[ltp] APM suspend with 2.6 kernel [long]

[Michael Hagemann]

On Mon, 05 Jan 2004 15:24:49 +0100 Thomas Hood wrote:

> On Mon, 2004-01-05 at 00:02, Michael Hagemann wrote: 
> > All right, I have pretty much isolated the problem.  The culprit is
> > somewhere in the "big IRQ lock" patchset.
> 
> But what is the important thing?  That patchset has a big footprint.
> Yet I have looked through it and I don't see anything that obviously
> could cause our problem.

The first thing I'll try is to reinsert the __cli/__sti calls right
around set_system_power_state in apm.c:1199.  That seems to be the most
promising try.  They also aren't there in 2.6.0-...


> Has anyone tried disabling preemption to see whether that helps?

I had it disabled, but judging from 2.4, where I turned it on and off
with various patches, it shouldn't matter that much.


> I see this without attempting a suspend.  It happens once in a while
> without any obvious cause:

Well, probably misleading in this context, but keeping an eye out can't
hurt.


michael