Re: mmotm 2009-04-10-02-21 uploaded - forkbombed by work_for_cpu

[Ingo Molnar <mingo@elte.hu>]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Mon, 13 Apr 2009, Andrew Morton wrote:
> >
> >  static void drv_write(struct drv_cmd *cmd)
> >  {
> > -	unsigned int i;
> > +	unsigned int cpu;
> >  
> > -	for_each_cpu(i, cmd->mask) {
> > -		work_on_cpu(i, do_drv_write, cmd);
> > -	}
> > +	for_each_cpu(cpu, cmd->mask)
> > +		smp_call_function_single(cpu, do_drv_write, cmd, 1);
> 
> Ok, that's just -wrong-.
> 
> Doesn't anybody else see anything odd in doing
> 
> 	for_each_cpu(cpu, cmd->mask)
> 		smp_call_function_single(cpu, ..);
> 
> and react to it?
> 
> IOW, why not just do
> 
> 	smp_call_function_many(cmd->mask, do_drv_write, cmd, 1);
> 
> here?

Uhm, yes. The ancient version did:

        for_each_cpu_mask(i, cmd->mask) {
                set_cpus_allowed(current, cpumask_of_cpu(i));
                do_drv_write(cmd);
        }

and we just kept that pattern while somewhat mindlessly converting 
it around, and never realized how it collapses to a nice masked API 
version when the right IPI-call primitive is used.

	Ingo