Re: [PATCH 6 of 6] PV on HVM suspend\resume

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, Apr 27, 2010 at 06:08:35AM +0100, Stefano Stabellini wrote:
> On Mon, 26 Apr 2010, Jeremy Fitzhardinge wrote:
> > On 04/25/2010 04:25 PM, Stefano Stabellini wrote:
> > >> This won't compile with CONFIG_PREEMPT enabled, because out_destroy_sm
> > >> is missing below.
> > >>
> > >>     
> > >  
> > > Now it is fixed,
> > 
> > Did you push out the fix (ideally as a new delta changeset on top of the
> > previous ones, rather than rebaseing them).
> > 
> 
> Yes but this time I rebased them; next time I'll keep the fix separated.
> 
> > 
> > >  however I tried suspend/resume with CONFIG_PREEMPT
> > > enabled and the kernel seems to hang at resume time; I wonder if it
> > > works for pv guests since the suspend/resume code is similiar...
> > >   
> > 
> > I think Pasi and IanC tested it quite a bit not too long ago, but its
> > possible they didn't test preempt kernels (and it wouldn't surprise me
> > at all if preempt breaks things).
> > 
> 
> all right, more debugging required here...
FYI. The Documents/SubmitChecklist mentions:

12: Has been tested with CONFIG_PREEMPT, CONFIG_DEBUG_PREEMPT,
    CONFIG_DEBUG_SLAB, CONFIG_DEBUG_PAGEALLOC, CONFIG_DEBUG_MUTEXES,
    CONFIG_DEBUG_SPINLOCK, CONFIG_DEBUG_SPINLOCK_SLEEP all simultaneously
    enabled.

amongs other things..