]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
sched/fair: Fix wrong negative conversion in find_energy_efficient_cpu()
authorLukasz Luba <lukasz.luba@arm.com>
Mon, 10 Aug 2020 08:30:04 +0000 (09:30 +0100)
committerPeter Zijlstra <peterz@infradead.org>
Wed, 26 Aug 2020 10:41:57 +0000 (12:41 +0200)
In find_energy_efficient_cpu() 'cpu_cap' could be less that 'util'.
It might be because of RT, DL (so higher sched class than CFS), irq or
thermal pressure signal, which reduce the capacity value.
In such situation the result of 'cpu_cap - util' might be negative but
stored in the unsigned long. Then it might be compared with other unsigned
long when uclamp_rq_util_with() reduced the 'util' such that is passes the
fits_capacity() check.

Prevent this situation and make the arithmetic more safe.

Fixes: 1d42509e475cd ("sched/fair: Make EAS wakeup placement consider uclamp restrictions")
Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Link: https://lkml.kernel.org/r/20200810083004.26420-1-lukasz.luba@arm.com
kernel/sched/fair.c

index abdb54e2339f454aae177c141f3f651bbd6f4559..90ebaa4ae16e28a84110d761defb1eec9a7aee63 100644 (file)
@@ -6594,7 +6594,8 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
 
                        util = cpu_util_next(cpu, p, cpu);
                        cpu_cap = capacity_of(cpu);
-                       spare_cap = cpu_cap - util;
+                       spare_cap = cpu_cap;
+                       lsub_positive(&spare_cap, util);
 
                        /*
                         * Skip CPUs that cannot satisfy the capacity request.