taskstats_tgid_free: fix usage
Subsystems (4)
PATCH 3/6 taskstats_tgid_free: fix usage
taskstats_tgid_free() is called on copy_process's error path. This is wrong.
IF (clone_flags & CLONE_THREAD)
We should not clear ->signal->taskstats, current uses it, it probably has a valid accumulated info.
ELSE
taskstats_tgid_init() set ->signal->taskstats = NULL, there is nothing to free.
Move the callsite to __exit_signal(). We don't need any locking, entire thread group is exiting, nobody should have a reference to soon to be released ->signal.
include/linux/taskstats_kern.h renamed from include/linux/taskstats_kern.h~3_free +2 -11
@@ -49,17 +49,8 @@ static inline void taskstats_tgid_alloc(static inline void taskstats_tgid_free (struct signal_struct *sig) { -struct taskstats *stats =NULL ; -unsigned long flags; - -spin_lock_irqsave (&sig->stats_lock, flags); -if (sig->stats) { - stats = sig->stats; - sig->stats =NULL ; - } -spin_unlock_irqrestore (&sig->stats_lock, flags); -if (stats) -kmem_cache_free (taskstats_cache, stats); +if (sig->stats) +kmem_cache_free (taskstats_cache, sig->stats); }extern void taskstats_exit_alloc (struct taskstats **,unsigned int *);
kernel/fork.c renamed from kernel/fork.c~3_free +0 -1
@@ -897,7 +897,6 @@ static inline int copy_signal(unsigned lvoid __cleanup_signal (struct signal_struct *sig) {exit_thread_group_keys (sig); -taskstats_tgid_free (sig);kmem_cache_free (signal_cachep, sig); }
kernel/exit.c renamed from kernel/exit.c~3_free +1 -0
@@ -128,6 +128,7 @@ static void __exit_signal(struct task_stflush_sigqueue (&tsk->pending);if (sig) {flush_sigqueue (&sig->shared_pending); +taskstats_tgid_free (sig);__cleanup_signal (sig); } }
Quoted 9 lines · Oleg Nesterov
taskstats_tgid_free()is called on copy_process's error path. This is wrong.IF (
clone_flags&CLONE_THREAD)
We should not clear ->signal->taskstats, current uses it, it probably has a valid accumulated info.
ELSE
taskstats_tgid_init()set ->signal->taskstats = NULL, there is nothing to free.
Quoted 5 lines
Move the callsite to
__exit_signal(). We don't need any locking, entire thread group is exiting, nobody should have a reference to soon to be released ->signal.Signed-off-by Oleg Nesterov <[email protected]>
Looks good!
---
Balbir Singh,
Linux Technology Center,
IBM Software Labs