Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

Merge tag 'sched_ext-for-6.17-rc7-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext

Pull sched_ext fix from jun Heo:
"This contains a fix for sched_ext idle CPU selection that likely fixes
a substantial performance regression.

The scx_bpf_select_cpu_dfl/and() kfuncs were incorrectly detecting all
tasks as migration-disabled when called outside ops.select_cpu(),
causing them to always return -EBUSY instead of finding idle CPUs.

The fix properly distinguishes between genuinely migration-disabled
tasks vs. the current task whose migration is temporarily disabled by
BPF execution"

* tag 'sched_ext-for-6.17-rc7-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
sched_ext: idle: Handle migration-disabled tasks in BPF code

+27 -1
+27 -1
kernel/sched/ext_idle.c
··· 856 856 return false; 857 857 } 858 858 859 + /* 860 + * Determine whether @p is a migration-disabled task in the context of BPF 861 + * code. 862 + * 863 + * We can't simply check whether @p->migration_disabled is set in a 864 + * sched_ext callback, because migration is always disabled for the current 865 + * task while running BPF code. 866 + * 867 + * The prolog (__bpf_prog_enter) and epilog (__bpf_prog_exit) respectively 868 + * disable and re-enable migration. For this reason, the current task 869 + * inside a sched_ext callback is always a migration-disabled task. 870 + * 871 + * Therefore, when @p->migration_disabled == 1, check whether @p is the 872 + * current task or not: if it is, then migration was not disabled before 873 + * entering the callback, otherwise migration was disabled. 874 + * 875 + * Returns true if @p is migration-disabled, false otherwise. 876 + */ 877 + static bool is_bpf_migration_disabled(const struct task_struct *p) 878 + { 879 + if (p->migration_disabled == 1) 880 + return p != current; 881 + else 882 + return p->migration_disabled; 883 + } 884 + 859 885 static s32 select_cpu_from_kfunc(struct task_struct *p, s32 prev_cpu, u64 wake_flags, 860 886 const struct cpumask *allowed, u64 flags) 861 887 { ··· 924 898 * selection optimizations and simply check whether the previously 925 899 * used CPU is idle and within the allowed cpumask. 926 900 */ 927 - if (p->nr_cpus_allowed == 1 || is_migration_disabled(p)) { 901 + if (p->nr_cpus_allowed == 1 || is_bpf_migration_disabled(p)) { 928 902 if (cpumask_test_cpu(prev_cpu, allowed ?: p->cpus_ptr) && 929 903 scx_idle_test_and_clear_cpu(prev_cpu)) 930 904 cpu = prev_cpu;