Software Suspend: 2.6.0-test9 vs. 2.4.22 patched with swsusp 2.0rc2

From: John Mock
Date: Tue Nov 04 2003 - 05:42:10 EST

In order to understand what's broken in 2.6.0 vs. what hasn't worked
earlier, i picked up the current stable kernel and patched it with the
current software-suspend release candidate. Some things which i have
found to be broken in 2.6.0-test9 on a VAIO R505EL seem to work at least
partially with 2.4.22 patched with the current software-suspend release
candidate (patch).

To wit:

'uchi-hcd' comes back into an almost plausible state with 2.4.22-sws-rc2
and fails as previously documented under 2.6.0-test9. However, it
won't sleep with a mouse or digital camera connected, and without
that, after sleeping, it freezes when i try to plug in a USB mouse.

Rather than dying when swap space isn't initialized, 2.4.22-sws-rc2
just quietly doesn't go to sleep.

X still has problems under 2.4.22-sws-rc2; however, it does not double-
fault or auto-reboot as it does under 2.6.0-test9. It does not
come back properly when hibernate is done from inside a native X
window, but it does if native X (DRI) is running, and hibernate is
done from a console screen rather tahn from X (and i don't use a
special hack to get X to run more 256 colors at 1024x786 on a VAIO

2.4.xx software suspend just feels closer to being production code (but
isn't a mainstream kernel).

So, 2.4.22-sws-rc2 doesn't solve my problems either, but looks much closer
than 2.6.0-test9, alas.

Also the problem with 'serial_cs' releasing a resource twice does not occur
under 2.4.22-sws-rc2, so that also may be a recent development.

'ohci1394/sbp2' worked for me on 2.4.19, but doesn't on 2.4.22-sws-rc2, so
that might not just be a 2.6.0 issue (although it appears to ve at least
partially fixed by the new upstream 'ieee1394' sources under 2.6.0). It
doesn't get a slab error, it just doesn't seem to do anything.

We need to release 2.6.0, so given that software suspend is listed in the
'config' comments as 'experimental', i think it shouldn't hold things up.

I just wish this could have gotten more attention earlier. *sigh*

-- JM

P.S. With Nigel working on this on a regular basis, no doubt it will now.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at