From: Nick Piggin <nickpiggin@yahoo.com.au>
"Overlapping" domains of the kind set up on some SGI IA64 systems cause
sched_balance_self to go into an infinite loop, because a CPU may not have
a domain which is a superset of the referring CPU's domain (apart from the
top level domain, in the case of SGI Altix).
Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---
kernel/sched.c | 4 +++-
1 files changed, 3 insertions(+), 1 deletion(-)
diff -puN kernel/sched.c~sched-consolidate-sbe-sbf-fix-3 kernel/sched.c
--- 25/kernel/sched.c~sched-consolidate-sbe-sbf-fix-3 2005-06-09 23:45:15.000000000 -0700
+++ 25-akpm/kernel/sched.c 2005-06-09 23:45:15.000000000 -0700
@@ -1047,6 +1047,7 @@ static int sched_balance_self(int cpu, i
cpumask_t span;
struct sched_group *group;
int new_cpu;
+ int weight;
span = sd->span;
group = find_idlest_group(sd, t, cpu);
@@ -1061,8 +1062,9 @@ static int sched_balance_self(int cpu, i
cpu = new_cpu;
nextlevel:
sd = NULL;
+ weight = cpus_weight(span);
for_each_domain(cpu, tmp) {
- if (cpus_subset(span, tmp->span))
+ if (weight <= cpus_weight(tmp->span))
break;
if (tmp->flags & flag)
sd = tmp;
_