[Pm-utils] second suspend/resume cycle hangs on resume

[Brian J. Murrell]

On Mon, 15 Dec 2008 17:08:47 +1100, Nigel Cunningham wrote:

> Hi Brian.

Hi Nigel,

> You might also carefully examine your system after the first cycle. This
> sort of behaviour is always the result of some driver properly saving
> and restoring its state (or not saving/restoring at all), and often
> results in the relevant hardware not working perfectly after the first
> cycle as well as dying properly on the second attempt at
> suspending/hibernating.

It all seems to work quite well, even after the first suspend/resume 
cycle.

> I'd therefore also suggest that you try building as much as possible as
> modules, and try to rule things out by suspending a couple of times
> without the appropriate module loaded.

One of the first things I tried in fact was unloading as many modules as 
I could, including bringing down network interfaces, etc.  ISTR I had it 
down to a dozen or less modules loaded and it still did the same thing.

> Hopefully you'll get to a point
> where you can say "if I suspend with module X loaded, it hangs".

Well, I have already gotten to a point where, if it's a module, I cannot, 
not have it installed.

> Start
> with testing after booting from init S, just in case it's something
> that's still built in after you've done the
> modularising-as-much-as-possible.

I'm not sure I'm following you on this instruction.  Can yo elaborate, if 
it's still relevant given my description above?

I think also, the work I did with /sys/power/pm_trace and detailed in the 
lkml posting that I pointed to in my original posting here points 
squarely at video doesn't it?  I certainly could be wrong.  I'm far from 
an expert at troubleshooting this suspend/resume issues.

I find in fact that we (the linux community) are still to this day 
struggling with suspend/resume quite saddening and frustrating.

b.