[ltp] Re: T21, hangs on ACPI resume

[Alexis Parseghian]

> I am also in much the same boat, (type 2674-9BG). Sleep (fn-F4) work OK s=
o
> long as I resume within an hour or so. More than that and it doesn't come
> back (in what sounds like much the same way as Alexis described)

Can't even get it back after a few seconds, here.

> > Thanks. I'll post the results of the tests here.
> please do.

Here we go. As Kevin suggested :

a) edited /etc/default/acpi-support "MODULES" list to unload/reload
the following:
"irda soundcore pcmcia_core 3c59x psmouse uhci_hcd"

a') regular boot to the ACPI-enabled setup, grub options :
/boot/vmlinuz-2.6.12-9-386 root=3D/dev/hda1 ro vga=3D791 (...)
   (...) acpi_sleep=3Ds3_bios nolapic noapic pci=3Dnoacpi
b) init 1
c) happily rmmoding... (rmmod -f)
3c59x, psmouse, speedstep_smi, pcmcia, uhci_hcd unload OK

unable to unload: snd/soundcore, bluetooth, irda : "Resource
temporarily unavailable".

At this point, buttons aren't supported anymore (unloaded when
switching to runlevel 1), so run /etc/acpi/sleep.sh, machine goes to
sleep.

Doesn't wake up, same as before.

d) regular boot to above grub line, edited to take out "vga=3D"
e) init 1
f) rmmod 3c59x, psmouse, etc... OK -- didn't try removing the ones
previously failing
g) tried rmmod -f vesafb : "Device doesn't have a release() function",
but is gone from lsmod
h) sleep.sh

Still doesn't wake up.

Can't see anything else to unload, short of rmmod'ing all the way to a
kernel panic...

> I havce been unable to find out how on earth one goes about
> debugging ACPI, especially when resuming is broken.

Agreed. Not even having a log trace (I guess the machine doesn't wake
up enough for that) and not having any idea *where* it fails doesn't
help.

Oh, and for what it's worth, BIOS rev is 1.14 ; according to the IBM
website, there's nothing relevant in the more recent updates.

A.