Re: [PATCH 2/3] stop_machine: simplify

[Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>]

* Rusty Russell (rusty@rustcorp.com.au) wrote:
> On Thursday 10 July 2008 10:30:37 Max Krasnyansky wrote:
> > Rusty Russell wrote:
> > > stop_machine creates a kthread which creates kernel threads.  We can
> > > create those threads directly and simplify things a little.  Some care
> > > must be taken with CPU hotunplug, which has special needs, but that code
> > > seems more robust than it was in the past.
> > >
> > > Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
> >
> > Rusty,
> >
> > You mentioned (in private conversation) that you were going to add some
> > logic that checks whether CPU is running user-space code and not holding
> > any locks to avoid scheduling stop_machine thread on it. Was it supposed
> > to be part of this patch ?
> >
> > Max
> 
> No... I tried it, and it killed my machine.  I didn't chase it for the moment, 
> but it's on the "experimental" end of my patch queue.
> 
> Will play with it again and report,
> Rusty.
> 

Hrm, I must be missing something, but using the fact that other CPUs are
running in userspace as a guarantee that they are not running within
critical kernel sections seems kind of.. racy ? I'd like to have a look
at this experimental patch : does it inhibit interrupts somehow and/or
does it take control of userspace processes doing system calls at that
precise moment ?

Mathieu

> 
> 
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68