Re: Fw: Re: [mm PATCH 4/6] RCU: (now) CPU hotplug

[Dipankar Sarma]

On Fri, Jan 26, 2007 at 12:17:39PM -0800, Andrew Morton wrote:
> On Sat, 27 Jan 2007 01:16:22 +0530
> Dipankar Sarma <dipankar@xxxxxxxxxx> wrote:
> > > The plan is, I hope, to rip it all out and do freeze_processes() on the
> > > hotplug side, so nobody else needs to worry about cpu hotplug any more. 
> > > But at present everyone seems to be in hiding.
> > 
> > This would be ideal. However, we don't seem to have any momentum
> > on this.
> 
> There's no point in expending effort on a fancy new lock until this option
> has been eliminated, so yeah, things are stuck.

The new lock (scalable refcount) is almost already there. 
This http://lkml.org/lkml/2006/10/26/65 can be used to implement
get_cpu_hotplug()/put_cpu_hotplug(). The unfairness issue
can be fixed. I am going to play with these patches and
see if I can come up with something useful quickly.

> > The other thing we would need to do in this case is to
> > check if all the users of cpu hotplug can tolerate a very slow
> > hotplug step if there are 10s of thousands of processes and threads.
> 
> Yes, that needs evaluation.  If it's a problem then we might need to

I will check with the dynamic LPAR people to see if they
have any requirements in this regard. I have also heard
about some people wanting to disable additional threads
in processors (running 1 h/w thread per core) using cpu hotplug 
for some type of applications. I will check with those
folks as well.

> introduce a more sophisticated user interface which enables userspace to
> take multiple CPUs online and offline in a single step, rather than
> one-at-a-time.  I expect that'd be fairly straightforward.

The worry I have is that all of this is a complete rewrite of
existing CPU hotplug and it took us 1+ years to get a decent
CPU hotplug implementation. Rusty and Vatsa can probably vouch
for how difficult it was to get it right. I don't think
it is a good idea to keep CPU hotplug broken until this
rewrite happens. Can't we just go with something that is
already there and works correctly for us without waiting
for the perfect solution ?

Thanks
Dipankar
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/