[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH v1 2/3] sched_credit2.c : runqueue_per_core code



>>> On 12.03.15 at 18:07, <dario.faggioli@xxxxxxxxxx> wrote:
> On Thu, 2015-03-12 at 16:21 +0000, Jan Beulich wrote:
>> >>> On 12.03.15 at 15:57, <uma.sharma523@xxxxxxxxx> wrote:
> 
>> > @@ -1940,10 +1946,14 @@ static void init_pcpu(const struct scheduler *ops, 
> int cpu)
>> >  
>> >      /* Figure out which runqueue to put it in */
>> >      /* NB: cpu 0 doesn't get a STARTING callback, so we hard-code it to 
> runqueue 0. */
>> > -    if ( cpu == 0 )
>> > -        rqi = 0;
>> > +    if ( opt_credit2_runqueue == CREDIT2_OPT_RUNQUEUE_SOCKET )
>> > +    {
>> > +        rqi = (cpu) ? cpu_to_socket(cpu) : boot_cpu_to_socket();
>> > +    }
>> >      else
>> > -        rqi = cpu_to_socket(cpu);
>> > +    {
>> > +        rqi = (cpu) ? cpu_to_core(cpu) : boot_cpu_to_core();
>> > +    }
>> 
>> Rather than extending the bad assumption of CPU 0 being the boot
>> CPU (What if it gets offlined and this or another one onlined back
>> as CPU 0?), can't you find a way to avoid depending on the numeric
>> value of "cpu"?
>> 
> BTW, while we're here, can we really offline CPU#0? I haven't played
> much with CPU on/offlining, so sorry if I'm asking something obvious...

No, we can't currently - due to similar wrong treatment elsewhere in
the tree. But we shouldn't put in more road blocks.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.