, and disable all the quirks if it's the
case.
The test is:
have_kms()
{
# if we are running with a KMS-enabled video driver, we should not
# attempt to run any quirks
[ -d /sys/class/drm/card0/device/graphics/fb0 ] || return 1
remove_all_video_quirks
add_parameters --quirk-no-chvt
}
But on the X31 (using a radeon card, and thus not using KMS) I have
the /sys/class/drm/card0/device/graphics/fb0 folder. This folder only
means there's a framebuffer running, I guess, and I think it then breaks
many machines.
Having the function return 1 unconditionnaly fix the problem here, so I
really think it's related to the remove_all_video_quirks stuff.
Cheers,
--
Yves-Alexis
signature.asc
Description: This is a digitally signed message part
