alinux: sched: Fix per-cgroup idle accounting deadlock
fix #29692432 seqcount assumes that writers are already exclusive, so it saves extra locking. However this critial section protected by idle_seqcount can be entered twice if an interrupt tries to wake up a task on this cpu. Once race caused by interrupt is avoided, writers are exclusive. So seqlock is unnecessary, and local_irq_save + seqcount is enough. Fixes: 61e58859 ("alinux: sched: Introduce per-cgroup idle accounting") Signed-off-by: NYihao Wu <wuyihao@linux.alibaba.com> Acked-by: NShanpei Chen <shanpeic@linux.alibaba.com>
Showing
想要评论请 注册 或 登录