copy_process: cleanup bad_fork_cleanup_sighand
Subsystems (3)
PATCH 1/2 copy_process: cleanup bad_fork_cleanup_sighand
The only caller of exit_sighand(tsk) is copy_process's error path. We can call __exit_sighand() directly and kill exit_sighand().
This 'tsk' was not yet registered in pid_hash[] or init_task.tasks, it has no external references, nobody can see it, and
IF (clone_flags & CLONE_SIGHAND) At least 'current' has a reference to ->sighand, this means atomic_dec_and_test(sighand->count) can't be true. ELSE Nobody can see this ->sighand, this means we can free it without any locking.
include/linux/sched.h renamed from include/linux/sched.h~1_SIGH +0 -1
@@ -1143,7 +1143,6 @@ extern void exit_thread(void);extern void exit_files (struct task_struct *);extern void exit_signal (struct task_struct *);extern void __exit_signal (struct task_struct *); -extern void exit_sighand (struct task_struct *);extern void __exit_sighand (struct task_struct *);extern void exit_itimers (struct signal_struct *);
kernel/signal.c renamed from kernel/signal.c~1_SIGH +0 -14
@@ -336,20 +336,6 @@ void __exit_sighand(struct task_struct *kmem_cache_free (sighand_cachep, sighand); } -void exit_sighand (struct task_struct *tsk) -{ -write_lock_irq (&tasklist_lock); -rcu_read_lock (); -if (tsk->sighand !=NULL ) { -struct sighand_struct *sighand =rcu_dereference (tsk->sighand); -spin_lock (&sighand->siglock); -__exit_sighand (tsk); -spin_unlock (&sighand->siglock); - } -rcu_read_unlock (); -write_unlock_irq (&tasklist_lock); -} -/* * This function expects the tasklist_lock write-locked. */
kernel/fork.c renamed from kernel/fork.c~1_SIGH +2 -1
@@ -1196,7 +1196,8 @@ bad_fork_cleanup_mm: bad_fork_cleanup_signal:exit_signal (p); bad_fork_cleanup_sighand: -exit_sighand (p); +if (p->sighand) +__exit_sighand (p); bad_fork_cleanup_fs:exit_fs (p);/* blocking */ bad_fork_cleanup_files:
Quoted 3 lines · Oleg Nesterov
[PATCH 1/2]
copy_process: cleanupbad_fork_cleanup_sighand
Err, sorry, should be "[PATCH 1/4]". The same for the next "2/2" patch in series.
Oleg.
Quoted 6 lines · Oleg Nesterov
The only caller of
exit_sighand(tsk)is copy_process's error path. We can call__exit_sighand()directly and killexit_sighand().This 'tsk' was not yet registered in pid_hash[] or init_task.tasks, it has no external references, nobody can see it, and
Quoted 7 lines
IF (clone_flags & CLONE_SIGHAND) At least 'current' has a reference to ->sighand, this means atomic_dec_and_test(sighand->count) can't be true. ELSE Nobody can see this ->sighand, this means we can free it without any locking.
This passes steamroller tests (which now work on ppc64, but that is another story). Looks good to me!
Thanx, Paul
Quoted 47 lines
Signed-off-by Oleg Nesterov <[email protected]>include/linux/sched.h renamed from include/linux/sched.h~1_SIGH +0
-1@@ -1143,7 +1143,6 @@ extern void exit_thread(void);extern void exit_files (struct task_struct *);extern void exit_signal (struct task_struct *);extern void __exit_signal (struct task_struct *); -extern void exit_sighand (struct task_struct *);extern void __exit_sighand (struct task_struct *);extern void exit_itimers (struct signal_struct *);kernel/signal.c renamed from kernel/signal.c~1_SIGH +0
-14@@ -336,20 +336,6 @@ void __exit_sighand(struct task_struct *kmem_cache_free (sighand_cachep, sighand); } -void exit_sighand (struct task_struct *tsk) -{ -write_lock_irq (&tasklist_lock); -rcu_read_lock (); -if (tsk->sighand !=NULL ) { -struct sighand_struct *sighand =rcu_dereference (tsk->sighand); -spin_lock (&sighand->siglock); -__exit_sighand (tsk); -spin_unlock (&sighand->siglock); - } -rcu_read_unlock (); -write_unlock_irq (&tasklist_lock); -} -/* * This function expects the tasklist_lock write-locked. */ kernel/fork.c renamed from kernel/fork.c~1_SIGH +2
-1@@ -1196,7 +1196,8 @@ bad_fork_cleanup_mm: bad_fork_cleanup_signal:exit_signal (p); bad_fork_cleanup_sighand: -exit_sighand (p); +if (p->sighand) +__exit_sighand (p); bad_fork_cleanup_fs:exit_fs (p);/* blocking */ bad_fork_cleanup_files: