suspend2 on PowerBook: keyboard doesn't work on resume

Hi,

I'm trying to get suspend2 working on my PowerBook G4 of Oct 2003 vintage.

I'm running kernel 2.6.15.6 with the 2.6.15.1 suspend2 patch (it applied
and built fine).

Before suspending, I quit X and logged in as root on the console.  I
unloaded all the modules I could - leaving just agpgart and uninorth-agp.
USB is kernelized and the modules are unloaded.  I started a loop that
printed "Running" every 10 seconds (and also started a job that after 60
seconds would reboot).  I ran the 'hibernate' script.  It completed
successfully and powered down the machine.

On the next startup, the kernel boots and it resumes.  I see the root
terminal again, complete with my "hibernate" command and the syslog
messages detailing the progress of suspend2.  The system does not respond
to keystrokes, however.  After a few seconds, the "Running" message is
printed out, and after 60 seconds, the system reboots.  To me this
suggests that the system is mostly resuming correctly, but something has
gone funny with the keyboard driver.

Keyboard-related dmesg:

MacIO PCI driver attached to Intrepid chipset
input: Macintosh mouse button emulation as /class/input/input0
adb: starting probe task...
[snip]
PCI: Enabling device 0002:20:0d.0 (0000 -> 0002)
adb devices: [2]: 2 c3 [3]: 3 1 [7]: 7 1f
ADB keyboard at 2, handler 1
Detected ADB keyboard, type ANSI.
input: ADB keyboard as /class/input/input1
input: ADB Powerbook buttons as /class/input/input2
ADB mouse at 3, handler set to 4 (trackpad)
input: ADB mouse as /class/input/input3
adb: finished probe task...

Does anyone on the list use suspend2 on a PowerBook?  Any success/failure
stories?  Any advice for generating more useful debugging information?

Thanks,
dustin.

Hi,

I'm trying to get suspend2 working on my PowerBook G4 of Oct 2003 vintage.

I'm running kernel 2.6.15.6 with the 2.6.15.1 suspend2 patch (it applied
and built fine).

Interesting. I was pretty sure that version wouldn't work at all :->

On the next startup, the kernel boots and it resumes.  I see the root
terminal again, complete with my "hibernate" command and the syslog
messages detailing the progress of suspend2.  The system does not respond
to keystrokes, however.  After a few seconds, the "Running" message is
printed out, and after 60 seconds, the system reboots.  To me this
suggests that the system is mostly resuming correctly, but something has
gone funny with the keyboard driver.

Keyboard-related dmesg:

MacIO PCI driver attached to Intrepid chipset
input: Macintosh mouse button emulation as /class/input/input0
adb: starting probe task...
[snip]
PCI: Enabling device 0002:20:0d.0 (0000 -> 0002)
adb devices: [2]: 2 c3 [3]: 3 1 [7]: 7 1f
ADB keyboard at 2, handler 1
Detected ADB keyboard, type ANSI.

There's an ADB keyboard -- this all probably suggests that the ADB
driver from pre-resume and original kernel are stepping onto each other.
Maybe they have no good enough power management?

Does anyone on the list use suspend2 on a PowerBook?

I used to, but on a later powerbook with USB keyboard, so...

johannes

Hi,

quoted

I'm trying to get suspend2 working on my PowerBook G4 of Oct 2003 vintage.

I'm running kernel 2.6.15.6 with the 2.6.15.1 suspend2 patch (it applied
and built fine).

Interesting. I was pretty sure that version wouldn't work at all :->

Do you mean that version of suspend2, or that combination of kernel and
suspend2 patch?  I can try a different combination of versions if you
suspect some other configuration will work...

There's an ADB keyboard -- this all probably suggests that the ADB
driver from pre-resume and original kernel are stepping onto each other.
Maybe they have no good enough power management?

Perhaps.  Nigel Cunningham, the main suspend2 developer, suggested that I
post this problem to this list to see if anyone knew what was happening.

Thanks,
dustin.

Hi,

To follow up on my previous post: after upgrading to suspend2.2.2 and
kernel 2.6.16.1, suspend/resume works perfectly on my PowerBook.

(I was just getting ready to do some kernel hacking to figure out why it
didn't work, and was pleasantly surprised to find that it did work!)

Cheers,
dstn.

On Sat, 2006-04-01 at 11:18 -0500, Dustin Lang wrote:

To follow up on my previous post: after upgrading to suspend2.2.2 and
kernel 2.6.16.1, suspend/resume works perfectly on my PowerBook.

Good to know. Nigel told me it now uses the mainline suspend code for
doing the hard stuff, so a lot of the hacks I had previously went out
and were replaced by some proper stuff :)

johannes