On Thu, Oct 21, 2021 at 11:32 PM Barry Song <21cnbao@xxxxxxxxx> wrote:
>
> On Thu, Oct 21, 2021 at 9:43 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> >
> > On Wed, Oct 20, 2021 at 10:36:19PM +0200, Peter Zijlstra wrote:
> >
> > > OK, I think I see what's happening.
> > >
> > > AFAICT cacheinfo.c does *NOT* set l2c_id on AMD/Hygon hardware, this
> > > means it's set to BAD_APICID.
> > >
> > > This then results in match_l2c() to never match. And as a direct
> > > consequence set_cpu_sibling_map() will generate cpu_l2c_shared_mask with
> > > just the one CPU set.
> > >
> > > And we have the above result and things come unstuck if we assume:
> > > SMT <= L2 <= LLC
> > >
> > > Now, the big question, how to fix this... Does AMD have means of
> > > actually setting l2c_id or should we fall back to using match_smt() for
> > > l2c_id == BAD_APICID ?
> >
> > The latter looks something like the below and ought to make EPYC at
> > least function as it did before.
> >
> >
> > ---
> > arch/x86/kernel/smpboot.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
> > index 849159797101..c2671b2333d1 100644
> > --- a/arch/x86/kernel/smpboot.c
> > +++ b/arch/x86/kernel/smpboot.c
> > @@ -472,7 +472,7 @@ static bool match_l2c(struct cpuinfo_x86 *c, struct cpuinfo_x86 *o)
> >
> > /* Do not match if we do not have a valid APICID for cpu: */
> > if (per_cpu(cpu_l2c_id, cpu1) == BAD_APICID)
> > - return false;
> > + return match_smt(c, o); /* assume at least SMT shares L2 */
>
> Rather than making a fake cluster_cpus and cluster_cpus_list which
> will expose to userspace
> through /sys/devices/cpus/cpux/topology, could we just fix the
> sched_domain mask by the
> below?
> It will be odd to users that a cpu has BAD cluster_id but has
> "meaningful" cluster_cpus and
> cluster_cpus_list in sys.
>
> diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
> index 5094ab0bae58..0f9d706a7507 100644
> --- a/arch/x86/kernel/smpboot.c
> +++ b/arch/x86/kernel/smpboot.c
> @@ -687,6 +687,15 @@ const struct cpumask *cpu_coregroup_mask(int cpu)
>
> const struct cpumask *cpu_clustergroup_mask(int cpu)
> {
> + /*
> + * if L2(cluster) is not represented, make cluster sched_domain
> + * same with smt domain, so that this redundant sched_domain can
> + * be dropped and we can avoid the complaint "the SMT domain not
> + * a subset of the cluster domain"
> + */
> + if (cpumask_subset(cpu_l2c_shared_mask(cpu), cpu_smt_mask(cpu)))
> + return cpu_smt_mask(cpu);
> +
> return cpu_l2c_shared_mask(cpu);
> }
needs to change this line as well, otherwise, sys is still using the
cpu_clustergroup_mask:
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index cc164777e661..c8cd4e9732bd 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -116,7 +116,7 @@ extern unsigned int __max_die_per_package;
#ifdef CONFIG_SMP
#define topology_cluster_id(cpu) (per_cpu(cpu_l2c_id, cpu))
#define topology_die_cpumask(cpu) (per_cpu(cpu_die_map, cpu))
-#define topology_cluster_cpumask(cpu) (cpu_clustergroup_mask(cpu))
+#define topology_cluster_cpumask(cpu) (cpu_l2c_shared_mask(cpu))
#define topology_core_cpumask(cpu) (per_cpu(cpu_core_map, cpu))
#define topology_sibling_cpumask(cpu) (per_cpu(cpu_sibling_map, cpu))
so sys will use topology_cluster_cpumask which describes hardware as it is;
cpu_clustergroup_mask() describes what is given to the scheduler.
>
>
>
> >
> > /* Do not match if L2 cache id does not match: */
> > if (per_cpu(cpu_l2c_id, cpu1) != per_cpu(cpu_l2c_id, cpu2))
>
> Thanks
> Barry
Thanks
barry
|