Skip to contentSkip to conversation

APM woes: IBM T20 Thinkpad

DISCUSSION Alex Hudson 2 participants 2 messages Dec 8, 2001 – Dec 9, 2001

lore
Share & tools
lore
#0 lore raw

I'm running a self-compiled Linux 2.4.16-pre1 on a IBM-T20, and am seeing problems with APM.

Most times I suspend (apm --suspend) I can't un-suspend successfully (I've managed it _once_ ;). This is since upgrading - I was previously using a Debian version of 2.4.9. I have gone down to the console, turning off most services (e.g., ps aux is about one screen-ful) and it still doesn't work, so it doesn't appear to be some other service.

When re-animating, the disks come on and the screen backlights, but nothing much else. Disk access is non-existant (or at least, sufficiently minimal that I can't detect it) and the network isn't responding.

I've also seen a couple of other apm-wierdos - the battery being at 99% full, for example, even after being on the charger for a while. I also tend to see a lot of disk activity - the light flashes several times a second - but then, I'm usually heavily into swap (running Gnome, etc..).

If someone could point out some tests I could try to pin this down, or has some other suggestion, I would be very grateful.

Cheers,

Alex.

#1 lore raw

Alex,

Quoted 2 lines · --On Saturday, 08 December, 2001 11:20 PM +0000 Alex Hudson…

I'm running a self-compiled Linux 2.4.16-pre1 on a IBM-T20, and am seeing problems with APM.

...

Quoted 2 lines

If someone could point out some tests I could try to pin this down, or has some other suggestion, I would be very grateful.

This is what I did to try and get a T23 to work (which isn't a T20, I know).

First upgrade your BIOS and/or system controller - available FOC from the IBM web site.

If that doesn't fix it, try changing whether or not interrupts are allowed during suspend. I think for a T20 they should *not* be, but the T23 crashes if they are *not* (and needs it the other way around). Don't just fiddle with the config option as pci_quirks will override it. Either look at doing something like
http://www.alex.org.uk/T23/T23-apm-patch.txt
or just frig the code yourself to force apm.allow_ints one way or another.

--
Alex Bligh