Suspend and 2.6.14

[fm2503]

Hi,

Since upgrading to the 2.6.14 kernel from kernel.org I can't get S3 suspend working anymore. (echo mem > /sys/power/state)

AGP is disabled in the kernel, and using NvAGP=1

The following is seen as the system attempts to sleep
NVRM: ACPI: unsupported event: 2
Could not suspend device 0000:01:00.0: error -1

My card is (in a dell D800):
0000:01:00.0 VGA compatible controller: nVidia Corporation NV34M [GeForce FX Go 5200] (rev a1)

Using 7676 with gentoo patches (r1) - the same thing happens with straight 7676 as well.

Any ideas?

Cheers

Matt

[Ferenc Szalma]

Hi,

I am getting the exact same error messages when trying the same "echo mem > /sys/power/state".

NVRM: ACPI: unsupported event: 2
Could not suspend device 0000:01:00.0: error -1

I am running gentoo 2.6.12-r2 kernel on a Toshiba m200 laptop. I have NvAGP=0 in xorg.conf. Nvidia bug report log attached.

Did you manage to get your suspend working. Or has anyone have an idea what the problem might be?

F.

[a7v]

IIRC the nvidia graphic drivers doesn't have full support for power management through ACPI. It's still using APM for that although they are working on it.

[rbirdman]

Originally Posted by fm2503

Hi,

Since upgrading to the 2.6.14 kernel from kernel.org I can't get S3 suspend working anymore. (echo mem > /sys/power/state)

AGP is disabled in the kernel, and using NvAGP=1

The following is seen as the system attempts to sleep
NVRM: ACPI: unsupported event: 2
Could not suspend device 0000:01:00.0: error -1

snip<

Any ideas?

Cheers

Matt

I see pretty much the same thing with the latest Fedora Core 4 kernel (2.6.14-1.1637_FC4)

For the time being I'm using the older kernel.

[philipl]

Originally Posted by fm2503

Hi,

The following is seen as the system attempts to sleep
NVRM: ACPI: unsupported event: 2
Could not suspend device 0000:01:00.0: error -1

I don't really understand why people are only seeing this now because I saw it with 2.6.13 and, I think, 2.6.12 as well.

For whatever reason, the driver is getting an S2 event when it should get S3. So, what to do? Edit the code and set it to handle S2 like S3. It's a pretty trivial edit to make. If people have trouble working it out, I'll post more details, but I have to run for now.

I have a Tecra M3 and with 2.6.14 and some sata patches, it will suspend and resume just fine.

[zander]

Recent Linux 2.6 kernels send PM_EVENT_* messages; PM_EVENT_ON and PM_EVENT_SUSPEND used to match PM_SUSPEND_ON and PM_SUSPEND_MEM, but in Linux 2.6.14, PM_EVENT_SUSPEND is #define'd to 2, hence the failure.

[Ferenc Szalma]

Hi,

Recent Linux 2.6 kernels send PM_EVENT_* messages; PM_EVENT_ON and PM_EVENT_SUSPEND used to match PM_SUSPEND_ON and PM_SUSPEND_MEM, but in Linux 2.6.14, PM_EVENT_SUSPEND is #define'd to 2, hence the failure.

So why is it that my gentoo-2.6.12-r2 won't suspend to ram, and gives the same error message. See my previous message and the attached nvidia-bug-report there.

F.

[rbirdman]

Originally Posted by zander

Recent Linux 2.6 kernels send PM_EVENT_* messages; PM_EVENT_ON and PM_EVENT_SUSPEND used to match PM_SUSPEND_ON and PM_SUSPEND_MEM, but in Linux 2.6.14, PM_EVENT_SUSPEND is #define'd to 2, hence the failure.

So it's a feature not a bug.

Can I grep through the source code for that define statement and change it to 3?

[rbirdman]

Originally Posted by Ferenc Szalma

Hi,

I am getting the exact same error messages when trying the same "echo mem > /sys/power/state".

NVRM: ACPI: unsupported event: 2
Could not suspend device 0000:01:00.0: error -1

I am running gentoo 2.6.12-r2 kernel on a Toshiba m200 laptop. I have NvAGP=0 in xorg.conf. Nvidia bug report log attached.

Did you manage to get your suspend working. Or has anyone have an idea what the problem might be?

F.

Try setting NvAGP=1.

http://members.westnet.com.au/rbirdman/suspend.html

I found a link that explained it better but lost it.

[bigsur2k4]

Originally Posted by rbirdman

Can I grep through the source code for that define statement and change it to 3?

I've tried it out a few minutes ago and it worked for me :-)

I editet /usr/src/linux/include/linux/pm.h on Line 214 and changed "#define PM_EVENT_SUSPEND 2" to "#define PM_EVENT_SUSPEND 3".
Then I recompiled my Kernel: "make clean && make-kpkg clean", "make-kpkg --initrd --revision ver2 binary-arch".

After reboot I tried to load the nvidia module and did an "echo -n "mem" /sys/power/state" and it seemed to work

After wakeup the display was distroyed, but I've found a script to get around that earlier this week. My /etc/acpi/actions/lm_lid.sh looks like:

#!/bin/bash
# lid button pressed/released event handler
/usr/sbin/laptop_mode stop
chvt 1
modprobe -r ehci_hcd
modprobe -r uhci_hcd
sync
echo "vbetool save"
vbetool vbestate save > /tmp/sleeping
echo -n "mem" > /sys/power/state
vbetool post
echo "restore"
vbetool vbestate restore < /tmp/sleeping
rm /tmp/sleeping
modprobe uhci_hcd
modprobe ehci_hcd
echo "brightness:3" > /proc/acpi/toshiba/lcd
/usr/sbin/laptop_mode start
chvt 7

I've a Toshiba Sattelite Pro M30 with a Geforce Go 5200, Debian Etch with Kernel 2.6.14.

Thank you so much for your hint, Zander!!!

Andreas

[rbirdman]

Originally Posted by bigsur2k4

I've tried it out a few minutes ago and it worked for me :-)

I editet /usr/src/linux/include/linux/pm.h on Line 214 and changed "#define PM_EVENT_SUSPEND 2" to "#define PM_EVENT_SUSPEND 3".

snip<

Thank you so much for your hint, Zander!!!

Andreas

Sounds like fun.

If I get a chance I'll try it tonight.

[zander]

@rbirdman: no, this is not a feature, the NVIDIA Linux graphics driver needs to be updated.