2005-11-01 17:01:37

by Paul E. McKenney

Subject: [PATCH] additional -rt RCU usage fixes

Hello!

I guess I need to be more careful when creating experimental RCU patches,
as people have been copying my mistakes. Here is a patch to fix some of
them in -rt.

Signed-off-by: <[email protected]>

---

signal.c | 11 +++++++----
1 files changed, 7 insertions(+), 4 deletions(-)

diff -urpNa -X dontdiff linux-2.6.14-rc5-rt2-ckhandRCUfix/kernel/signal.c linux-2.6.14-rc5-rt2-signalRCUfix/kernel/signal.c
--- linux-2.6.14-rc5-rt2-ckhandRCUfix/kernel/signal.c 2005-10-31 22:28:45.000000000 -0800
+++ linux-2.6.14-rc5-rt2-signalRCUfix/kernel/signal.c 2005-10-31 22:40:07.000000000 -0800
@@ -338,7 +338,7 @@ void exit_sighand(struct task_struct *ts
write_lock_irq(&tasklist_lock);
rcu_read_lock();
if (tsk->sighand != NULL) {
- struct sighand_struct *sighand = tsk->sighand;
+ struct sighand_struct *sighand = rcu_dereference(tsk->sighand);
spin_lock(&sighand->siglock);
__exit_sighand(tsk);
spin_unlock(&sighand->siglock);
@@ -353,13 +353,14 @@ void exit_sighand(struct task_struct *ts
void __exit_signal(struct task_struct *tsk)
{
struct signal_struct * sig = tsk->signal;
- struct sighand_struct * sighand = tsk->sighand;
+ struct sighand_struct * sighand;

if (!sig)
BUG();
if (!atomic_read(&sig->count))
BUG();
rcu_read_lock();
+ sighand = rcu_dereference(tsk->sighand);
spin_lock(&sighand->siglock);
posix_cpu_timers_exit(tsk);
if (atomic_dec_and_test(&sig->count)) {
@@ -1140,7 +1141,7 @@ void zap_other_threads(struct task_struc
}

/*
- * Must be called with the tasklist_lock held for reading!
+ * Must be called under rcu_read_lock() or with tasklist_lock read-held.
*/
int group_send_sig_info(int sig, struct siginfo *info, struct task_struct *p)
{
@@ -1422,7 +1423,7 @@ send_sigqueue(int sig, struct sigqueue *
{
unsigned long flags;
int ret = 0;
- struct sighand_struct *sh = p->sighand;
+ struct sighand_struct *sh;

BUG_ON(!(q->flags & SIGQUEUE_PREALLOC));

@@ -1442,6 +1443,8 @@ send_sigqueue(int sig, struct sigqueue *
goto out_err;
}

+ sh = rcu_dereference(p->sighand);
+
spin_lock_irqsave(&sh->siglock, flags);

/*

2005-11-02 13:12:11

by Ingo Molnar

Subject: Re: [PATCH] additional -rt RCU usage fixes

* Paul E. McKenney <[email protected]> wrote:

> Hello!
>
> I guess I need to be more careful when creating experimental RCU
> patches, as people have been copying my mistakes. Here is a patch to
> fix some of them in -rt.

thanks, applied - will show up in -rt3. Should be done for -mm too,
which now includes rcu-signal-handling.patch?

Ingo

2005-11-02 16:51:22

by Paul E. McKenney

Subject: Re: [PATCH] additional -rt RCU usage fixes

On Wed, Nov 02, 2005 at 02:12:29PM +0100, Ingo Molnar wrote:
>
> * Paul E. McKenney <[email protected]> wrote:
>
> > Hello!
> >
> > I guess I need to be more careful when creating experimental RCU
> > patches, as people have been copying my mistakes. Here is a patch to
> > fix some of them in -rt.
>
> thanks, applied - will show up in -rt3. Should be done for -mm too,
> which now includes rcu-signal-handling.patch?

I believe it should. I will submit any needed catch-up patches against
2.6.14-mm1 when it appears.

Thanx, Paul