tjh.dev / kernel
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.

sched_ext: Fix is_bpf_migration_disabled() false negative on non-PREEMPT_RCU

Since commit 8e4f0b1ebcf2 ("bpf: use rcu_read_lock_dont_migrate() for
trampoline.c"), the BPF prolog (__bpf_prog_enter) calls migrate_disable()
only when CONFIG_PREEMPT_RCU is enabled, via rcu_read_lock_dont_migrate().
Without CONFIG_PREEMPT_RCU, the prolog never touches migration_disabled,
so migration_disabled == 1 always means the task is truly
migration-disabled regardless of whether it is the current task.

The old unconditional p == current check was a false negative in this
case, potentially allowing a migration-disabled task to be dispatched to
a remote CPU and triggering scx_error in task_can_run_on_remote_rq().

Only apply the p == current disambiguation when CONFIG_PREEMPT_RCU is
enabled, where the ambiguity with the BPF prolog still exists.

Fixes: 8e4f0b1ebcf2 ("bpf: use rcu_read_lock_dont_migrate() for trampoline.c")
Cc: stable@vger.kernel.org # v6.18+
Link: https://lore.kernel.org/lkml/20250821090609.42508-8-dongml2@chinatelecom.cn/
Signed-off-by: Changwoo Min <changwoo@igalia.com>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

authored by

Changwoo Min and committed by
Tejun Heo 0c4a59df 090d34f0

+19 -12
+19 -12
kernel/sched/ext_idle.c
··· 860 860 * code. 861 861 * 862 862 * We can't simply check whether @p->migration_disabled is set in a 863 - * sched_ext callback, because migration is always disabled for the current 864 - * task while running BPF code. 863 + * sched_ext callback, because the BPF prolog (__bpf_prog_enter) may disable 864 + * migration for the current task while running BPF code. 865 865 * 866 - * The prolog (__bpf_prog_enter) and epilog (__bpf_prog_exit) respectively 867 - * disable and re-enable migration. For this reason, the current task 868 - * inside a sched_ext callback is always a migration-disabled task. 866 + * Since the BPF prolog calls migrate_disable() only when CONFIG_PREEMPT_RCU 867 + * is enabled (via rcu_read_lock_dont_migrate()), migration_disabled == 1 for 868 + * the current task is ambiguous only in that case: it could be from the BPF 869 + * prolog rather than a real migrate_disable() call. 869 870 * 870 - * Therefore, when @p->migration_disabled == 1, check whether @p is the 871 - * current task or not: if it is, then migration was not disabled before 872 - * entering the callback, otherwise migration was disabled. 871 + * Without CONFIG_PREEMPT_RCU, the BPF prolog never calls migrate_disable(), 872 + * so migration_disabled == 1 always means the task is truly 873 + * migration-disabled. 874 + * 875 + * Therefore, when migration_disabled == 1 and CONFIG_PREEMPT_RCU is enabled, 876 + * check whether @p is the current task or not: if it is, then migration was 877 + * not disabled before entering the callback, otherwise migration was disabled. 873 878 * 874 879 * Returns true if @p is migration-disabled, false otherwise. 875 880 */ 876 881 static bool is_bpf_migration_disabled(const struct task_struct *p) 877 882 { 878 - if (p->migration_disabled == 1) 879 - return p != current; 880 - else 881 - return p->migration_disabled; 883 + if (p->migration_disabled == 1) { 884 + if (IS_ENABLED(CONFIG_PREEMPT_RCU)) 885 + return p != current; 886 + return true; 887 + } 888 + return p->migration_disabled; 882 889 } 883 890 884 891 static s32 select_cpu_from_kfunc(struct scx_sched *sch, struct task_struct *p,